CRO: structured experimentation, not vibes
Ecommerce CRO is a research process: quantitative funnel analysis, qualitative user research, hypothesis backed by evidence, correctly sized A/B tests, and honest guardrails. Anyone who talks about tests before they've done the research is selling you vibes.
The average Shopify or Woo store leaves 30-60% of revenue on the table between add-to-cart and thank-you. Finding that revenue is not about the latest checkout hack it's about the right test, run at the right sample size, on the right page, in the right sequence.
This hub is the CRO stack we run against every account: research → hypothesis → build → measure → next. Sub-topics cover PDP, cart, checkout, PLP, personalisation and the tooling that makes it repeatable.
Guides, playbooks and case studies
Heatmaps, session replay, on-site polls, GA4 funnels what to measure before you test.
Hero image, price framing, USP strip, sticky ATC, size guides, review pattern.
Drawer vs page, express payments, upsell offers, shipping thresholds.
Guest-first, address autocomplete, error handling, trust markers, VAT clarity.
Filter UX, out-of-stock handling, badge system, sort defaults.
Segment-based hero, category-based recommendations, returning-visitor journeys.
Convert vs VWO vs GrowthBook the trade-offs at each stage.
How to run 6-10 winning-or-learning tests per quarter without breaking the site.
The playbook, step by step
- 01Baseline the funnel
GA4 exploration + Shopify/Woo funnel export. Find the biggest drop-off in absolute revenue, not percentage.
- 02Layer qualitative
Heatmap + session replay + on-site poll on the two worst-performing steps. Look for confusion, not just clicks.
- 03Write the hypothesis
'Because X, we believe Y will move metric Z by W%.' No hypothesis, no test.
- 04Size + guardrail
Pre-calculate sample size for a realistic MDE. Set guardrails on revenue-per-visitor, refund rate, and mobile bounce.
- 05Ship + document
Winning tests roll out, losing tests are documented in the experiment log. The log is the moat it's how tests get faster over time.
Frequently asked questions
How many tests do we need per quarter?
6-10 well-designed experiments per quarter is a healthy cadence for a £100k-£1m/month brand. Volume above that adds cost faster than it adds learning.
What's a good conversion rate?
Depends on category, price and traffic mix. Median UK Shopify ecommerce is ~2.1% (Littledata 2025). Anything above 3.5% is strong; above 5% is exceptional.
Should we do CRO before we scale ads?
Yes. A 15% CRO lift multiplies against every future pound of ad spend. Skipping CRO to scale ads first is buying revenue at a permanent discount.
Which tool should we use?
GrowthBook (open-source, cheap) up to ~£300k/month, Convert or VWO past that. Optimizely and Adobe Target only if you already run enterprise stacks.