Price testing can be the most profitable a/b tests your company will run. Yet optimizing pricing, packaging, and the product line-up can be tricky. Here's an outline of what to consider, based on years of a/b and MVT experiments. Every CRO roadmap should include pricing strategy tests.
3. About Me
I have run over 500 marketing experiments
Many of the most successful tests have been related
to pricing & packaging
I teach a/b testing courses around topics such as
statistics and design of experiments
Have run experiments for...
4. Intro
Price testing isn’t just the price. It encompasses:
Price levels
Price gaps
Price anchors
Discount strategy
Line up strategy
Feature composition
Cannibalization
Pricing by Marketing Channel
Pricing by Customer Segment
Pricing by Vertical
and more
5. Broader Impact
Pricing decisions don’t just impact immediate
conversion. It also impacts...
customer satisfaction and loyalty
ability to sell other things in the short-term
refund rates
future revenue stream possibilities
brand perception
customer expectations
6. Can you? Should you?
There’s some debate around the legitimacy of price testing
against live customers.
It is done all the time.
Be careful not to discriminate.
Randomization is not discrimination.
Be careful around consistency of experience
Test targeting can be tricky in particular for logged-out
experiences
Test quickly and have a back-pocket offer.
7. Toolset
Tougher to do with off-the-shelf testing tools
Often requires building own testing and reporting
capabilities
8. Goals
As with any experiment, what are the key success metrics? Just
because it’s a “price test” doesn’t mean everyone is necessarily on
the same page at the start.
KPI options
Top line revenue
Price charged at checkout
Bottom line profit short-term
Bottom line profit long-term
New customers
NPS / WTR / WOM etc.
etc.
9. Effort vs Reward
Price testing is at the harder end of the optimization
spectrum
The payoff is well worth the investment however
I have seen > 40% lifts from price testing
Price testing has the broadest impact: not a page or
segment usually but an overlay on top of the whole
business – huge leverage
10. Lab vs Live
Ideal is to start testing in the lab (surveys, focus
groups, conjoints, etc.)
Once you have pretty good confidence in certain
directions, then move it to live tests
Does not have to be done in that order however.
11. Controlling Variables
Simplest approach is to test in phases:
Packages (e.g., components of the offer)
Lineup (e.g., how many packages there are)
Messaging (e.g., how to sell each package)
Prices for Core Offering (the actual price levels)
Price levels for Add-On Offerings
Timing or Sequence of Sales Cycle
Discounts
Be sure that you are either testing just one of these factors at a
time, or else establish an appropriate experimental design to tease
out the effects of the different factors.
12. Continuous vs Discrete Variables
Price is continuous, can test more points on the
curve without additional sampling overhead
Number of packages is not quite continuous
1 means no choice
Odd numbers are best
Limited to single digits
13. When to Price Test
You have new products
Your products have changed
Your competition has changed their pricing
You’re at a new part of the market adoption curve
You changed strategic directions
14. Subscriptions
SaaS? Have to use cohorts
Trial period? Have to track free to paid
Monthly subscription? Get a few points on the curve
and estimate
Annual subscription? Harder to estimate
15. Choosing Your Customers
In some cases, the prices you choose determines the
kinds of customers you’ll get.
The are downstream implications of acquiring different
types of customers.
Lower list prices or frequently discounted prices bring in
more more conscious customers. The may be:
less loyal
less likely to buy future offerings
more likely to cancel
more likely to use customer support
16. Discounting
You should study the
frequency of discounts
steepness of discounts
duration of discounted period
predictability of discounts
impact on brand perception
impact on loyalty
impact on margins
discounting wars with competitors
17. Scarcity
Scarcity is closely related to price
The price will go up soon
The discount will end soon
There are a limited number
This offer only available for customers from... /
who...
18. Fake Door Testing
Possible in a modified manner with price tests, but
can be tricky. Can pair with limited quantity
messaging when done authentically.
More easily done with product
19. Free
A free version can bring in lots of potential customers
Offering a free version can be viewed as an advertising expense
Free can be important or even necessary from a competitive standpoint
Free can also impact brand perception and ability to sell very high end products
Free can also incur support costs
Free can be a way to get valuable customer feedback to improve the product
Free can be a way to build a network effect for certain offerings
Free has a cannibalization impact that can be measured
20. Free Trial
To test:
Credit card required vs not
Trial period
Auto-charged or not
Feature limits
What happens if customer doesn’t end up paying. Do
any features remain? Any data saved?
21. Refer a Friend as Pricing
“Refer 3 friends to get an additional 1 GB of space”
is a form of price... instead of paying you money,
they are paying you in-kind with social-based
marketing
How else can you “charge” without asking for
dollars?
content uploads
P2P disk usage
data or other ways to help build/grow the product
22. Price Display
Actual Price
Total amount
Per Month, Per Day etc.
Perceived Price
Valued at
Compare at
Price before of after discounts?
Component Value
Price Display
Strike though
List
Discount Display
% off
$ off
You save
Badges
Best value
Most popular
24. Package Hiding
Companies tend to ever expanding product flavors and
package sets. Over time what started out as a rational
for the business becomes a drag on conversion
Simplest way to approach price testing is to trim back the
lines offered to a more management set (1 to 3)
Supplementary packages can be hidden completely or
pushed down to deeper levels or segment-based
marketing – excluded from general pop messaging
This approach will give some idea of loss due to hazards
of excessive choice
27. Indecision Risk
Too many price value trade-offs and prospects bail
out
Must also find right level of product detail to not
overwhelm, but still sufficiently inform
28. Component Values
Hard to market all the features or benefits
Should quantify relative value of each and present
as such
Each package must have one sufficiently high value
component for important customer segments
30. Negative Example
Strike through only on one package
Huge jump between low and mid package
Highlight (“best value”) on most expensive package
Package/Pricing but no CTA based on that info
State fees buried elsewhere
Inconsistent messaging about differences between
packages
Combination of transaction, subscription/trial, and lead
gen not easily parsed
32. Positive Example
Odd number of packages
Center package is recommended
Pricing displayed at the lower-seeming “per month” amount
Original strikethrough price, % off highlighted & amount saved
Dead-simple benefits list; Clutter free design
Visual hierarchy of CTAs
Risk-free messaging
Minimal asterisking
This page was tested extensively:
33. Game Changing Examples
Amazon Prime
Amazon partner / affiliate store
Apple App store
Southwest Airlines
Netflix
Adobe Creative Cloud
34. Conclusion
Price testing encompasses a huge array of testing
possibilities
Price testing requires a serious commitment and
some finesse to get right
Price testing is likely the biggest lever you have