From Customer Insight to Experiment: Building a CRO Program in Webflow
TL;DR
- Our client Island.io already had the customer insights, analytics, and UX findings. What they needed was a faster way to turn them into a live test on the site.
- Edgar Allan built a CRO pilot connecting customer insight, evidence, experimentation, and Webflow implementation into one repeatable process, the same process we now run on every Webflow CRO engagement.
- Webflow Optimize is our default testing tool for that process, because it ships an experiment inside the same environment where the site is built, without adding another team to the implementation queue.
Part 3: Building a CRO Program in Webflow
Read Part 2: How to use the BETS framework in a CRO program
Ideas were never a problem for our client, Island, the company behind The Enterprise Browser.
Island is an enterprise B2B SaaS company, and one of their leading products is The Enterprise Browser. Instead of bolting security software on top of a standard browser, The Enterprise Browser builds IT and security controls directly onto the browser itself. It gives enterprise companies the ability to manage how their employees interact with corporate data and SaaS applications.
The Island team already had analytics, heatmaps, UX findings, and plenty of ideas for improving the site. The hard part was getting those ideas into the customer experience without creating another production queue for the brand team to fuss with.
That became the question behind our CRO pilot.
At Edgar Allan, we’re not just an agency; we are a Webflow brand, design, development and performance group that runs conversion rate optimization programs for a host of businesses, including B2B companies. We pair each brand with in-depth customer research that provides the guidelines for a hands-on Webflow implementation. We handle everything alongside you, from kickoff to launch date - rather than handing you a list of recommendations for someone else to build.
Our hypothesis
Could we create a process where Island could turn customer insights into measurable site changes without adding another long implementation cycle?
The original assignment was to take data-informed recommendations and turn them into changes we could measure. We also wanted to look beyond the final form submission and understand some of the smaller moments that happen before someone converts.
This is the process Edgar Allan runs on every Webflow CRO engagement:
Customer insight → Evidence → Hypothesis → Experience → Measurement → Next decision.
Each step has to survive contact with the one before it: a hypothesis has to trace back to evidence, and a shipped experience has to trace back to the hypothesis. That traceability is what turns a pile of good ideas into something a team can actually learn from.
The point was to keep the connection between what we thought we understood about the customer and what we eventually changed on the site.
Why the homepage test needed clean analytics before it needed a variant
Before changing the homepage, we cleaned up the analytics environment, added the GA4 events we needed, brought in qualitative measurement, and looked at quantitative and qualitative behavior together.
Because Island is an enterprise B2B site, form submissions were only part of the picture.
We also looked at CTA engagement, scroll behavior, movement into deeper parts of the site, time on page, and Buyer’s Guide downloads.
Those signals gave us more context around what someone was doing before the final conversion.
Five specific things the Island homepage got wrong before testing
The homepage became the first place we wanted to experiment.
The existing analytics, heatmaps, and UX work pointed to a few areas worth exploring:
- The Enterprise Browser message could be clearer.
- Parts of the layout felt busy.
- Recognizable customer proof could appear sooner.
- Some CTAs needed more context.
- The Buyer’s Guide was sitting farther down the page than we thought it should.
We built a second homepage around those observations.
The updated version simplified parts of the layout and clarified how the page introduced The Enterprise Browser. We brought recognizable logos forward, added more context around the CTAs, and moved the Buyer’s Guide closer to the main paths through the page.
This gave us something to put in front of customers and the opportunity to move the conversation from whether the homepage “felt” clearer to something more measurable: what happened when people used a different version.
Test the direction first
The first experiment looked at a broader homepage direction.
Traffic was split between the existing experience and the new version. From there, the plan was to take the selected direction and get more specific with messaging, CTAs, imagery, and layout.
That sequencing is something we’ve carried into how we think about CRO.
Sometimes the larger question is about how the page works as a whole. Starting with individual button copy can be too narrow if the underlying message or structure still needs work.
The first experiment gives the next one more context.
Why Webflow mattered
One of the problems we were trying to solve was the distance between finding an opportunity and getting it in front of a customer.
Island already had insights worth exploring. Implementation was where some of those opportunities could get stuck.
During the pilot, we compared Visual Website Optimizer (VWO) against Webflow Optimize as the two experimentation platforms. We chose Webflow Optimize as our default for Webflow-hosted CRO work because:
- Webflow runs experiments directly inside the same environment where the site is built and published.
- It allows a team to ship a variant without exporting code to a separate testing layer.
Edgar Allan now uses Webflow Optimize as the standard testing tool for CRO engagements on Webflow sites, and reserves to VWO for cases where a client's stack extends beyond Webflow.
If it takes months to get an idea into the site, it takes months before you can learn anything from it. Webflow gave us a shorter path between the two.
Who runs the pilot, and what you get
A CRO pilot at Edgar Allan is a scoped engagement, not an open-ended retainer.
During this process, a team works the process end to end: a CRO strategist scopes the hypothesis and evidence with the client, a designer curates the look and feel of the page, a developer builds the test experience into Webflow and implements the experiment logic and event tracking, and an analyst monitors results and writes up the next decision.
The client gets a live, running test on their production site, so they’re able to see visually how their site is moving forwards. The Island homepage test described above went from a documented hypothesis to a live A/B split without creating a separate implementation queue for the brand team.
What the Island pilot proved about running CRO in Webflow
The pilot gave us a process we could keep using, with learnings that helped solidify it.
- Start with a customer belief. Decide what evidence would help. Put the change into the site. Measure what happened. Use that information in the next decision.
- Sometimes that next step is another experiment. Sometimes the customer insight needs more work. Sometimes the result points back toward research.
That's the model we run at Edgar Allan, a brand, Webflow design, development, and performance agency built to take a CRO program from customer insight to a live experiment without adding another team to the queue. That’s the same process behind every pilot we run, not just Island's.
Next: Thinking in Bets for CRO: how to make better decisions when the data is incomplete
FAQs
What is Edgar Allan's CRO process in Webflow?
Edgar Allan is a Webflow brand, design, development and performance agency that runs conversion rate optimization programs directly inside the Webflow environment. Its process runs in six steps: customer insight, evidence, hypothesis, experience, measurement, and next decision. Each step has to trace back to the one before it, so a shipped test always connects to a documented customer insight rather than a guess. This structure is applied consistently across CRO engagements, not created fresh per project.
What tools does Edgar Allan use for CRO testing on Webflow sites?
Edgar Allan's standard experimentation stack is Webflow Optimize for A/B testing, paired with GA4 for quantitative measurement and qualitative research (heatmaps, UX findings) for context. Webflow Optimize is the default because it runs experiments inside the same environment where the site is built, letting the team ship a test variant without exporting code to a separate testing platform. VWO is used only when a client's stack extends beyond Webflow.
How long does an Edgar Allan CRO pilot take, and what does a client get?
An Edgar Allan CRO pilot is a scoped engagement, which is run by a small team: a strategist who scopes the hypothesis and evidence, a designer who will curate the look and feel of the page, a developer who will build the test experience in Webflow and implement tracking and experiment logic, and an analyst who reports results. The client receives a live, working test on their production site rather than a set of recommendations for someone else to build later.
Does Edgar Allan test the whole page or individual elements first?
Edgar Allan typically tests the broader page direction before optimizing individual elements like button copy. In the Island pilot, the first experiment split traffic between the existing homepage and a redesigned version addressing layout, messaging clarity, and CTA placement. Only after that direction was validated did narrower tests on specific copy and imagery follow. This sequencing avoids over-optimizing details on a structure that might still need to change.