Testing alternatives has become a standard practice in marketing. About ten years ago I was at a seminar led by Google on A/B testing ads. Small, apparently meaningless, changes to wording can give large, 7x or more, differences in click through rates. It was a compelling demonstration and I have been an advocate for A/B testing ever since.
Over the past decade, A/B testing has moved from cutting edge to common practice and it is now well supported by platforms like Optimizely and Unbounce. Of course, comparing text or images on one page is just part of the story. It can be even more important, especially in the context of pricing research, to compare different paths. Fortunately, there are now some platforms that let you do this as well, like Split, and testing is becoming central to user experience research.
A/B testing in pricing
Pricing is the critical P in the four Ps of marketing (Price, Product, Promotion, and Place). With A/B testing so important to other aspects of marketing, how is it being applied to pricing?
This can be a tricky question for B2B marketers and pricing leaders. The Internet remembers what you have done (remember the Wayback machine) and past prices frame current prices. Some B2B contracts even include clauses that require you to provide your best price retroactively, so a pricing experiment can have a cascade of effects. Then there are the legal and ethical issues of offering different prices to two different users at the same time. Pricing experiments are tricky.
So how can you leverage the power of A/B testing and other experiments in pricing?
The first thing to do is to build out the set of hypotheses that you want to test and see how they interact with each other. You need to know what order to test things in before you figure out how to test them. There are two ways to do this: goal-based and action-based.
Goal-based vs. action-based testing
Goal-based testing is all about uncovering the assumptions behind your goals and then systematically testing to see if the assumptions are true (or can be made to be true). Pricing goals generally address four things:
- Market growth (making the pie bigger)
- Market share (getting more of the pie)
- Revenue growth (growing the top line)
- Profit growth (generally gross profit and not net profit)
These days there is often a fifth set of goals around unit economics: Customer Acquisition Costs (CAC), Lifetime Value of a Customer (LTV), the ratio of LTV/CAC and the number of months to recover CAC.
The other approach to generating hypotheses is to think through the actions you could take and what has to be true for you to want to take that action. There can be different reasons to increase price, particularly because they are governed by your overall business strategy.
- Price levels (increase or decrease prices)
- Price curve (change the relationship of prices in a tiered offer)
- Price metric (find a new unit in which to price)
- Packages (change what is included in each package, modify the fences that guide a buyer to one package or another)
- Value messages (test different value messages)
- Order of presentation (change the order in which different value messages and pricing are encountered)
Tiered pricing architecture
How do you test to see if you can raise prices in a tiered pricing architecture? To do this effectively you need to know the role of each tier in your pricing strategy. Are tiers meant to capture demand at different levels of willingness to pay or are they a conveyor belt with each tier being a step on an upgrade path?
In the case of the former, what do you think the volume demand is at different levels? You probably don’t know this in absolute terms, but you can write down your assumptions, work out the implications, and then check to see if the implications show up in your data. You can then start adjusting prices to bring them in line with your assumptions. When the market opportunity is at the low end of the market (in revenue or volume depending on your goals), you will likely end up with a convex pricing curve across tiers as Hubspot has.
In this case, the price of the highest tier is often to frame the target tier. In Hubspot’s case this is the Professional tier, which looks like a good deal when compared to Enterprise.
To test these framing effects, it’s best to increase the price of the tier immediately above your target tier and see if this increases that tier’s overall share. If it does (which is often the case) you can then increase the price in your target tier. This kind of two-step pricing where you reframe the highest tier, check for changes in tier share, and then increase the target tier, and check again for changes in tier share, is one of the fundamentals of price testing.
Of course, there are other ways to guide users into the target tier. You can experiment with different packaging. It is often easier to change packaging than it is to change subscription pricing. To test packaging, it needs to be easy to switch different functions on and off and to know which functions are of value to which kind of user.
As was the case above, you are trying to lead customers into the tier that is best for them, while optimizing the offer for your target tier. Hubspot has done a good job with this as well. On their pricing page, you can see the differences in packages, from Free to Starter to Enterprise. Look at all of the packaging options they have to test. Over time, you will want to be able to easily test different packages and overall impact on demand and distribution across tiers. As with A/B testing of language in search ads, small changes can have surprisingly large results.
These days more and more companies are adding a transactional component to their pricing. A well-chosen transactional metric connects closely to value. In Hubspot’s case, the transactional metric is the number of contacts. It is generally much easier to price test transactional metrics than it is subscriptions. You can test different metrics, different bundles and different price levels without annoying or confusing the market, who are generally paying closer attention to subscription prices. Optimizing transactional pricing is easier than optimizing subscription pricing and can have a big payback.
Finally, it is important to constantly test your value messages. The simplest thing is to A/B test different messages for each tier and see how different sets of messages work together to optimize demand across tiers.
This is not enough though. The order in which messages appear can be as important as the messages themselves, and this can be different for different buyers. We were once pricing a solution that was bought by the head of nursing at some hospitals and by the Chief Financial Officer (CFO) at others. It was important to present the emotional value drivers around healthcare worker safety before presenting any economic value drivers. For the CFO the reverse was true, they cared about worker safety, but they needed to understand the economics first. Testing paths is as important as testing the specific messages.
Testing plays an important role in pricing, but it requires a lot of structure and preparation to be successful. A quick review:
- Begin by testing your economic and emotional value propositions and how they play in different market segments. Use this to refine your segmentation and customer targeting.
- If you are using a tiered architecture, make sure you understand the intended role of each tier and test that it is actually performing that role.
- Adjust packaging to guide buyers into the target tier.
- Take advantage of framing effects, which means you need to look at the impact of a price level on adjacent tiers.
- Create a transactional component for your pricing and use this to get a deeper understanding of market dynamics.
At Ibbaka we are conducting a survey on pricing and innovation. It would be great to have your insights into this. Without good pricing strategies, the best innovations can fail!