A page tested on real traffic,
not on an opinion in a meeting
A landing page redesign based on internal opinion often performs worse than the page it replaced. It changes everything at once, with no way to know which change helped or hurt. We test one variable at a time against real traffic. That way every result teaches something that carries into the next test.
Where testing programs go wrong
The most common failure is a full redesign that changes the headline, layout, images and offer all at once. If conversion improves, nobody can say which change caused it. If it drops, the same is true. The business has just spent a redesign budget to learn nothing it can repeat. The second failure is calling a test after a few days, because one version looks ahead before the sample size actually supports that conclusion. That leads to decisions based on noise.
The third is testing on desktop only, or treating mobile and desktop as one traffic pool. The two often behave differently enough that a page optimised for one can quietly underperform on the other.
A page redesign and a page test are different activities, with different risk profiles. A redesign can lose ground invisibly. A test, done properly, can only teach something. That is the case for testing over wholesale redesigns, whenever traffic allows it. Even a page that clearly needs visual modernising benefits from testing the highest-risk elements, headline and offer, before the rest of the redesign ships.
How we run each test
- One variable at a time. Headline, offer, form length or layout isolated per test, so a result is attributable to a specific change, not a bundle of them.
- Significance before a verdict. Sample size and test duration checked properly, with an early-call warning rather than a confident-looking but premature conclusion.
- A prioritised backlog. Test ideas ranked by likely impact using heatmaps, session recordings and funnel drop-off data, so testing time goes to the highest-impact changes first.
- Mobile and desktop split. Performance checked and, where warranted, tested separately, since the two audiences often respond differently to the same page.
- Findings that compound. Each test’s result gets recorded and built into the next hypothesis. That way the program gets smarter over time, instead of restarting from zero with every new idea.
What we need to start
- A landing page we can edit directly, or a developer we can hand precise test specifications to.
- Enough paid or organic traffic to reach a valid sample size within a reasonable window.
- Analytics access, heatmap or session recording tools if you already use them, or agreement to add one.
- Patience to let each test run its full course rather than stopping early.
- Agreement on the primary conversion metric before testing starts.
What’s in the monthly report
We report which variant won, the lift it produced and how confident that result is, following a proper significance check. Where traffic allows, results are split by device and by traffic source. A page that wins for paid search traffic does not always win for paid social traffic.
Monthly reporting includes the test backlog and what is queued next, so the program’s direction is visible, not just its latest result.
We also segment results by new versus returning visitors, where volume allows. A page change that helps a first-time visitor can sometimes hurt a returning one, who was already familiar with the old layout.
Cost and timeline
| Option | Price | What’s included | Timeline |
|---|---|---|---|
| Launch or audit | from $800 | Testing framework and prioritised backlog built, first test launched. | 1 to 2 weeks |
| Monthly management | from $800 / month | Ongoing test execution, significance checking, monthly result report. | monthly, no lock-in |
| Full control, handover to your team | from $1,800 | Testing framework, backlog and significance methodology documented and handed to your team. | 3 to 4 weeks |
More on testing and CRO
This page sits under our Performance marketing and Funnels and CRM work. See also Creative testing program, Conversion tracking and capi setup, Meta ads management for adjacent paid-channel services.
For the automation side of reporting and budget decisions, see ab test analysis. For real numbers behind these benchmarks, read the sports-nutrition sales x2.7 case study, one client, results vary.
Ready to see what this would look like for your account? Get in touch and we will look at your current setup in the first call.
FAQ
How much does this cost?
Both launch and monthly management start from $800, with the pace of testing scaling to your traffic volume and ambitions.
How long until we see results?
The program runs continuously. A first statistically valid test result typically takes 2 to 4 weeks, depending on how much traffic the page receives.
What ad budget do we need alongside this?
Testing needs traffic more than a large budget. A page with modest paid spend plus organic traffic can still run a meaningful program, just at a slower cadence than a high-traffic page.
What do you need from us?
Edit access to the landing page, or a developer we can brief. Analytics and heatmap tools if available. And patience to let each test reach a valid sample size.
How do you report on performance?
Monthly reporting on which variant won and by how much, split by device and traffic source where volume allows, plus visibility into the test backlog.