CRO run as a program,
not a one-off redesign
A redesign is a one-time bet on taste. A CRO program is a repeating cycle. Find where visitors actually drop off, test one change against a real baseline, keep what wins, move to the next leak. We ran this cycle on a funnel that went from 1.5% checkout conversion back up to a number that supported a 2.7x sales quarter.
Four small leaks, no homepage glance will find them
Most sites lose conversion in a handful of specific, findable places. A checkout step asking for information nobody wants to type on mobile. A price shown without context next to a competitor’s lower one. A form that fails silently. A page that loads in four seconds instead of two. None of this shows up in a glance at the homepage. It shows up in session recordings and step-by-step funnel data, exactly where most redesign projects never look.
We have seen checkout conversion swing by a wide margin on the same account within months. Not because anything visibly broke, but because small frictions added up: a feed error, a slow page, a missing trust signal. Nobody noticed, because nobody was watching that number on a schedule.
A one-off redesign fixes whatever the designer happened to notice, then stops. A CRO program keeps watching. The leaks that matter six months from now are not the same ones that matter today.
How the program runs
We audit the real funnel first: session recordings, step-by-step drop-off data, a walkthrough on both desktop and mobile. Mobile friction is usually worse, and usually ignored. That audit produces a backlog of specific hypotheses, ranked by how much traffic hits that step and how large the expected impact is. Not by what looks interesting to fix.
We run one or two tests at a time against a real baseline. We read them for statistical significance, instead of calling a winner after a day of good-looking numbers. We ship the design and copy changes ourselves, rather than handing you a slide deck of suggestions. Each month ends with a readout: what we tested, what won or lost, what that tells us about the next test. Even a flat result is useful information.
Checkout gets specific attention, since that is usually where the highest-value friction lives. A one-point improvement there is worth the most in absolute revenue.
What we need from you
Analytics access, session recordings if your tool supports them. Either the ability for us to implement changes directly, or a developer who can ship what we design within the test cycle. Tell us your current conversion numbers honestly. This program works best on sites that admit the number is bad.
What you see in the report
Each test’s result against its baseline, read with a proper significance check, not an eyeballed dashboard. Over a quarter, the cumulative lift in checkout or landing page conversion against where you started.
Price and timeline
| Option | Price | What it covers | Timeline |
|---|---|---|---|
| Launch or audit | from $900 | Funnel audit, session recordings, prioritized test backlog | 2 to 3 weeks |
| Monthly management | from $1,500 / month | One to two tests running, monthly readout, implementation included | monthly, no lock-in |
| Full control, handover to your team | from $1,900 | Full test framework and backlog handed to your team, training included | 4 to 5 weeks |
Related
Pair this program with A/B testing done correctly and landing page design and optimisation, and run a marketing funnel audit first if you have never had one. The full build is on the funnels and CRM service page. For the statistics layer, see automated A/B test analysis. Real example: the sports nutrition relaunch, one client, results vary.
Think your funnel is leaking but cannot point to where? Get in touch and we will look at your real drop-off data in the first call.
FAQ
How much does a CRO program cost?
From $1,500 a month for one to two tests running at a time with a monthly readout, no lock-in. A one-off funnel audit without the ongoing testing cycle starts at $900.
How long until we see results?
The first audit and test backlog take 2 to 3 weeks. A single test typically needs 2 to 4 weeks of traffic to read correctly, so expect the first confirmed win in 4 to 7 weeks, not days.
What traffic volume do we need for this to work?
Tests need enough visitors to reach significance in a reasonable time. A few hundred checkout starts a month is usually enough for monthly cycles. Less than that, and we recommend fixing bigger, obvious leaks first.
What do you need from us?
Access to analytics and session recordings if you have them. Access to edit the site, or a developer who can ship our changes. The ability to run a proper A/B test on your platform.
How do you report?
A monthly readout in plain language: which test ran, what the result was, whether it reached real significance, and what we are testing next and why.