How to Use the BETS Framework to Decide What to Test in CRO
TL;DR
- BETS is a framework Zach Pousman created at Helpfully that sorts customer beliefs into four confidence levels: Speculation, Theory, Educated Guess, and Backed.
- The right next test depends on how much evidence already supports the underlying customer belief, not on which idea feels most urgent.
- As a Webflow branding, design, development, and performance agency, Edgar Allan uses BETS to decide whether a customer insight for enterprise B2B SaaS companies needs more research, is ready for a Webflow experiment, or already has enough evidence to shape the live site.
Part 2: How to use the BETS framework in a CRO program
Read Part 1: How to Create Customer Insights Before You Start a CRO Program
This is the second article in Edgar Allan's five-part series on how we build CRO programs for clients, combining customer research, experimentation, and Webflow implementation under one team.
What should you test next?
I think there’s another question worth asking before opening an experimentation tool: How confident are we in the customer belief behind the test?
In the latest episode of Building the Next Web, my friend Zach Pousman from Helpfully and I discussed a useful framework for answering that question. Zach calls it BETS, and the Edgar Allan team uses it in our own CRO practice. The idea behind it is simple: treat each customer insight as a bet and move it as the evidence changes.
What is the BETS framework?
BETS turns a customer belief into a graded bet instead of a guess. It sorts what a CRO (Conversion Rate Optimization) team believes about customers into four evidence levels: Speculation, Theory, Educated Guess, and Backed, measuring how much a belief has earned its place in the experiment queue rather than how interesting it sounds.
BETS organizes customer insights by confidence.
The thresholds will be different from company to company. For example, three purchases might be meaningful for one business and barely register for another. What matters is that the team agrees on where each belief sits and what evidence would move it. Once the team agrees, tracking a belief's confidence doesn't require special software: a whiteboard, spreadsheet, FigJam board, or Trello board works fine.
How does a customer belief move through BETS?
Let’s say the starting belief is:
Enterprise buyers want implementation information earlier in the buying process.
That starts as speculation because someone on the sales team mentioned it. It becomes a theory once you review several sales calls and hear the same issue surface early in evaluation, and it strengthens further when site behavior shows enterprise visitors repeatedly moving into implementation content before requesting a demo, backed by interviews pointing in the same direction.
At that point, it’s an educated guess, and you have enough context to design a CRO experiment around it.
For example:
Moving implementation information earlier in the product journey will increase qualified demo requests from enterprise visitors.
If the pattern continues across research and experimentation, the original belief may eventually be backed and start shaping the default enterprise experience. A customer belief should get sharper as you learn, not stay fixed the way it was first written.
How to use the BETS framework in your CRO program
You can keep the process simple with four steps:
- Write down the customer belief behind the test.
What do you think is happening with the customer? - Put it in a BETS category.
How much evidence do you have today? - Decide what evidence would change your confidence.
Maybe that’s analytics, interviews, usability work, sales calls, or an experiment. - Update the belief after you learn something.
If new evidence supports the belief, move it up a category, from Theory to Educated Guess, for example. If new evidence complicates or narrows it, rewrite the belief so it matches what you now know, then run the confidence check again from wherever it lands.
Once a belief moves through these categories, a CRO test carries its own history. Anyone on the team can see where the idea came from, what evidence built it, and why it was worth the engineering time.
What happens when a test doesn’t support the bet?
Go back to the customer belief.
Let’s say you move implementation content earlier and the result doesn’t move much. Maybe the original idea was too broad. So, you go back into the research and learn that implementation matters much more to larger enterprise companies because more internal teams are involved.
With that in mind, the updated insight might become:
Larger enterprise buyers look for implementation information early because adoption involves more internal teams.
That gives you a more specific customer belief and a different experiment to try. The test still gave you your evidence, but BETS gives you somewhere to put it.
Why this helps a CRO team
What we like about BETS is that it gives the assumptions behind a CRO program somewhere to live. Instead of a test idea living only in one person's head, it's written down with a confidence level anyone on the team can check before committing engineering time in Webflow. At Edgar Allan, this confidence check happens before every CRO experiment we build into a client's Webflow site.
It lets us see what the team believes, how much evidence sits behind it, and what we still need to learn. And that gives us a better question to ask before deciding on the next experiment: What are we betting on?
Next: From customer insight to experiment: building a CRO program in Webflow
FAQs
What framework does Edgar Allan use to gauge confidence in a test idea before running a CRO experiment?
Edgar Allan, a Webflow design, development, and performance agency, uses the BETS framework (Backed, Educated Guess, Theory, Speculation) to sort customer beliefs by evidence strength before deciding whether to research further, run an experiment, or use the insight to shape the live site. BETS was created by Zach Pousman at Helpfully; Edgar Allan applies it as a decision filter before committing Webflow development time to a test.
What does BETS stand for in conversion rate optimization?
BETS stands for Backed, Educated Guess, Theory, and Speculation, four confidence levels for a customer insight. Speculation means you've noticed something with no evidence yet; Theory means you have an explanation and some early evidence; Educated Guess means several signals support it; Backed means there's enough evidence to make decisions around it without further testing.
Who created the BETS framework?
The BETS framework was created by Zach Pousman at Helpfully. It categorizes customer insights by how much evidence supports them: Speculation, Theory, Educated Guess, or Backed, so a team can decide whether an idea needs more research or is ready for an experiment.
How do you decide whether a customer insight is ready for an A/B test?
An insight is ready for a test once it reaches "Educated Guess" status in the BETS framework, meaning several independent signals (for example, sales call patterns, site behavior, and customer interviews) all point in the same direction. Below that threshold, the better move is more research, not a live experiment, since testing an unproven belief wastes development time on a hypothesis unlikely to hold.