Read on day 29
- Rides booked, out of bookings opened
- Variant ahead, significant
- Rides attended, of rides due by day 28
- Held
- Decided before launch
- Roll it out on phones
We find where visitors leave and why, then design, build and measure the change.
Google Analytics 4
Most phone visitors who open booking leave before they pick a slot. On desktop, far fewer do.
Microsoft Clarity, with consent
Visitors page through the month, tap grayed-out days and close the calendar.
Field by field
The date takes longer than any other field, and most unfinished bookings stop there.
One question on exit
“I couldn’t tell which days had a bike available.”
Ideas are ranked by evidence, expected impact and effort.
Next: A/B test
If we show the next free slots at the nearest store, more phone visitors who open booking will book a ride, because they no longer hunt through the month for a free day.
Our designer draws the variant and our developer builds it in your site’s code or CMS. The winner goes live as it was tested.
The step’s traffic decides how a change is measured, and it is chosen before launch.
Half the visitors see the original, half the variant, in the same weeks.
For less traffic: two equal periods. It shows what changed, not why.
A bug that blocks bookings isn’t tested. It is fixed, then measured.
The sample’s first test, agreed before any visitor sees the variant, so no one can change it to fit the numbers.
More rides booked per booking opened on phones, with no drop in rides attended. Otherwise the original stays.
Phone visitors who open booking, split 50/50 at random between the original and the variant.
Four whole weeks, from a sample size worked out before launch. Read once, on day 29.
Only if booking breaks, the 50/50 split drifts or rides booked drop below a set limit.
Early in a test, either version can pull ahead by chance. The result is read when the planned sample is in, and judged by what the business counts.
The sample test runs four whole weeks. On day 9 the variant was ahead; the test was not stopped. It is read once, on day 29.
Read on day 29
The price follows the scope and is agreed before work starts.
A CRO audit of your main journey, ending in a ranked backlog and the first test plans.
Choose it when you don’t know which step loses people, or why.
One step, one hypothesis: the variant designed, built and measured against its plan.
Choose it when the step and the idea are already clear.
Research, tests and rollouts month after month, with every result recorded in a learning log.
Choose it when you want testing to continue after the first result.
Need tracking set up first, or a new page rather than a better one?
Tell us which journey matters most. We will reply with questions and the first places we would look.
Discuss your projectIt depends on the journeys in scope, the state of your tracking, the research methods, whether your traffic supports A/B tests and who builds the changes: our developers, or yours from our designs and specs. We agree on the price for a clear scope before work starts, and a CRO audit can be ordered on its own. If a testing tool is needed, its license is in your name.
Enough conversions on the tested step to reach a clear result in a few weeks. Before launch we calculate the sample size from the step’s current rate, the smallest change worth acting on and the significance level and power you choose. If the traffic can’t get there, we make research-led changes and compare the same measure before and after.
Until it reaches the sample size set before launch and has run for at least one full business cycle, in whole weeks so every weekday and weekend is counted. We don’t stop a test because one version is ahead on a given day: checking again and again and stopping at the first good result makes false winners far more likely.
A UX audit is an expert review of how easily people complete tasks, delivered as a ranked list of fixes. CRO starts from the step where your funnel loses inquiries, sign-ups or sales: research, hypotheses, then changes that are tested or measured against that number. A UX audit can feed a CRO backlog.
Yes. We check that each step of the journey is recorded as an event, with the inquiry, sign-up or sale marked as a key event in Google Analytics 4 and, for leads, checked against your CRM. Data that was never collected can’t be added later. If the tracking isn’t there yet, it comes first: see analytics and conversion tracking.
Yes. Landing page optimization starts from the campaign: what the visitor searched for, what the ad promised and what the page asks them to do. We align the offer, headline, proof and form, then test or measure the result. For a new page, see landing page design; for the campaigns themselves, Google Ads management.
Not when it follows Google’s guidance on testing: search engines see what visitors see (no cloaking), variant URLs carry a canonical tag pointing to the original, redirect tests use temporary (302) redirects, and the test code is removed once the test ends. We set tests up this way from the start.
Yes, with their consent. In the EU, recording tools need the visitor’s consent before they record, so the recording script loads only after the visitor agrees, and form fields stay masked. The wording of the banner and your privacy policy stays with your legal team.