Transcript Slide 1

®
How much should that
doggie in the window be?
Strategies for maximizing value through
pricing research
Scott Laing, PRC and Chris Robson, PRC
NorthWest MRA Spring Educational Conference, May 8th 2008
Group Exercise: The BMW 330i






If you were buying this car, what would
you expect to pay?
Sticker price on this is $42,000. Is that
a ‘fair’ price?
Would YOU personally buy this car new
at $42,000?
At $30,000? $20,000? $10,000
The corresponding Audi A4 is $36,000.
Does knowing that change how you look
at the price?
If (like the Trabant in old East
Germany) this were the only car model
available in the US, would you be more
likely to consider it?
Image: FreeDigitalPhotos.net
© Parametric Marketing LLC 2008 Client/Parametric Confidential
So how much should BMW charge?


BMW would obviously sell a lot more if they priced it at $10,000 –
why don’t they?
How do BMW decide what to list it at?






Cost to manufacture and distribute
Expected return from service and extras
Competitive prices and positioning
Company profit goals
Why is ‘sticker’ never what you pay?
Every constituent has an opinion, some grounded in data, some
based on experience and intuition
© Parametric Marketing LLC 2008 Client/Parametric Confidential
The Perils and Pitfalls of Pricing



Pricing is hard. And getting it wrong can be disastrous
There is no ‘one size fits all’ formula: However, there are
many tried and tested approaches
Research can (and should) play a major part in any
substantial pricing task

However, it’s really easy to do bad pricing research…

This is just a brief introduction to the subject
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Research vs. Business

Most of us here are in the MR profession

We want to believe that research and business are tightly integrated



A lot of our worries about pricing studies center on methodology
and client timelines
Clients or end users – the people doing the pricing – also have to
worry about business context
There is a huge opportunity for us to step up and integrate more
understanding of the economics and business implications of
pricing, and providing more support to our business clients in this
vital area
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Key Concepts


We’ve found that it’s really helpful to ensure that we’re all on the
same page with respect to a few key concepts.
Many of these terms are banded about without people fully grasping
their meaning


In this area, terms like “Price Elasticity” are often used as imprecisely
and incorrectly as, say, “Statistical Significance” in our field 
Often, people are uncomfortable admitting that they really don’t
fully understand the meaning of these terms. You don’t need to
understand everything about them, but it’s good to have a
“statistically significant” understanding!
© Parametric Marketing LLC 2008 Client/Parametric Confidential
The Demand Curve
The staple of the pricing research business, a demand curve is
nothing more than a graphical description of the relationship
between the price of a good or service and the number of buyers or
amount that would be bought
Units Bought

$
Note: Economists usually flip these axes
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Price Elasticity of Demand

“Price Elasticity” – or “Price Sensitivity” – is a fancy way of
describing the shape of the demand curve. Elasticity refers to the
relationship between change in price and change in demand level.
For example, a “necessity” is often relatively “inelastic” as changes
in price often result in little or no change in demand.
Units Bought

This good is inelastic: Large
price changes result in small
changes in demand
$
Note: Economists usually flip these axes
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Impact of Price Elasticity on Revenue
The Price Elasticity of Demand is the ratio of the percentage
increase in demand given a corresponding percentage increase in
price – in effect the slope of the curve

We say that a good is Elastic if increasing price decreases total revenue,
and Inelastic if increasing price increases revenue
Units Bought

In an Elastic good the demand falls
faster than the price, so total revenue
falls as price increases
Revenue =
Price x Units
$
Note: Economists usually flip these axes
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Price Awareness

Consumers have internal models regarding what things should cost



These internal models may not be rational or connected with actual
current pricing
Price Awareness is how well these internal models match the
current market realities
As researchers, we need always be aware of the likely disconnect
between customers’ internal models and reality.

Do customers (respondents) really have any clue about the general price
level of a 2GB RAM module? It can make a big difference in interpreting
results
“While few believe that price knowledge is perfect, research has indicated that both academics
and industry personnel overestimate its extent. As a consequence, pricing decisions and market
research may be based on an erroneous assumption.”
Binkley and Bejnarowicz 1995
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Willingness To Pay


Willingness to Pay is a measure of what a (prospective) customer is
prepared to exchange for a good or service.
This matters deeply. It’s easy for a respondent to say a product
should cost $5, but have no interest in paying $5


Mixing up price awareness with willingness to pay is a common pitfall
Often a good is priced lower than an individual’s “Willingness to
Pay”


Economists call the difference between the price consumers are willing to
pay and the actual price of a good the “Consumer Surplus”
Sellers often use the tactic of offering premium goods (e.g. coffee with
froth) to identify people with a higher “Willingness to Pay”
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Factors Affecting Willingness to Pay

It is important to remember that ‘Willingness to Pay” is affected by
many other factors, including:





Price Awareness
Perceived Utility (relative satisfaction from or desirability of a good)
Current financial circumstances and income
Competition and substitutes
“Willingness to Pay” is notoriously difficult to measure, and is
generally hidden in our usual purchasing transactions


In usual markets, consumers keep their ‘Willingness to Pay’ hidden,
whilst retailers have to make their prices public
Some markets – such as housing, cars and e-bay – work by different
rules
The ‘Holy Grail’ of pricing research is to determine Willingness to Pay effectively
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Purchase Intent



Purchase Intent can be thought of as the “decision plan” to buy a
good or service
I may have “price awareness” that an appropriate price for
something is $5. I may be “willing” to pay $5 for it. But that does
not mean that I am going to rush out and buy it this second. I may
never get around to buying it.
Measuring purchase intent can be slippery (and beyond the scope of
today’s session), but here are some pointers:



People don’t actually behave the same way that they respond
There are transformations that can help map between stated purchase
intent and actual purchase rates
Specifying time restrictions in survey questions can greatly improve value
of results
© Parametric Marketing LLC 2008 Client/Parametric Confidential
A Few Research Techniques

Direct Questioning

Monadic pricing Studies

Van Westendorp Price Sensitivity Meter

(Hedonic) Regression

Conjoint Techniques
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Direct Questioning


Q1: How much would you be willing to pay for that Doggie in the
Window?
Direct questioning, either in Quant or Qual studies, is frequently
used and is almost always meaningless

The problems with asking this type of question get even worse in open
focus groups

Please, please don’t use direct questioning!
“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”
– The Strategy and Tactics of Pricing, Nagle and Hogan
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Monadic Price Surveys

In a Monadic Price Survey a respondent would be asked to state
whether they would buy at one specified price:





Would you be willing to buy that Doggie for $5?
The respondent group is split so that several different price points
are tested – but each respondent only sees one price
A Purchase Probability Curve can be built by aggregating the
responses
Pros: Produces a plausible approximation to the demand curve
Cons: Needs a large sample to produce meaningful results. May not
fully take into account competitive and market effects
© Parametric Marketing LLC 2008 Client/Parametric Confidential
The Van Westendorp Price Sensitivity Meter

Van Westendorp (1976) proposed asking four very specific
questions of each respondent:





At what price would you begin to think That Doggie was too expensive to
consider?
At what price would you begin to think That Doggie was so inexpensive that you
would question the quality and not consider it?
At what price would you begin to think That Doggie was getting expensive, but
you still might consider it?
At what price would you think That Doggie was a bargain - a great buy for the
money?
This question set has the advantage that it elicits a price window for
consideration from the respondent – based on their price awareness
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Traditional Analysis of Van Westendorp Data

Traditionally, responses to Van Westendorp questions have been
analyzed by drawing the cumulative curves for the “Too Expensive”
and “Expensive but” responses, and the inverse cumulative curves for
the “Too Inexpensive” and “Bargain” responses
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Traditional Analysis of Van Westendorp Data
Acceptable Price
Range: $16-$24
Indifference Price
Point ($21)
Point of Marginal
Cheapness ($16)
Point of Marginal
Expensiveness ($24)
Optimal Price
Point ($19)


Traditionally the crossover points are given deep and meaningful
names…
…but this approach has been challenged by many analysts, and it’s
theoretical foundation is not regarded as particularly sound
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Rehabilitating Van Westendorp


The dichotomy with Van Westendorp is that the question set is very
powerful and the real-estate taken in a survey is small, but the
standard interpretation is suspect
Newton, Miller and Smith proposed extending the question set to
include a pair of purchase likelihood questions



How likely would you be to purchase That Doggie if it were offered at
[BARGAIN] price?
How likely would you be to purchase That Doggie if it were offered at
[GETTING EXPENSIVE] price?
Using these two additional questions we can build a Consideration
Curve by aggregating the individual consideration curves for each
respondent

This also gives us the opportunity to relax the
‘positive elasticity’ assumptions inherent in the
‘too cheap’ question
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Individual Consideration Curve
Examples of Alternative VW Analysis
Aggregate Consideration Curve
Margin Curves
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Recommendations for Using Van Westendorp

Do not use for categories that are not familiar to respondents

Technique relies on a level of price awareness

Use the extended question set

Ignore the traditional analysis, instead build consideration curves

Use the consideration curves to estimate expected margin curves

Note: The VW questions can be very powerful in small-n/focus
groups, if administered individually

See “Building Price Elasticity Curves”, Evans 2007, for a Monte-Carlo
approach to this
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Hedonic Regression

Sometimes we want to understand pricing patterns in existing
markets


Hedonic Regression treats price as a linear function of a number of
contributing features. Specific prices on the market are used as
‘observations’, and a regression model is used to build the
contributions of each feature.


E.g. How would you price a PC with a particular specification?
Caveat: We need to be careful of correlated input variables and other
problems typical with regression techniques
Hedonic models are often used in Real Estate valuations (where the
‘features’ could be square footage, number of bedrooms etc.)

They can also be used in any other established, feature-driven category
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Conjoint Techniques



In conjoint techniques, respondents are shown ‘bundled’ offerings of
‘attributes’ and asked to respond with their preference (and, in
some cases, whether they would purchase)
Price is usually one of the attributes, and the data can be used to
evaluate pricing options in a very rich manner
There are many different techniques and approaches to conjoint
studies. Recent improvements in raw processing power have made
several new techniques viable

We will give an example of a Choice-Based Conjoint (CBC) with DualResponse None (DRN), analyzed using Hierarchical Bayes (HB)
techniques
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Example CBC/DRN Conjoint Task
Imagine that you happened upon a window and saw the following three
Doggies: Which of these would you be most likely to buy?
Scruffy Brown
Doggie with:
Mangey Black
Doggie with:
Spotty Doggie
with:
No Tail
Non-Wagety Tail
Wagety Tail
And Fleas
$10
And Dog Breath
$20
$15
Assuming that these were the only Doggies available, would you
actually buy the one you chose?
Yes, I would buy this Doggie
No, I would wait to see if I could find a better one
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Conjoint Simulation Example: Price Sensitivity
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Conjoint Simulation Example: Competitive Scenario
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Conjoint Simulation Example: Compare Scenarios
© Parametric Marketing LLC 2008 Client/Parametric Confidential
Application & Strategy


As researchers, we often focus on the price that elicits most
consideration or purchase
When we work on pricing with clients, we must focus on pricing
strategy that is best for business




Margin should generally be more important than units
It is important that we understand how our clients make money, and
provide support for business strategies that maximize that
Establishing consumer demand at different prices and under
different competitive situations is vital, but only part of the business
story. A pricing strategy needs to take into account other factors,
such as supply capability, channel capacity and dynamics, consumer
education needs…
As MR Professionals, we have a lot we can offer our clients
© Parametric Marketing LLC 2008 Client/Parametric Confidential
How Much is that Doggie in the Window?

How much is that doggie in the window?


The one with the wagglely tail.


SKU has been differentiated and utility identified
How much is that doggie in the window?


Protagonist, having identified SKU of interest, desires to know
price supplier would be willing to sell at
Protagonist repeats request for information regarding supplier’s
asking price, underlining the fact that in our usual markets the
price at which a supplier is willing to sell is public, whilst a
consumer’s willingness to pay is generally hidden
I do hope that doggie's for sale.

Protagonist does, however, acknowledge that the supplier is not
bound to sell any SKU, if supplier feels that the level of the
consumer’s willingness to pay does not make adequate profit
© Parametric Marketing LLC 2008 Client/Parametric Confidential
30
Parametric Marketing LLC


Parametric® provides research services to the Consumer Electronics and
technology industries.
We are a research consultancy that specializes in analyzing consumer
behavior and advising companies on the financial implications of product,
brand and promotional decisions. As specialists in advanced analytic
techniques such as conjoint, discrete choice modeling and customer
valuation, we are often called upon by other marketing research companies
to provide client tools and analytics consulting. Parametric was founded in
2003 by Chris Robson PRC, and Scott Laing PRC
400 E Evergreen Boulevard, Suite 303
Vancouver WA 98660
Phone: +1 360.696.2929
[email protected]
www.paramktg.com
© Parametric Marketing LLC 2008 Client/Parametric Confidential