Conversion Rate Optimisation (CRO)
More revenue from the traffic you already have.
CRO built on behavioural data and structured A/B tests, not redesigns, opinions, or best-practice checklists.
Most CRO engagements start with a redesign or a list of "best practices" copied from a blog post. Real CRO starts with data: where are users dropping off, what are they telling you in session recordings, and what hypotheses does that evidence suggest? Every test I run has a documented hypothesis, a minimum detectable effect calculation, and a decision rule, so results are actionable, not anecdotal.
Why most CRO efforts produce no lasting results
Conversion Rate Optimisation fails in predictable patterns. The most common: a redesign is commissioned based on design preferences and platform best-practice checklists rather than behavioural data. The new design launches, conversion rates do not improve meaningfully, and no one knows why because there was no structured test with a control condition. The second common failure: A/B tests are run without minimum sample size calculations, results are declared after a few days when the sample is far too small for statistical significance, and a winning variant goes live based on noise rather than signal. The third: tests are run one at a time with no documented hypothesis, so when a test produces a result, it is unclear whether the finding is generalisable or specific to one audience and one moment. CRO that compounds over time starts with specific, testable hypotheses derived from behavioural data, runs each test with statistical discipline, and builds an institutional knowledge base of what the specific audience responds to.
The behavioural data foundation that makes testing hypotheses accurate
Before running any tests, the measurement infrastructure needs to be in place and producing reliable data. GA4 funnel reports need to show the full conversion path from landing page to confirmation page with drop-off rates visible at each step. Session recordings need coverage on the pages that matter most for conversion. Heatmaps need to show where users click, where they stop scrolling, and which elements attract attention relative to the CTA. Form analytics need to show which fields have the highest abandonment rates and how long users spend on the form before submitting or leaving. Without this data layer, test hypotheses are based on assumption rather than observation of what the actual audience is doing. The observation period typically requires 2 to 4 weeks of data collection at meaningful traffic volumes before the test backlog is prioritised. The patience this requires is justified by the quality of the hypotheses it produces: evidence-based hypotheses outperform assumption-based ones because they identify the actual behaviour causing the drop-off rather than a theoretical explanation of it.
Statistical significance: what it means and why running tests to completion matters
A/B test results are only meaningful when the required sample size has been reached at the target confidence level. Statistical significance means the probability that the observed difference between the control and the variant occurred by chance is below a defined threshold, typically 5 percent at 95 percent confidence. This calculation must be done before the test starts, not after the data looks interesting. The required sample size depends on the baseline conversion rate of the control, the minimum detectable effect considered meaningful for the business, and the confidence level required. Tests that end early because the variant appears to be outperforming, or because the allocated test window has expired before the sample was reached, produce results that are not reliable and often fail to hold when rolled out to full traffic. The discipline of running every test to the predetermined sample size, and only then making a decision based on the result, is what separates CRO that improves conversion rates over time from CRO that produces activity without reliable insight.
GA4 funnel analysis, heatmap review, session recording analysis, and form drop-off identification to find the highest-leverage pages.
Prioritised test backlog ranked by expected impact, test complexity, and traffic volume, so you run the most valuable tests first.
Each test designed with a clear hypothesis, primary and secondary metrics, minimum sample size, and statistical significance threshold.
Ad landing pages tested for headline, value proposition, social proof placement, and CTA copy, matched to specific ad creative and audience.
Multi-step form and checkout flow optimisation to reduce abandonment at the highest-value conversion points.
Every test result documented with lift, confidence level, and roll-out decision, creating an institutional knowledge base of what works for your audience.
- 01Data collection: GA4 funnel setup, heatmap installation, and session recording collection across the conversion-critical pages.
- 02Opportunity mapping: identify pages where a 10% lift in conversion rate would have the largest revenue impact.
- 03Hypothesis development: write specific, testable hypotheses for each page based on what the data shows users are and are not doing.
- 04Test execution: run A/B tests using a proper statistical framework, 95% confidence threshold, no peeking before the required sample.
- 05Iteration: winners become the control; losing variants are documented with learnings that inform the next hypothesis.
Ready to get started?
Book a 30-min call