3.1 Pricing: Lecture Notes

1. Pricing Beyond the 3 Cs

1.1. Why Study Pricing?

• Even slight improvements in pricing can yield significant results. For ex­ample, for a company with 8% profit margins, a 1% improvement in price realization, assuming steady unit sales volume, would boost the company’s profits by 12.5%. By contrast, decreasing fixed costs by 1% would only lead to an increase in profits of 4%. (Dolan 1992).
  • Continental Airlines had 44 million passengers in 2001, at an average ticket price of $193. Charging $2, or 1.04%, more per ticket would have transformed a loss into a profit.
• 80% of managers know how much it costs to produce their product. 23% say they know their customers’ willingness to pay for the product.
• 41% of firms have identified their inability to recruit employees with the right pricing tool sets as their major barrier to implementing pricing prac­tice overhauls.

1.2. Three Cs and Pricing

Typically an introductory marketing class teaches that any pricing strategy should reflect the ‘3 Cs’ of pricing:

• Costs
• Customer
• Competition

Advanced pricing analysis, however, views the 3 Cs as describing a set of­ constraints that pricing strategies must overcome to succeed. The 3 Cs also describe three bad pricing strategies.

• Cost-Based Pricing. Or, pricing on the basis of what it costs you to make the product.
• Customer-Based Pricing. Or, allowing your customers to dictate your pric­ing policy.
• Competition-Based Pricing. Or, choosing your pricing strategy exclusively on the basis of what your competitors do.

1.2.1 Cost-Based Pricing

Cost-Based Pricing involves setting a price such that

Practical implementation problems:

• You have to know costs.
  • A baby sleepsuit company made a loss despite engaging in cost-plus pricing, because they did not realize how much their packaging cost.
  • For example, Diamond Deliveries in Philadelphia. The bicycle division, which management thought of as Diamond’s core business, generated just 10% of total revenues and barely covered its own direct- labor and insurance costs. Diamond was charging customers $4.69 per job, but with fully allocated costs of $9.24 per job, the company was losing $4.55 every time a cyclist picked up a package.1
  • You need to be able to estimate the costs of everything.
• Costs are a function of sales which are in turn a function of prices. This makes such calculations circular. Imagine a firm whose average costs de­ crease with sales.
  • If they sell less, their costs go up. But is a time of declining sales really the right environment to increase prices? For example, Wang Computer
  • If they sell more, their costs go down. But pricing at average cost for small-scale capacity means that the firm may never discover this. This is an issue for firms that face large economies of scale or unknown experience curves. For example, an aeroplane manufacturer found that every time they managed to double their sales, they reduced labor time per aeroplane produced by 10 percent.
• Cost-based pricing is misplaced in industries where there are high fixed costs and near-zero marginal costs. Distributing fixed costs is hard.
  • Complicated when products affect fixed costs in different ways.
  • Activity-based pricing is where the customer agrees to pay the price when the cost is determined with an agreed mark-up based on the contribution to fixed costs. Activity-based pricing is better than regular cost-based pricing in such situations. For example, Owens Minor, a medical supply equipment distributor, improved pricing performance by adjusting its mark-up to be higher on adult diapers compared to replacement ultrasound film.
• Ignoring the value you create leads to underpricing
  • Pet Rock

Rather than asking what prices firms need to charge in order to cover their costs and achieve their profit objectives, firms should ask how their pricing strategy will affect their cost structure.

1.2.2 Customer-Based Pricing.

There are many versions of customer-based pric­ing.

• B2B: Salesforce allows purchasing agents to dictate their prices. An aquarium cleaning product firm sold below their cost because they were scared of losing the custom of a big-box retailer.
• B2C: Giving away a valuable product or add-on for free because the com­pany fears a customer revolt. Flickr hosts 4 billion images (as of Oct 2009). However, the user base is so passionate that managers are reluctant to charge more.

The problems with customer-based pricing are that:

• Customers do not reveal how much they value the product. Radiohead launched a ‘pay what you want’ scheme for their album ‘In Rainbows’. 62 percent paid nothing. Average price of $2.26 an album.
• Customers need to be educated about the value of the product. For example, it was hard for customers to understand what TiVo of­fered over a traditional VHS recorder.
• When customers are used to being in control of a firm’s pricing, they revolt at price changes. Turbotax starting charging users $9.95 for each additional return they filed. Users revolted and gave them a 1.5 star rating on Amazon. The company retreated, reverted to charging nothing and refunded the money. Resident Evil 5, a computer game, tried a different approach: ‘This is the part where I get to say ‘BS’,’ the vice president of Strategic Planning & Business Development wrote on the Capcom forums. ‘RE5 is well worth every penny of $60. A huge game, with tons of replay value, loads of unlockable, new weapons, co-op, mercenaries mode, etc. If any game warrants its price point, it’s RE5’

Instead of asking themselves what their customers would pay, firms should ask themselves what are the prices at which they can convince their customers are supported by their product’s value. How can they segment to reflect differences in customer valuation?

1.2.3 Competition-Based Pricing.

Competition-based pricing describes the situation where a firm does not have a pricing policy that relates to its product, but instead a pricing policy that reflects its competitors’ pricing decisions.

Sometimes this simply takes the form of a firm copying their competitor’s pricing and not conducting their own pricing research.

Sometimes such pricing can take the form of a firm setting a market share ob­jective and discounting their price relative to their competitor until they attain it.

The problems with competition-based pricing are that:

• It encourages firms to ignore their unique value proposition. A radiology software firm with a unique 3D capability lost millions because they priced at the same price as the marketing leader.
• It can lead to price wars if a firm sets a market share objective when the market size is fixed or declining, then this immediately signals that this gain in market share will come at the loss of a competitor. This leads to price wars. For example, the 2005 price war between Anheuser-Busch and Miller was provoked by flat sales.
• Focusing on market share does not necessarily lead to maximum profits even in industries with large economies of scale, some segments of the population are more profitable to serve. In the UK, cable companies have often ignored this and focused on extending their market share to rural villages.

Instead of setting market share objectives, firms should focus on identifying the most profitable segments to serve, and finding ways of profitably serving them while protecting themselves from price wars.

1.3. To Summarize

The 3 Cs are often thought of as pricing ‘dos’, but they can just as well describe pricing don’ts.

Generally, the problem is that each of the 3 Cs encourages a powerless approach when it come to pricing. In all cases, the firm is reacting to a constraint, rather than shaping that constraint. The rest of this course will focus on how we can shape customer and competitive reactions to our pricing strategy, in a way which means that firms do not have to worry about hitting the cost floor.

A more attractive reworking of the 3 Cs would be:

• Create Value
• Calibrate Value
• Communicate Value
• Capture Value

The course is based around these new 4 Cs.

2. Pricing Economic Value to the Customer

2.1. Definition

Economic Value to the Customer (EVC) is based on the insight that a customer will buy a product only if its value to them outweighs the value of the closest alternative, or when Utility_a ≥ Utility_b. The utility of a product depends on its value to the customer minus its price.

Rearranging gives:

Rewriting [Value_a − Value_b] as the differentiation value between product a and b allows us to summarize product a’s price ceiling as

Therefore, to sell a product, a firm needs to price at or below its competitor’s price plus the value advantage its product has to the customer over the rival prod­uct.

2.2. EVC Example

Atlantic Computer has developed software that allows their servers to host twice as much webspace as its rivals. How should they price this new software-server combination?

• Relative to buying two servers from Atlantic’s competitor, by buying one doubly efficient server from Atlantic, a firm would save $4,000 in labor costs,$500 in electricity and $1,500 in software licenses. This suggests that the differentiation value relative to the closest competitive offering is $6,000.
• The price of two servers from the competitor is $6,800.
• This suggests that the EVC of a server with the software is $6,000 + $6,800= $12,800.

2.3. Practical Implementation of EVC analysis in a firm

• Identify what benefit your product provides.
  • Make sure you define a benefit, not a feature. Skill here reaps large rewards. For example, Xerox became very profitable because it priced per photocopy.
• Identify closest competitive offering and calculate its price.
  • Be honest about who your closest competitor is. For example, even though a B2B online video-calling service may want to compare itself to dedicated-line video-conferencing systems, its real closest competitor is Skype, which is free. 
  • Be honest with yourself about the price that competitors sell their product for. For example, a web back-up company calculated an EVC price that was 100 percent greater than what they were charging. However, they omitted to include the fact that all their rivals discounted at 80 percent from list. 
  • Get the units and time-horizon right. This is far easier to do when you have defined the product’s benefit properly. For example, an cleaning product manufacturer left money on the table by not reflecting the fact that its cleaner cleaned twice as much as its competitor.
• Identify potential sources of differentiation value
  • Be specific, not vague. It is important that your firm’s value proposition does not sound like it was churned out by generic software.
• Measure how much value these create
  • Have an independent lab do measurable testing. For example, an air-filter company got a lab to measure the number of dust particles in the air and estimate the effect this would have on asthma sufferers.
  • Often current customers are willing to take part in benchmarking studies, as they view it as free consulting. For example, W.W. Grainger offered consulting and measurement incentives to 15 clients who agreed to participate.
• Add the reference price and differentiation values together to get EVC.

The next challenge is to determine how far below EVC the firm’s ideal price point will be.

2.4. How far should a firm discount from EVC?

It is a common misperception that EVC analysis implies that a firm should price at the EVC. This is not the case. The EVC describes only the maximum price a firm might theoretically charge. Generally, a firm will decide to forego some of their differentiation value.^[1] There is some neurological basis for this ‘fairness effect’.^[2]

This suggests a 50 percent discount rule. For example, Atlantic Computer ultimately priced not at the suggested EVC, but instead discounted its differentiation value by 50 percent ($3,000), so that it priced the product at $9,800.

Set price closer to EVC if:

• End-Benefit effect: When your product is an essential part of a known end-use.
  • Since Bridal Bouquets are a small but essential part of a large and expensive end product, florists are able to charge a great deal more for them than for other bouquets.
• Shared-Cost effect: When the cost is shared.
  • Business hotels have been able to capture the customer’s differentiation value for pillow-top mattresses and expensive bedding because usually the firm is paying for the hotel stay.
• Advantageous mental account: Customers have different mental accounts which they evaluate separately when they make purchase decisions. Price closer to EVC if your product falls into an advantageous mental account such as ‘Vacation Money’ or ‘Tax Refund Money’.
  • For example, success of travel agency selling Las Vegas packages that offered a $300 tax rebate special to coincide with the 2008 Bush tax rebate checks.

May have to price well below EVC if:

• Uncertainty Effect: Uncertainty over realization of attributes
  • For example, a new anti-dandruff shampoo was unable to command a price premium because its claims could not be verified by experts and consumers knew that its effectiveness would vary.
• Newness Effect: Attributes are new and therefore hard to understand
  • For example, Boxee is a ‘freeware cross-platform media center software with social networking features that is a fork of the open source XBMC media center software with some custom and proprietary additions’. Since customers are unlikely to understand the value that implies, they may have to price below EVC when they release their hardware (allegedly) in 2010.
  • Make clear this is temporary. 1969 study: Five brands in two stores. One had discount from initial price, other had no sign of discount. In all five cases, the store that had marked the discount as being from the list price did better in the long run.
• Expenditure effect: When it is a larger proportion of expenditure.
  • For example, an industrial carpet cleaner could charge higher prices to office managers than dedicated office cleaning companies.
• When it is ethically right or legally appropriate to do so
  • Bottled water priced at $35 for a 12 pack during Hurricane Rita
  • North Carolina ‘Price Gouging Law’. North Carolina defines price gouging as intentionally charging an unreasonably excessive price un­der the circumstances for goods or services that are used by North Carolinians during an emergency to preserve, protect or sustain life, health, safety or economic well-being.

May wish to price below EVC for strategic reasons if:

1. Winner-takes-all market
2. Initial Customer Lock-In
3. Pre-empt competitive response

There are three ways that EVC analysis gets used (as a formula, as a guideline and as a diagnostic), which we now discuss in turn.

2.5. Using EVC as a Pricing Formula

It is best to use EVC as a pricing formula

• When competitor’s prices are well-known and concrete
• When a product’s differentiation value is easy to calibrate
• When a product’s differentiation value is easy and believable to communi­cate

The Atlantic Computing example is an almost ideal setting for the use of EVC as a formula. The ideal setting for the use of EVC as a formula is one where the seller knows precisely how the customer will use the product, and intimate details of their likely cost savings or net benefits of using the technology. This generally implies a man­ufacturing or engineering setting. This is no surprise, since consultants developed EVC analysis for these kind of industries. EVC can be particularly useful as a tool to help the sales force sell and prevent price erosion by allowing them to answer questions like:

• How can you justify charging that much for your product?
• Why should I buy your product when your competitor’s product is $3,000 cheaper?
• What is most worthwhile for our firm to keep in mind about your offering?

However, the system should not be so complex that sales people bypass it.

2.6. Using EVC as a Pricing Guideline for new products

EVC can be useful for a firm that is choosing the initial price of its product. The classic example of this is Zantac. As a second entrant to the stomach ulcer market, it may have been natural for them to price at or just under the incumbent’s (Tagamet’s) price. However, EVC analysis demonstrated large differentiation value because of the twice-daily dosage format and fewer side effects, and as a consequence they priced at 50 percent above the incumbent’s price. In two years they were the market leader.

2.7. Using EVC as a Pricing Diagnostic for existing products

EVC is a useful tool for establishing whether an existing product that is per­ forming poorly is overpriced or mis-promoted

• Overpriced
  • Segway recognized that compared to its EVC, segways were felt by the majority of customers to be overpriced at a price point of $3,000. Therefore the Segway company focused on finding ways of reducing the average price, such as the use of Segway on one-day city tours.
• Value message is not focused on crucial differentiation value
  • Michelin initially emphasized all its points of differentiation value such as longevity, speed and safety, in its advertising. Customer tests re­ vealed that their most salient and easy to communicate differentiation value was safety, and Michelin switched its advertising to emphasize that, with great success.
• Value message is not being communicated effectively
  • Duracell found out that its value message that it lasted for longer was not believed by customers. Therefore they switched from a quan­ titative message to one that emphasized that consumers should use Duracell batteries when reliability was crucial.

It is essential to conduct a form of EVC analysis before instituting a price cut, to protect management from making the wrong decision.

3. Pricing under Consumer Uncertainty

In the last lecture, we focused our attention on how firms can use EVC as a conceptual framework to guide their thinking about pricing. We emphasized the importance of value communication, given that what matters is ‘perceived differentiation value’ rather than the firm’s notion of differentiation value. In this lecture, we cover how firm strategy has to change when customers stop making decisions based on price and start using price to make decisions. A reworking of the EVC framework to emphasize consumer behavior would be.

We discuss in turn:

1. How to make consumers feel comfortable about a price when they are not clear about the reference price
2. How price can influence the perceived differentiation value of the product

3.1. When customers do not know reference prices

Many customers do not know prices. A recent MIT study interviewed people in line at the store. They found that while 80% of customers thought they knew the price of the objects in their basket, only 50% did know the price. Given their relative lack of price knowledge, customers often seek ‘price cues’ to inform them as to whether it is a good deal or not. This lack of price knowledge means that they constantly look for clues as to whether this price they are seeing is going to be a good ‘deal’ relative to an unknown reference price.

3.1.1. Product Line Strategies.

The compromise effect is a crucial behavioral impulse when designing prices. Customers often choose the mid-priced option to protect themselves from making a bad choice. The implication here is that one can increase profits by adding a low-price or high-price option in addition to an existing product.

3.1.1.1. Considerations for Compromise Effect.

• Not just for B2C markets.
  • Works successfully in B2B markets. Purchasing agents are never crit­icized for purchasing the mid-price option.
• Needs to be the same brand.
  • Lab evidence shows that if you add a premium brand as a decoy product it doesn’t have the shift consumers from low-price option to mid- price option.
• Everyone at the firm needs to know what the intention of introducing ‘decoys’ is.
  • Architectural software firm lost money after sales force manager mis­understood its decoy premium product and started discounting it to match the mid-price product.

3.1.2. Discounting Strategies.

One way firms can signal that their price will be cheaper than the unknown reference price is to advertise a discount, thereby anchoring the reference price upwards.

3.1.2.1. Proportional Discounts.

• People tend to think of prices in terms of proportion or percentage changes rather than absolute changes. Take for example the following decision. There is no absolute difference in money saved, but consumers respond differently.

• When discounting, discount cheapest part of the product bundle
  • More effective to discount off dessert than total meal.
  • More effective to discount car financing than entire car price.
• The percentage change from 0 is infinity, which perhaps explains why customers like free products so much. An MIT experiment just showed that going from free to 1 cent for a chocolate can reduce demand by 17 percent, even if competitive products also increase by 1 cent.

3.1.2.2. Credible Discounting.

• MIT research suggests that effectiveness of sales signs decreases at around 30 percent saturation.
• Pricing software needs to reflect the need for credibility
  • JCPenney recently started discounting earlier but less (10 percent dis­counts) rather than 20 percent discounts, on basis of projections from pricing software
• Firm needs to give credible reasons
  • GM’s employee discount program was so successful because it signaled a ‘large’ discount.
  • Similarly, liquidation sales are successful because again there is a cred­ible reason for the firm to be discounting.

3.1.3. Price Cues.

3.1.3.1. Key Item Pricing.

Most popular strategy used in retail. Takes advantage of the fact that consumers know the price of a few items and retailers price these aggressively. Works well in the lab. Less real-world evidence it works well because repeat purchase items often have huge fluctuations in prices rendering customers price knowledge imperfect.

3.1.3.2. Price Endings.

$9 endings are effective at signalling value relative to an un­ known reference price.

• Previously, $9 endings have been taken as reflecting left-to right process­ing by customers (for example, a customer believes that a sweater that is marked $9.99 is closer to $9 than $10).
• However, more recent research suggests a more nuanced relationship with the company’s brand message. (e.g. Discount Stores vs. Art Galleries)

3.2. When pricing affects a customer’s valuation

The price of a product itself can send important signals about the product.

3.2.1. Timing Strategies.

Getting the timing of pricing right can actually aid how much consumers enjoy your good. Basically, consumers prefer to avoid a payment that is timed when either they are enjoying the good, or when they expect in particular to not enjoy the good.

3.2.1.1. Practical timing considerations.

• Decouple the pain of paying from consumption for experience goods
  • For example, people are willing to pre-pay and pay a premium for things they enjoy (such as vacations)
• Decouple the pain of paying from the pain of learning how to use a tech­nology good.
  • Customers are more likely to switch if they do not have to pay full price for a technology service in the same month they are learning how to use it.
• Decouple pain of paying from possibility of bad experience of products.
  • Cellphone companies gain more customers if they preannounce an au­tomatic rebate when there is service interruptions.

3.2.2. Asymmetric information about product quality.

High prices may imply high quality if quality is uncertain.

• Do you want the cheapest surgeon to perform laser eye surgery on you?

However, the question remainsL: Why does this work? It isn’t credible that these prices reflect costs (for example having better-quality lasers). The bad surgeon could have bad lasers and still charge a high price. Generally, high prices as a signal of quality work well with:

• Customers of middling sophistication who are uncertain about quality
  • Some survey evidence of limited ability to think iteratively through the credibility of price as a signal of quality
• Scenarios when people who are not knowledgable anticipate that there will be repeat or more knowledgable purchasers in the market too.
  • I might buy more expensive cigars as I assume they are priced right for cigar fans
• Capacity constraints
  • Show promoters in Vegas sell more seats at higher prices

3.2.2.1. Other occasions that price influences quality.

• High prices may signal ‘quality’ of fellow shoppers
  • If you subscribe to a cheap dating service, you may just meet cheap people.
• High price being equated with high quality is so ingrained in customer behavior that it has a neurological component.
  • For example, customer perceptions of the quality of wine are affected by price. This has been confirmed by neural imaging.

3.2.3. Signaling by the customer.

One potential source of differentiation value is for Veblen or ‘snob’ goods, where part of the product’s appeal is its high price.

• High prices allow customers to signal their worth to other individuals
  • When a fountain pen manufacturer raised its prices, it sold more.
• It also allows customers to signal their worth to themselves
• It also allows gift-buyers to signal the value of their present.
  • Scottish whiskies had difficulty in Japan when they tried to enter with lower prices.
  • L’Oreal has had huge success with the ‘Because you’re worth it’ campaign

To summarize: There is a clear use for EVC but there is also a clear use for price framing. The art of pricing is realizing which to emphasize and when.

4. Measuring Customer Reactions to Prices

In this lecture, we learn both how to price new products by using survey tools and how to improve the pricing of existing products by measuring price elasticities.

4.1. Uncontrolled Price Surveys

Very early in the development of survey techniques for marketing, researchers learned that it was futile to ask consumers outright what they would be willing to pay for a product.

4.1.1. Unstructured Questions.

The most obvious way of finding out customers’ feelings about price is to ask them a question like ‘How much should this can of Coke cost?’. Such a question has problems from the outset, because the question does not demand that the customer actually buy the can of Coke. Instead it conflates ‘price awareness’ with willingness to pay.

4.1.1.1. Van Westendorp’s Price Sensitivity Meter.

Marginally better is: ‘How much would you be willing to pay for a can of Coke’? This type of price questioning is used in a species of pricing research named ‘Van Westendorp’s Price Sensitivity Meter’. Here, customers are asked:

• At what price would you consider the product too expensive?
• At what price would you consider the product to be so cheap you would not consider it?
• At what price would you consider the product to start to get expensive?
• At what price would you consider the product to be a ‘good value’ ?

This was originally proposed in 1976 as a way of gauging how important the kind of price-quality signal that we discussed in the previous lecture is for a product. However, it has been distorted from its original intention and is commonly used as a way of getting a price range for a product. The lower bound is the intersection of ‘too cheap’ and expensive. The upper bound is the intersection of ‘too expensive’ and ‘cheap’.

This methodology has been discredited. I suspect its persistent popularity it lies in the facts a) that it produces a figure with intersecting lines that is reminiscent of some kind of demand and supply curve and b) ‘Van Westendorp’ had an impressive sounding last name. Problems in general with open-ended questioning approach:

• The problem with this is that a customer will almost always say: it depends. After all, my willingness to pay for a can of Coke would probably vary from $0.20 in a supermarket if I was in a hurry and I had plenty of Coke cans back home, to $5 if I was in a hotel room and the only place I could get a drink was from a hotel minibar.
• People are not very good at this kind of open-ended question since we are not called upon to make this kind of judgement often. Suppose you weregiven a suitcase and asked to guess how much it weighed. People have a tough time with questions like this. However, when on the other hand people are given two suitcases and asked to say which one is heavier, they get it right 100 percent of the time.
• Lyon (2002) suggests that 20% of respondents say that the price that would be ‘right’ is lower than the price at which they say they would find the product too ‘cheap’ at.
• Such huge price ranges are they give clients an excuse to go with their gut.

4.1.1.2. Monadic Pricing Studies.

Monadic pricing studies are a fancy name given to ‘single cell’ pricing research where respondents are asked a single question about a product.

• ‘If a Coke can is priced at $0.99, how likely are you to buy it?’
• ‘If a Coke can is priced at $0.99, how likely are you to buy it at the snack-shop at work?
• ‘If a Coke can is priced at $0.99, would you buy it at the snack-shop at work’?
• ‘If a Coke can is priced at $0.99 and Pepsi is $1.09 at the snack-shop at work, would you buy Pepsi, Coke or neither supposing, you had no soda so far that day?’

This kind of phrasing is less likely to confuse survey-takers as they just have to make a choice rather than naming a price. This would at least get you more accurate answers for your particular setting. One advantage if you test a broad enough range of prices, you can recreate a demand curve. Potential Problems:

• By focusing the customer’s attention on a reasonable price, you may be censoring extreme responses such as $5 which may be the basis for profitable segmentation.
• Need a lot of respondents to get an answer. Do not be tempted to use a price ladder, however. Asking follow-up questions such as ‘how about $0.89?, How about $0.79?. Just lead the customer to behave as though they are negotiating.
• Need to ensure that the question is not overly focused on the price.

4.1.2. General issues with price survey techniques.

There are three main flaws to this approach

• Customers may deliberately understate their willingness to pay
  • For example, a chemical distributor was told by its customers that they would pay 75 percent less than the market price for waste disposal services.
  • Customers may say a price that reflects how they want the person conducting the survey to view them, but does not reflect their actual willingness to pay. For example, Phillips conducted a survey where the teenage participants said they wanted yellow boom boxes. However, when they came to leave the room, they all chose black boom boxes.
  • Customers may simply be unsure about their willingness to pay. For example, an American bridal survey found that brides overstated willingness to pay five-fold.

4.1.3. Implementation.

• In person
  • The key is to find the right setting. Rushed settings produce worse data. The best data comes from times when people take time and survey-taking becomes a communal activity.
• Online
  • The problem here is reliability of data. We found that when we asked survey participants what browser they used, 40 percent lied. This figure actually increased when we raised their compensation.
• Focus Groups
  • A crucial problem here is that the incentives of the focus group facilitator can often lead to misleading results and analysis. However, without a facilitator no pricing questions get answered.

4.2. Improving Price Surveys by Using Conjoint Analysis

Choice-based conjoint is a laboratory survey of how customers would value dif­ferent attribute bundles. It is usually (and best) used for product design.

4.2.1. Conducting conjoint analysis.

• You choose a limited selection of product features to permutate in questions. Price should be one of these features.
• Software then gives survey respondents a selection of choices between differ­ ent product bundles. It is important that you make sure that respondents always have the option not to buy.
• Software then conducts logit analysis and gives regression output.
• The estimates of how much a customer values each product feature can be converted into rough dollar amounts by multiplying by the ratio of the coefficient on price. This allows us to establish EVC amounts.

4.2.2. Conjoint advantages.

• Conjoint analysis reduced the rather artificial focus on price in surveys by including information about features.
• Conjoint output allows researchers to give ‘utility-values’ to various com­ ponents which can help accurately determine the EVC.
• Quick and relatively cheap

4.2.3. Conjoint disadvantages.

Do not use conjoint if

• Customers are likely to take price as a signal of the quality of the attribute. This problem is often severe and means that you need to severely limit the number of attributes. More than 3 is often too many.
• Product features are not well understood by customers.
• Part of a more complex system of related products.

4.3. Using Historical Data to Improve Existing Pricing

4.3.1. Price Elasticities.

The price elasticity of a product measures the responsive­ness of sales to a change in price. Price elasticity is defined as the percent change in quantity sold given a 1% change in price. This helps analysts figure out whether revenues will be the same, higher or lower after a change in price. If elasticity =1, revenues will be the same from a price change. If elasticity is >1, revenues will be higher with a price decrease. If elasticity<1, revenues will be higher with a price increase.

On average, price elasticities are around −2, but ‘on average’ is not very useful. The standard formula is:

4.3.2. Estimating a constant price elasticity.

One problem with the formula above for the price elasticity is that it can take two different values for whether P or Q are the original or final values. This shouldn’t matter if you are comparing price elasticities across segments and are always consistent about what values you use (for example the value that is used in the denominator, should be the one used to calculate the numerator).

This occurs because we are assuming a linear relationship between price and quantity. One (smart) way of getting around this is by assuming that the demand curve instead has a constant elasticity.

The functional form for a constant-elasticity demand curve is

Taking logs

We can then use spreadsheet analysis to find the value of the constant b which summarizes the price elasticity.

4.3.3. Spreadsheet analysis.

It is quite straightforward to use a spreadsheet tool like excel to calculate a price elasticity from historical data. The steps are:

1. Create new columns for log(sales) and log(prices). Use the excel function= LOG() to convert the raw sales data into log form.
2. In a new cell, insert the function ‘=SLOPE (known y’s, known x’s)’.
3. Choose for the known y’s the log of sales, and for the known x’s the log of prices.
4. You should get a negative number that represents the price elasticity.

One drawback to this procedure is that is not clear how reliable your estimate is. As an alternative you can also use the Data Analysis Add-In for Excel and the ‘Regression’ option to get measures called standard errors (which tell you how precise your estimate is), and R-squared which is a statistic that helps you know how much of the variation in sales is explained by price.^[3]

4.3.4. Why is a price elasticity ever useful?

• Relative margins (More on this in our next lecture):An electronics retailer priced batteries the same all over. [PriceElasticity Analysis] showed the battery that had the highest ”price sensitivity” in Dallas had the lowest price sensitivity in Boston. In other words, while Texans would buy this particular battery only within a narrow price range, Bostonians were far less picky about it. The store altered its prices accordingly, sold more batteries and made more money at it.

Associated Press, 2007

• Rule of thumb pricing tool, especially in retail sector with a large number of SKUs. This caries a weighty health warning since you are effectively assuming away your competitors, that you are already optimizing and that you have increasing marginal costs.

Revenue for a firm: Revenue = P × Q. If they maximize revenue with respect to the quantity they sell:

We can simplify since the price elasticity E_d

If you are a profit maximizing firm, then you of course set price such that MR=MC.

Dividing by P and rearranging yields:

Let us suppose you estimate a price elasticity of E_d = 2, then you can calculate a rough mark-up such that your price should be twice cost. If E_d = 3 then price should be 3/2 marginal cost (or a 50 percent mark-up). If demand is only somewhat elastic E_d = −1.5, then price should be three times cost.

4.3.5. Drawbacks of historical pricing analysis.

• Need accurate pricing data
  • For example, the marketing division of a Web server space company faced problems when the sales force didn’t keep record of discounts.
• Need variation in price
  • A drapes manufacturer could not calculate an elasticity because it had kept the same price for the last 10 years.
• If the data goes back too far it may describe an old scenario. Don’t use pricing data more than 5 years old.
  • A book catalog company calculated an artificially low price elasticity by including pre-internet-era data.
• Misleading because of changes in other confounding factors
  • For example, ice-creams are priced higher and sell better in hot weather. However, if one performs multivariate analysis rather than univariate analysis when calculating price elasticities, it is possible to control for these ‘observable’ sources of bias.

4.3.6. Ways of improving historical pricing analysis.

• CRUCIAL. Calculate different price elasticities for each type of customer, each region, each product.
• Use more data than just aggregate sales and prices
  • DHL employed software that included the reactions of customers who called and got a quote but didn’t ship – that is, a failed sale. By in­ cluding data from this group of customers, they improved their ‘quote to book ratio’ from 17 percent to 25 percent.
• Use panel data econometrics where you include controls for places and times in your regression analysis. The problem is that this can get very expensive both in terms of personnel and costs of acquiring data.

4.4. Improving Pricing Analysis by Using Field Experiments

A lot of the problems with historical data analysis happen because the price changed for reasons that were also associated with changes in sales (for example the weather and ice-cream sales). This can be controlled for by using controlled price experiments. Can be very successful. Ambassadors travel group discovered they had a price elasticity of 0.3 and were able to increase profits by 15 percent. Usually we tend to think of field tests as being experiments, but new tools such as Google Analytics are actually putting them easily in the reach of firms with smaller budgets.

4.4.1. Challenges with field experiments.

• Ensure will have enough price-variation.
  • One experiment I encountered had lowered all prices by 15%, which is only useful if your aim is to drop all prices by 15%.
• If too visible, can anger customers
  • Amazon had to refund $7,000 to angry customers after its DVD pricing experiments were spotted
• Stock-outs or problems with an experimental cell (such as mailing problems) can be detrimental to establishing concrete conclusions.
• Multiple pricing experiments make interpretation of one cell difficult
  • Fingerhut runs 30,000 pricing experiments a year, making interpreta­tion of results that involve complements and substitutes difficult.

Math Appendix

Mid-point elasticity formula. If you are concerned about the approximations involved with the simple elasticity formula but are not doing the kind of detailed historical analysis which is amenable to the log transformation, you can use the midpoint elasticity formula.

Using calculus to move from logs to price elasticity formula. Our constant elasticity demand function is

5. Pricing to Segment Customers

5.1. Why the demand curve can be misleading as a concept for pricing

In the last lecture, we thought hard about how to measure the demand curve. But it is important to also think about why it slopes downwards. The answer is variation in how much our customers value our product.

The most crucial insight of this class is that we shouldn’t think about where we should price along our demand curve. Instead we should think, how many different prices to different customers can I charge along my demand curve?

In this familiar setting, by pricing at or below the ‘Low’ price a firm is leaving money on the table. The crucial question is always how your firm can charge a lower price to low-valuation types and get them to ‘enter’ the market, but still persuade the high-valuation types to pay the high price. During this class we will talk about ‘high’ and ‘low’ types. This is just a short-hand for the multiple types along the demand curve, and the aim is always to construct price strategies to price to all of them.

5.2. Strategic assessment of whether a firm should move from a single price strategy

• Does my product offering have differentiation value?
• Can I identify 2+ customer value profiles who theoretically have different valuations for my product?
• Is there empirical evidence that these different customers actually have the EVC differences you expect them to have?
  • Very different price elasticities indicate that they do have different EVC.
• Is there empirical evidence that customers in this value profile have similar enough EVC?
  • You can find out whether you have segmented enough, by trying to segment again. If the new price-elasticities that you calculate are noticeably different from each other then you have not segmented enough.

When marketers talk about customer segments, they are often thinking of a certain type such as ‘Denim Dads’, ‘Yoga Mommies’, or ‘Gen Y-ers’. However, when pricing analysts use the term ‘segmentation’, they are thinking of some­ thing far more specific. The aim of empirical analysis at all times is to identify and charge different prices to ‘High’ and ‘Low’ types. There are two main means of implementing segmentation: Customer-based and Product-attribute based. In customer-based price segmentation, the firm selects which prices the customer will pay based on their observable characteristics. In product-based price segmentation, the customer decides what price they will pay based on the product’s observable characteristics. The firm’s job is to create an appropriate pricing schedule and price fences between the different products.

5.3. Customer-based price segmentation

5.3.1. Implementation challenges.

In customer-based price segmentation, the firm selects which prices the customer will pay based on their observable characteristics.

• Unambiguous indicator of group membership
  • For example, veterinarians have veterinarian school certificates that the firm could demand be produced for a discount.
• Characteristic must be correlated with EVC. This is not always clear cut. We saw this in the empirical example that we discussed in class, where the doctors had widely different valuations. This is problem too in the real world.
  • For example, Quickbooks had been charging existing customers less because it reckoned that they would value an upgrade less since they already had it. However, there was actually a split between those who valued it less and those who valued it more.
• Product must not be tradeable across groups
  • An enterprising veterinarian could purchase the software and resell at a profit to a hospital.
  • This explains why customer-based segmentation is found mostly for services or experiences.

5.3.2. Price discrimination with anti-competitive intent.

A seller charging competing buyers different prices for the same ‘commodity’ or discriminating in the provision of ‘allowances’ – compensation for advertising and other services ­ may violate the Robinson-Patman Act. This kind of price discrimination may give favored customers an edge in the market that has nothing to do with their superior efficiency. Price discrimination is generally lawful, particularly if they prices reflect the different costs of dealing with different buyers or are the result of a seller’s attempts to meet a competitor’s offering.

5.3.2.1. Specific Legal Tests.

• The Act applies to commodities, but not to services, and to purchases, but not to leases.
• The goods must be of like grade and quality.
• There must be likely injury to competition (that is, a private plaintiff must also show actual harm to his or her business).
• Normally, the sales must be in interstate commerce (that is, the sale must be across a state line).

5.3.2.2. Anti-trust authorities look for.

• Below-cost sales by a firm that charges higher prices in different localities
• Price differences in the sale of identical goods that cannot be justified on the basis of cost savings or meeting a competitor’s prices; or
• Promotional allowances or services that are not practically available to all customers on proportionately equal terms.

5.3.2.3. Outcomes from Robinson-Patman.

• Merck would sell the drugs to hospitals at a 92% discount from the cat­alog price, but only if the hospitals reached certain market shares for the drugs. Prosecuted for swindling Medicare/Medicaid by not offering these discounts.
• Robinson-Patman explains why in Costco you enter a bizarre alternate universe of products for sale. Cereal comes in twin packs. Coke is available in cases containing 32 cans, not 24. Printer cartridges are sold in three- packs, not as singles or in pairs as at other stores.
• Since the Brooke Group Ltd. v. Brown & Williamson and Volvo Trucks North America, Inc. v. Reeder-Simco GMC rulings there have been fewer successful prosecutions. Plaintiffs have to show harm rather than intent.

5.3.3. Ethical issues with customer-based price discrimination.

• Price discrimination must be legal
  • In addition to Robinson Patman Act (1936): B2B regulation, EU de­velopments
  • Wheeler Lea Amendment: Bans unfair or deceptive acts in commerce
  • Racketeer Influenced and Corrupt Organizations Act (RICO). Victo­ria’s Secret’s differing catalog prices were prosecuted as mail fraud under this act.
  • Some state laws forbidding gender based price-segmentation.^[4]
• Price discrimination must be culturally acceptable. It may be perceived as unfair or illegal even if it is not.
  • 2/3 of adult internet users believe that it is illegal for online retailers to charge different people different prices.^[5]
  • Bloomberg waged a war on a Chinese restaurant that charged $1 more on its English menu, though it was difficult to find a law to prosecute under.
  • Controversy of supermarket chain that had unique ‘10 percent’ mark­ up for stores targeted at Hispanic shoppers3
  • Sometimes, just sometimes, this can be resolved by better framing.
    • Disney customer-based discrimination would be a lot less palatable if they said they are going to charge more for adults, than if they said, as they do now, they are going to offer discounts to kids.
• Privacy Issues
  • Customer-based price discrimination that makes use of past customer data is often felt to encroach on a customer’s privacy. This is becoming more important in Europe, and latterly so in the US.
• We worry about the ethical nature of price discrimination. I would argue, however, that when a company does not price-segment because it can’t find appropriate metrics, then this can lead to just as bad consequences. For example, in the early 2000s 90% of those who suffer from AIDS could not afford the prices charged for AIDS drugs.

The feeling that such a practice is not fair can be ascribed partly to a feeling of customer powerlessness. This has led firms to redesign their price segmentation strategies to emphasize that customers have a choice into which bucket they fall.

5.4. Product-based price segmentation

In product-based price segmentation, the customer decides what price they will pay based on the product’s observable characteristics. The firm’s job is to create an appropriate pricing schedule and to maintain appropriate price fences between the different products.

5.4.1. Necessary criteria for success.

• Can identify an unambiguous component of differentiation value correlated with overall EVC (e.g. convenience). Structure product accordingly.
• ‘Distortion’: This component of differentiation value must be correlated strongly enough with overall EVC that high-valuation types will pay a premium rather than not have it. Or in other words, you are going to force your low-valuation types to signal that they have low-valuations because you are going to distort your product quality downwards.
  • Comfort, Speed, Reliability, Ease of use are good places to start.
• ‘Compensation’: This component of differentiation value must be not so essential that low-valuation types will never buy the product without it. Price to compensate them. This is why we see discounts for economy class discomfort.

I illustrate analytically the concepts of ‘Distortion’ and ‘Compensation’ in a detailed mathematical illustration in the appendix to this handout. Though this is a highly analytical concept based on economic ideas of signaling, this is actually the place where pricing becomes more of an art than a science. You will need the skill to identify a product feature or aspect of your product, where you can distort quality downwards and ultimately not annoy your customers. An obvious place to start is to reframe the product choice as the idea that you are distorting product quality upwards from the low-value product, rather than vice versa.

5.4.2. Different types of product attribute segmentation.

• Reliability
  • Commercial oven thermometers cost 10 times more than oven thermometers aimed at the domestic market, because they promise within 2 degree accuracy rather than within 10 degree accuracy. Here the low-valuation type have to signal that they are willing for their baked goods to not always come out in a perfectly consistent manner.
• Capacity
  • Charging $199 for the 8GB IPhone, and $299 for the 16GB IPhone. Here the low-valuation types have to signal that they will not be using a full range of the multimedia capacity of the IPhone.
• Pleasantness
  • Sales rep in charge of ‘economy line’ at printing company that delivered materials to non-profits was deliberately chosen to be unpleasant.^[6]
• Convenience of Purchase
  • The classic example here is coupons. There were 253 billion coupons issued in 2005. Low-valuation types have to signal their low valuations by going to the bother of clipping and remembering the coupon.
• Quantity
  • Low-valuation types signal their type by forgoing bulk discounts.
• Channel
  • Bestbuy.com has better prices than Best Buy stores, but a low-touch environment. Those with low valuations have to signal this by being prepared to not see the product before they purchase it.
• Geography
  • Supermarkets charge different prices depending on location. Often, however, such price differences are more a reflection of local competition than strategic price-segmentation.^[7]
• Add-on Pricing
  • The classic example of add-on pricing is where a hotel separates out its low-valuation types (vacationers) from its high-valuation types (business people) by charging for incidentals such as internet access. Low- valuation types have to signal their low values by being willing to go without internet.^[8]
  • This kind of price-segmentation is very like feature-based price seg­ mentation except that the feature is explicitly optional.
  • The problem is that it can begin to feel like extra fees, which are separately painful each time a customer incurs them.
• Timing of Delivery
  • High-valuation types are willing to pay more for early delivery of goods. This can be as simple as charging a premium for express shipping. Or it can be as complex as skimming your initial market and pricing high for early adopters, and then lowering prices for late adopters. Low- valuation types therefore have to signal their low valuations by waiting for the product to go mass-market. These kind of skimming strategies, however, can lead to customer unhappiness, as we saw in the case of Apple with their IPhone pricing.

5.4.3. Challenges with product-line price segmenting.

• Finding attributes that correlate with EVC
  • For example, a skateboard manufacturer sold a premium line of skate­ boards with better wheels and colors, and also a discount stripped-down line where there were few patterns and the wheels were obviously plastic. The stripped down basic skateboards sold out, and actually resold for higher prices on eBay than the non-stripped premium version.
• Ensuring the integrity of the different products within the product line is maintained.
  • Customers can re-engineer low-segment products. For example, GPS for boating was re-engineered to work for airplanes.
• Perceived fairness for low-segment users
  • A catalog company found a negative effect on all sales for including differential pricing for ‘plus-sizes’
  • Dell faced a backlash from all customers for its new XPS pricing scheme, where ‘high-value’ customers got access to faster technical support.
• Price segmentation can unwittingly reveal your costs.
  • Fast-setting bandage material manufacturer revealed just how cheap its product was to make when it started selling its product as a novelty craft item.
• Can create bad publicity if price fences are found out
  • Best Buy got caught out with a secret website that customer service reps showed in-store customers, that supported its policy of charging less on the internet than in stores.^[9]
• Cost-Effectiveness of Separation
  • Maintaining different product features can be expensive. This can mean that it may in limited circumstances be necessary to discon­tinue product-based price segmentation. For example, a baby carrier company had to discontinue organic line as it was too expensive to manage.
  • Sometimes product segmentation goals can be lost to short-term profit goals. For example, Alamo and National decided to co-locate their car rental services.
  • A similar story is true for rebates, which started off with the same premise as coupons, but retailers got addicted to slippage and people forgetting to send their rebate forms in.
• Need pricing power
  • The problem here is generally non-strategic competitors. You are only able to maintain price fences while your competitors co-operate.
    • Some surprising industries are able to sustain the kind of pricing power that you need for price segmentation. For example, dry cleaners persist in charging more for women’s blouses than men’s shirts.^[10]
    • However, in ad-metrics industry a profitable add-on used to be trademark infringement monitoring services, since these were of interest predominantly to lawyers and other ‘high-valuation types’. However, profitability of this segment has been challenged by ‘one-price-fits-all’ start-ups.

6. Pricing and IT

In today’s lecture we discuss two separate issues. The first is how software platforms can transform pricing. The second is pricing for software platforms.

6.1. Pricing and IT Systems

Theoretically, IT advances have the potential to revolutionize pricing. However, there has been mixed success. Generally, they perform two functions:

• Automating the calculation of price elasticities. This is very useful if you have multiple difference SKUs or different sources of data. (See earlier lecture’s battery example).
• Suggesting a price based on these price elasticities given some kind of constraint (such as limited or perishable capacity). This involves price opti­mization.

6.1.1. Great uses for price optimization software.

6.1.1.1. Avoiding overcrowding.

One optimal use for IT pricing software systems is when there is an externality problem. A good example is road congestion pricing, where a city prices roads higher during peak hours to deter drivers from driving then. This is also the case for water parks, theme parks and so on. These are examples of goods that have a very high fixed capacity and highly variable demand, leading to congestion at certain times that they need to manage.

6.1.1.2. Avoiding spare capacity.

Most firms want to operate at something close to full capacity whenever they can. By booking a hotel room, I am directly preventing another customer from taking that hotel room.

This trade-off between pricing a room low and selling it now against waiting and potentially selling a room at the last minute for a higher price, has given rise to the practice of revenue management.

The good news is that Sloan is including more revenue management examples in the operations management class that is taught by Professors Vivek Farias and Jeremie Gallien. Those of you who are thinking of working in the airline industry, the travel industry, or any industry that has a fixed number of goods to sell should take this class. Also, those of you with a operations management bent will learn useful optimization techniques.

In a general sense, while optimization techniques for revenue management are important, equally important is constructing the right price fences to optimize around. Revenue can be improved by:

• Identifying new pricing fences
• Preventing arbitrage across existing pricing fences
• Improving customer perceptions of pricing fairness

6.1.1.3. Sometimes there are success stories outside of traditional revenue management domains.

Generally most success stories outside of industries with fixed ca­pacities rest on a better understanding of price elasticities. Theoretically, optimization systems could optimize over cost schedules, but I have come across no software systems that do this very well primarily because managers simply do not know cost schedules.

6.1.2. Some problems these systems have faced.

• Used a price management rather than optimization tool.
  • Large department store invested in expensive pricing optimization soft­ ware. Merchandizing set key commands on override and used it pre­ dominantly to manage the prices that they set using their ‘super gut’.
• Sometimes management uses it for uses it was not intended for…
  • Quote from USA Today (2007)
    • A few years ago, 7-Eleven Inc. noticed that when it fought to ”beat down” certain costs, so its con­ venience stores could reduce some of the prices they charged consumers, the effort wasn’t always worth­ while, says Kay Trapp, manager for merchandise pric­ ing. It turned out that several items with newly low­ ered prices saw no change in sales. The chain bought price-optimization software to get such insights in ad­ vance. ”We decided it made us smarter,” Trapp says.
• Limitations to the use of optimization software created for selling airplane seats in industries that have no capacity constraints..
  • Example of software that created artificial constraints in order to be able to say that it was optimizing.

6.2. Case Study: Google Adwords Pricing

A lot of these problems are caused because IT systems model people’s pricing behavior in the future as being like it was in the past. They also assume that people will not try and outsmart the pricing software. However, it makes a lot more sense if IT systems instead were focused on giving incentives so that people were actually willing to tell the firm what they were willing to pay. This is what Google has been able to do with its automated auctions for advertising search terms ‘Adwords’ system. By doing this, Google has transformed itself into the market leader in the provision of advertising.

6.2.1. Adwords Pricing Process.

First, a quick description of the Google search term auction mechanism:

• Firms enroll in Adwords
• They pick the particular search terms that they want their ads to display next to
• They then place a bid for each of these terms.
• (Roughly) the firm that bids the highest amount has their ad displayed first, and pays the price bid by the second highest bidder.
  • I say roughly because Google also adjusts the price bidders pay based on their ‘Quality Score,’ that is, how many customers click through on the ad.
• The second highest bidder’s ad displays second, and they pay the price bid by the third highest bidder, and so on.

6.2.2. Lessons to learn from Google.

• Auctions can help price your good
  • Auctions are incredibly successful at encouraging buyers to state their true valuations. Think of the typical auction, where the winning bidder pays the price bid by the second-highest bidder. Can you think of any circumstances under which the highest bidder would choose to bid less than they did?
  • It is no surprise then that firms that sell goods much prefer to use auctions when possible. ‘About 80 percent of private equity firms in a recent poll said they prefer to use auctions when acting as sellers. Meanwhile, about 90 percent of the same firms said they preferred to avoid auctions when acting as buyers. (Auction Process Roundtable, Mergers and Acquisitions, December 2006, pp. 31-32.)
  • Auctions therefore represent an attractive alternative to implementing revenue management techniques. Rather than using historical data and assumptions, the firm simply gives customers the right incentives to state their true willingness to pay.
  • Items that were difficult to price through other means (for example, Art and Houses). These goods were characterized by high enough prices to make it worthwhile to run an auction.
  • Items that needed to be sold quickly and in bulk (for example trade-in cars, various financial instruments and Dutch flowers). It was worth­ while running an auction in order to be sure of quick sales.
• Identifying the right type of network interaction to price can be crucial
  • Pay-per-click big success compared to pay-per-impression
  • Pay-per-action is less clearly a success
• Pricing on the most micro scale possible is essential for profitability
  • (Goldfarb and Tucker 2009) show that crucially the most profitable search terms are the ones at the most micro level with the fewest searches.
  • We provide evidence that Google is able to charge even more when the market is thin – that is, when the market is part of the long tail.
  • The crucial take-away is that to emulate Google’s success in using electronic auctions, the key thing to do is to use auctions to price the most micro and specific version of your product that you can.

6.3. Pricing for Software Platforms

6.3.1. Two-Sided Platforms.

6.3.1.1. What is a two-sided platform?

The majority of new business models on the internet are actually two-sided platforms. This is especially true for companies that MIT Sloan graduates set up. These are sometimes referred to as two-sided networks or two-sided markets. A two-sided platform is one where two groups interact over a common platform or market.

In the Keurig Case, we thought hard about how to price strategically with lock-in. Generally, there are large ties between markets which have ‘lock-in’ or razor-blades pricing and markets that are two-sided platforms. For example, Keurig would have been a two-sided platform if it just acted as an intermediary for the roasters and the customers and did not attempt to sell coffee itself. The key difference in markets where there is just lock-in is that the customer’s decision to purchase a product gives the firm purchasing power over him. By contrast, in a two-sided platform, the customer’s decision to purchase a product also gives the network operator power over the purchasing decisions over those on the other side of the network.

There are three major different kinds of two-sided platforms (Evans and Schmalensee 2007).^[11]

• Exchanges. These are typically networks that bring together disparate groups of buyers and sellers.
  • Examples include craigslist.org, a Mall, eBay.com.
• Software systems. Generally the platform is the operating system, and the two groups are computer users and program developers.
  • Linux, Windows
  • XBox, Playstation
• Content-Based Markets. These are markets, such as newspapers, where on one side there are advertisers and on the other side users who want access to the content.
  • The content can be user generated like on Facebook.
  • Or it can be generated by the website as in the case of yellowpages.com, Salon.com or (perhaps) Google.

These markets tend to be characterized by high fixed costs and strong network effects.

6.3.2. Pricing strategies in two-sided platforms.

Pricing strategy for two-sided platforms is more complicated, because there are potentially two sets of access fees and usage fees to set.

• An access fee is the price paid to use the network technology, which is independent of usage
• A usage fee is the price paid to interact with someone else over the network

(1) Identify the group that has larger spillovers for the other group

• For example, a nightclub having more women customers is more likely to attract a large number of men than a nightclub having a large number of men is to attract a large number of women. Browsers bring greater benefits to advertisers than vice versa.
• We can use our Excel technique again to gauge the differences in elas­ ticity between groups. We need to subsidize the group that has the lower absolute elasticity. In our example, men have a higher elasticity for the presence of women than vice-versa.

(2) We also need to identify the group that is more price-sensitive.

• For example, lawyers’ clients are more price-sensitive than lawyers themselves.

(3) If price-sensitivity and spillovers go in the same direction, that is great. If not, then it makes strategic sense to reduce access fees on the price-sensitive side and raise the usage price for the less network-sensitive side.

Another strategy is to target subsidies so as to encourage a particular user of the platform to agree to use your platform. For example, when Universal Studios agreed to make only HD-DVDs.

6.3.3. Implementation of two-sided platform pricing strategies.

(1) Exchanges

• EBay is the classic example of a successful two-sided exchange network. They charge both listing fees and transactional fees to their sellers.
• Chemdex provided an online marketplace for sales of chemicals, en­ zymes, lab equipment, biotech products like peptides, and many types of chemical reagents. Chemdex signed up nearly 150,000 users who ordered products from more than 2,000 suppliers. The Chemdex data­ base listed around a million products from beakers to specialty bio­ chemicals. At the point of its IPO Chemdex recorded quarterly sales of$165,000, mostly to Genentech, and had lost $6.8 million in the same quarter. Yet it raised $112.5 million at IPO, and its stock rose 60% on the first day, valuing the whole company (including those shares still privately held) at over $750 million. Its value eventually shrunk to 1% of its previous value. It made its money by charging a 1 percent fee on transactions. The problem with this company was that all their marketing efforts (‘Get a Free T-shirt!’) were aimed at consumers. However, the real challenge was actually persuading suppliers to pay the 5% listing fee. They simply could not see the value proposition.

(2) Software systems

• Adobe’s revolutionary price model was to give away the reader for free and make people pay for Adobe Distiller.
• Opentable charged restaurants to join its reservation service. After a $1000 to $1300 system setup fee, restaurants paid $100 to $200 a month plus $1 for each honored reservation. This restricts the number of restaurants that it can sell for. Customers receive reward points. For each reservation a customer made, they earned $1. By 2009, Opentable has served 75 million diners at 8,400 restaurants.

(3) Content Markets

• Platforms have to decide whether to charge people or use an advertising- based model.^[12] This is not always obvious.
• Facebook’s big challenge is to think of a way of monetizing the interac­tion between users and advertisers. The key issue they face is privacy concerns.

6.3.4. Challenges with two-sided platform pricing.

(1) Make sure that you are subsidizing an attractive group of customers

• FreePC learned this lesson in 1999 when it provided computers and Internet access at no cost to consumers who agreed to view Internet- delivered ads that could not be minimized or hidden. Unfortunately, few marketers were eager to target consumers who were so cost-conscious. FreePC abandoned its offer after losing $80 million.^[13]
• Youtube.com may just have the wrong kind of content to successfully monetarize. Too much is highly personal.^[14]

(2) Product supply issues

• One interesting side note is to think why it is that in a two-sided technology market such as a Play station or an Xbox, firms persistently un­dersupply the market initially. The real winners in such situations are the game console profiteers who hawk the systems on eBay. Play sta­tion 3s sold roughly for $1,500 on eBay just after the launch. The retail price was $599.

(3) Lack of broad understanding of these issues can lead to legal problems.

Sometimes usage fees are very large. ‘Rewards Network’ is a two- sided platform that allows restaurants to offer coupons to potential customers through its website. To acquire restaurants, Rewards Net­ works offered restaurants advances of $10k to $25k. The restaurant rewards network received dining credits of 150% to 200% of the amount of the advance. So for example, if a restaurant was advanced $10k, the Rewards Network would be entitled to sell its members $20k of credit at the restaurant. They signed up 3 million customers and 10,000 restaurants. However, they were then hit by an unfavorable class-action suit for usurious interest rates.

(4) Competitive Concerns

• Why is the Kindle priced so high? Some observers have speculated it is to appease Amazon’s standard publishing market. However, Amazon is also having trouble pricing e-books high where high is above $9.995
• However, the iPad is priced even higher ($499 iPad costs around $290 to make)

7. Out-pricing

Anheuser-Busch executives last month said they were lowering prices to win back drinkers who are choosing wine and spirits over beer. Miller, the No. 2 U.S. brewer, will match Anheuser-Busch’s cuts to avoid losing market share heading into the summer season.

In this lecture series we have intentionally ignored competitors. We did this to avoid a natural tendency to let competitors dictate prices. Instead, we have focused on pricing techniques intended to erect and protect price fences that allow us to defend our value proposition from competitors.

7.1. Legal Regulation of Pricing

7.1.1. Price fixing and bid rigging.

In the United States, price fixing can be prosecuted as a criminal felony offense under Section 1 of the Sherman Antitrust Act. Price fixing is an agreement among competitors to raise, fix, or otherwise maintain the price at which their goods or services are sold. It is not necessary that the competitors agree to charge exactly the same price, or that every competitor in a given industry join the conspiracy. Price fixing can take many forms, and any agreement that restricts price competition violates the law.

Any communication directly with competitors about prices or other elements of competition is strictly forbidden under US and EC antitrust laws. Even indirect communication about prices may be treated as evidence of possible collusion which, combined with other evidence of collusive intent, could result in an antitrust indictment.

7.1.1.1. Patterns that the FTC looks for when detecting price fixing.

• Identical prices may indicate a price-fixing conspiracy, especially when:
  • Prices stay identical for long periods of time
  • Prices previously were different
• Price increases do not appear to be supported by increased costs.
• Discounts are eliminated, especially in a market where discounts historically were given.
• Vendors are charging higher prices to local customers than to distant customers. This may indicate local prices are fixed.

7.1.1.2. Red Flags to FTC in employee behavior.

Each of the following situations has triggered a successful criminal antitrust prosecution:

• Any reference to industry-wide or association price schedules.
• Any statement indicating advance (non-public) knowledge of competitors’ pricing.
• Statements to the effect that a particular customer or contract ‘belongs’ to a certain vendor.
• Statements that a bid was a ‘courtesy,’ ‘complementary,’ ‘token,’ or ‘cover’ bid.
• Any statement indicating that vendors have discussed prices among them­ selves or have reached an understanding about prices.

7.1.1.3. Examples of successful price fixing prosecutions.

• Alfred Taubman, the former chairman of Sotheby’s auction house, served 9 months of a one-year jail sentence for rigging fees charged to clients. He was also fined $7.5m. Taubman acknowledged a dozen meetings with his Christie’s rival but insists they never colluded. ‘We were careful never to discuss anything that was illegal. And we knew the law and I knew it well,’ he said. ‘Whatever we do, we mustn’t discuss prices,’ Mason said. ‘Tennant said that was the first thing that Taubman mentioned to him.’
• Airlines colluded between 2004 and 2006 to levy heavy fuel surcharges on transatlantic flights. Virgin Atlantic escaped financial punishment because it came forward to expose the collusion. BA was fined $200 million by the Office of Fair Trading, Britain’s competition watchdog, and $300 million by the DoJ. Virgin Atlantic escaped financial punishment because it came forward to expose the collusion. Keith Packer, cargo manager, got eight months in US prison and a $20k fine for his part in rigging the cost of price services to and from America.
• Peter Baci, Senior VP of Yield Management for Sea Star Line, LLC, got 4years in prison for fixing the prices of cargo shipped between the US and Puerto Rico. Baci claimed in court that he was acting on orders, but this didn’t help him at sentencing.

7.1.2. Predatory pricing.

Pricing low with an intent to drive your competitor out of business can be prosecuted under the anti-monopolization regulations of the Sherman Act. However, the Supreme Court ruled in 1986 that fare wars that do not drive out competitors are ‘a boon to consumers,’ and the government must be very careful not to ‘chill the very conduct the antitrust laws are designed to protect.’

7.2. Preempting Competitive Price Erosion

One thing to note is that in some sense competitive price erosion is becoming less of a concern empirically. According to RSR Research (January 2010), the extent to which ‘pricing aggressiveness from competitors’ was considered a challenge fell from 48% to 38%. Instead, price erosion became more driven by internal pressure to improve sales.

7.2.1. Assessing risk of price erosion.

7.2.1.1. At risk.

• Falling Demand
• Spare Capacity
• Few differences in differentiation value

Beer is classic example of a product where there is a potential for price wars. It is an undifferentiated product with high fixed costs. Falling demand due to increased demand for wine and other alternatives.

7.2.1.2. Not at risk.

• Tight Capacity
• Steadily Increasing Demand
• Plenty of differentiation value. Razor-Blade or Two-Sided Markets are ideal.

In April, 2009, in the UK, Amazon offered 100 popular MP3s at just 29p (0.50c) per download. Apple’s iTunes store responded by raising prices on the most pop­ ular MP3s to 99p, or $1.50. Apple could do this because of switching costs and superiority of iTunes experience.

7.2.1.3. Cross-price elasticities.

A cross-price elasticity measures the percentage change in unit sales for a firm’s product to the percentage change of a competitor’s price.

It is also possible to calculate a cross-price elasticity for your product using the log-log transformation and the slope formula in Excel. The ys are the log of sales of your good, the xs are the log of price of your competitor’s goods. Ideally it would be zero. Anything over 0.5 is worrying.

7.2.2. Assessing competitor’s pricing sophistication.

• What are their price/cost structures?
• Are they strategically sophisticated? You can only play a pricing game at the level of your least strategic competitor.
• Pricing Game:
  • You need to pick an selling price between $0 and $50 for this object which will maximize your chances of winning this competition.
  • This is a tricky market where the ideal price-discount is 20%. Any higher, then people do not switch their brand allegiance. Any lower, then people begin to doubt the quality of the product.
  • The person who picks a selling price at stage 1 which manages to undercut the average price by 20% (i.e. charges 80% of the average price) will win this object.

Strategic pricing relies not only on an understanding of game theory but an understanding of how well your competitors understand game theory.

7.2.3. Pricing threat.

Selectively revealing information about capabilities and future plans can be useful in avoiding harmful negative-sum game. The goal of such information is to dissuade competitors from reducing prices and ever challenging the firm’s resolve.

• Some strategists say that expanding capacity makes you so fearsome that no competitor dares cut prices for fear of your response.
  • However, the airline industry only regained profitability after they cut capacity after 9/11. In 2003, the industry load factor had recovered from 9/11 and was the highest ever: 73.4% of available seats were filled.
• Talking about a cost advantage can sometimes backfire:
  • An early car computer software firm stated that competitors would not be able to under-cut its prices since it had a cost advantage due to out-sourcing of call centers. Its competitors also moved their ser­ vice departments overseas, which led to price increases from the call- centers.
• Most effective strategy is to commit in public to fight back with pricing.
  • In the case of Winn-Dixie and Big Star vs. Food Lion, price matching commitment lead to doubling of prices on 79 commonly purchased brand items after two years as Food Lion raised its prices.
  • However, the recent ‘secret memo’ that was revealed about Best-Buy’s attempts to circumvent its own price-matching policies shows problems of these policies.

7.2.4. Talking to the press about pricing.

Firms should take steps when communicating with the press to make sure that their pricing strategy is advantageously perceived by competitors, in the same manner as they manage perceptions of stock­ holders and securities analysts.

• Pre-announce pricing intentions and justify increases and decreases
  • Good Communication example: ‘Serguei Beloussov…. , the CEO said he’s now taking a cue from Apple, positioning Parallels [desktop virtual machine software] as a premium product with improved quality and service. What he says he won’t do, is get in a price war with close competitor VMWare Fusion.’ ‘Beloussov is convinced that the best path to more market share is improving the product and support while adding more features than the competition. Mac users, he says are not as price-conscious as their Windows counterparts, and are willing to pay more for a product that works well.’
  • Bad Communication example: In 1997 Compaq Computer Corp CEO Eckhard Pfeiffer claimed that his company’s new business model and aggressive price cuts will not lead to the price war among major PC makers that some were anticipating. The company says…the announced prices make the Desk Pros 6-8% cheaper than the comparable Dell offerings. Pfeiffer [says]… all-out war will be averted because cuts by other manufacturers ‘will come from their margins,’ obviously something Pfeiffer feels won’t carry too far given current market pressure.
• Explain yourself if you are changing your prices to avoid retaliation
  • Heart monitor manufacturer had to sell off multiple units that didn’t comply with proposed FTC regulation, so they reduced price. This set off a price war as their competitors did not understand what they were doing.

7.3. Responding to Competitive Price Erosion Using the Other 3Ps

Sabotage Customer Perceptions of Price Cut by exploiting Product, Promotion and Placement

7.3.1. Product.

• Develop a Fighting Brand. Read ‘Should You Launch A Fighter Brand?’ by Mark Ritson in Harvard Business Review, October 2009 before you do.
  • Intel used its Celeron chip to fight price wars with rather than its premium Pentium chips reasonably successfully.
  • Delta’s Song airline did not stave off JetBlue.
• Improved Product
  • Norton Anti-Virus. After McAfee came in offering ‘Net-to-Zero’ re­ bate, Symantec examined customer research showing that customers wanted more manageability across products (AV + Utilities). Instead, they responded by shifting to a value bundle. Norton System Works, which was a well integrated suite of great point products with a sin­gle, easy-to-use user interface. Symantec built a lead in profitability from retail segments, while McAfee retreated and continued to invest in corporate solutions.

7.3.2. Promotion.

One danger of price wars is that a promotional message can get lost. However, promotion can actually be an effective tool to counter price erosion.

• Southwest Airlines, for example, responded to American’s value-pricing move not with a price move of its own but rather with an advertising campaign proclaiming, ‘We’d like to match their new fares, but we’d have to raise ours.’
• Don’t get too complicated. A Medical device manufacturer told its client hospitals that if he was forced out of the market, they would be at the mercy of the larger company. With a monopoly position, the CEO warned, the customers would find their short-term price cuts transform into long- term price increases. This was not every effective as the hospitals were not sure how credible this was.

7.3.3. Placement.

• Form strategic partnerships with distributors and suppliers or complementary good sellers. Do not tell your distributors not to distribute a low-cost good. That leads to jail time.
  • 2003, 3M was found guilty under Section 2 of the Sherman Act for a de facto exclusive dealing arrangements whereby 3M offered customers large cash payments to purchase their transparent tape exclusively from 3M. Damages totaled $68 million.
  • In 2001 Bed, Bath and Beyond planned to introduce Waterford/Crystal and China. However, Federated and May Co. secured the exclusive deal for the purpose of preventing Bed Bath & Beyond from selling the Lenox/Waterford products. Successful prosecution in NY state and million-dollar fines.
  • Criteria for legal action is any two of
    • Product boycott is horizontal (e.g., encompassing two or more competitors) as opposed to vertical (e.g., covering entities at different levels of the distribution chain)
    • Designed to suppress price competition
    • Involves firms with market power or control over a critical source of supply.
• Instead work with key buying influencers. For example, when Microsoft threatened to enter the small business financial software market, Intuit started a special program to reach out to the 200,000+ accountants who provide support for QuickBooks.

7.3.4. Why not lower price?

(1) Shallow-Pockets Trap

• You may have deep pocketed competitors. Wal-mart vs K-mart in 1990s.
• Bromine manufacturer Dow defeated predatory pricing by a government- supported German cartel, Bromkonvention. Bromkonvention had been flooding the US market with below-cost bromine, at an even lower price than Dow’s. But Dow simply instructed their agents to buy up US Bromine at the very low price, then sell it back in Germany at a profit but still lower than Bromkonvention’s price. In the end, the cartel could not keep up selling below cost, and had to give in.

(2) Low Quality Trap

• Customers assume quality is low

(3) Fragile Market Share Trap

• The consumers you attract are not loyal. Sprint has found that its recent 25% discount plan has led to higher churn. The problem is that the defecting customers are only 2.6% of their total number of year­ end subscribers. The risk is that the other 97.4% of the customers are going to want discounts too.

8. Summarizing Pricing

8.1. Pricing Aim

8.2. Pricing menu

8.3. 10 Challenges to Pricing Intuition

(1) If we use cost-plus pricing, we will make a profit

• Only if you are incredibly accurate with your sales projections

(2) Our customers always prefer low prices

• Customers will always say they like lower prices. However, in many markets price serves as a guide to quality, and pricing too low can send out a negative signal.

(3) Our customers do not know prices, so our pricing strategy is unimportant.

• Lack of customer price knowledge makes how you present the price even more important. This is where strategies such as good-better­ best and price ending cues are key.

(4) Simpler pricing structures are better.

• You always want three prices. Simplifying pricing structures means giving away money. Due to the taxi-meter effect, customers may not always appreciate it when firms do price simply.

(5) If we are profitable, we do not need to price-discriminate.

• On the contrary: Price segmentation is actually going to be most effective for already-profitable firms, because effective segmentation requires the market power implied by profitability.

(6) Razor Blade pricing works because our customers are stupid

• Razor blade pricing works because it is actually subtle price segmen­tation. High-value, high-usage customers pay more. Low-value, low- usage customers pay less.

(7) Our product has network effects, so we need to set a low price

• Your product may well not have network effects at all.
• Even when a product has network effects, price segmentation is key. The crucial questions for network goods are: Whom do I set a low price to and whom do I set a high price to?

(8) Firms need to adjust prices until I fill capacity.

• It can be more profitable to have unused inventory or capacity.

(9) Industries need to work together to ensure that they avoid harmful price wars

• Statements like this lead to jail time. It is the responsibility of the firm, and the firm alone, to avoid a price war.

(10) Our competitors understand our pricing strategy

• Firms have to actively manage perceptions of their pricing by com­petitors, regulators and other stakeholders.

9. B2B Negotiations

B2B is pretty unique to the extent that pricing negotiations are key to most pricing outcomes. Consumer products (with the exception of cars) very rarely have negotiated prices.

9.1. Price Waterfall Analysis

Firms must understand how sensible the prices are that your current negotiation process is leading to.

9.1.1. Action plan.

• Plot the price paid against the number of units bought by the customer. Often it becomes evident that your best (high-volume) customers are paying more per unit than your less profitable customers. This is not a recipe for happy and long-term relationships.
• One often-recommended solution is to construct a ‘price waterfall’ chart. This allows you to work out the true net price for each customer.
  • Start with ‘Transaction price’
  • Net Rebate
  • Net Allowance
  • Net Discount
  • Calculate final ‘Pocket Price’

9.2. Negotiation

Price often looms larger in sales reps minds than in repeat buyers minds. You do not know if the price is too high until you lose the sale

• Establish economic value before discussing price
• Be the first person to bring up price: If you don’t talk price, your customer will.
  • 94% of sales reps do not talk price until their customer does
  • 1/3 will not say price ever, will only write it down or point

9. 2.0.1. Implement Pricing Policy via a Price Menu.

• Establish consistent, transparent criteria for discounting – Fixed prices, Flexible offers
• Never give ad-hoc discounts to repeat customers
• Reward loyalty and trial, not sheer volume

Study of 64 firms who dropped vendors

• Price was reason vendor was dropped 8.1% of the time
• 70.2% delivery problem
• 9.4% quality problem
• 6.2% service problem

9.2.1. Say: Our price is X.

Be firm. Do not be descriptive. Ban Phrases like:

• Regular, normal, list, book, lowest, best, reduced, basic, usual price
• The best I can do
• We can work a little on this price
• Our price is lower than anyone else
• Our price is less than a bag of chips per widget
• How does $100 sound to you? Am I in the ballpark?
• Tell me where I need to be. What do I have to do to get your business?

9.2.2. A Final Story.

Two village idiots wanted to invest $100 they won in the
lottery. They went to a pumpkin farmer they knew and bought 100 pumpkins at $1.00 each. After finding a good place to park and sell the pumpkins from their truck’s flatbed, they started selling them at $1.00 each. When they sold that load they went back to the farmer for more. After selling several truck loads they counted their money and realized they still had only $100. After counting it several more times to be sure, Charley said to Bobby, ”This is getting us nowhere! We sold all our pumpkins, but we still only have $100. Something’s wrong but I surely can’t figure it out.” Bobby said to Charley, ”You idiot, I was gonna let you figure it out, but you’re just too dumb since you didn’t even finish the third grade. It’s as plain as the nose on your ugly face that the only way we can make money is to get a bigger truck!”

10. Non-linear Pricing

Sometimes we need to sell more than one unit of our good to each customer.

10.1. Quantity Discounts

The basic theory about offering quantity discounts is that it allows you to serve high-valuation and low-valuation customers. High-valuation customers signal the fact they are high-valuation customers to you by purchasing in bulk. Low-valuation customers signal that they value your product less by buying fewer units but at a high unit price.

There are some practical problems:

• There are often practical problems when it comes to managing returns, as customers may buy more units to get the discount but expect to be refunded at the average per-unit price for the units they return.
• The ideal product is something like a banana, where resellers cannot store or hoard the product and it is difficult and expensive to resell.
• There must be no realistic chance of hoarding of the product by consumers.
• The Robinson-Patman Act was explicitly designed to try and prevent firms using these kind of non-linear tariffs to benefit chain stores at the expense of ‘Mom and Pop’ stores

10.2. Metered Pricing

Often some form of metered pricing arrangement is preferable where the buyer purchases goods simply on the basis of usage. Variations along this theme are razor-blade models where the customer buys a low price durable (the razor) and then subsequently purchases high price consumables (the blades).

10.3. Value-Based Pricing Metrics

With any form of pricing, the key to pricing multiple units of a good to each customer is to get the pricing metric right.

• For example, for most information services it makes sense to price ‘per download’ rather than per minute.
• It very rarely makes sense to charge per hour.