Case study · CRO · Xero.com
What 25 documented experiments taught us about the first page most people see, and where it landed.
The setup
The homepage gets the most traffic on xero.com, so small changes can matter. Almost every test was an A/B against the live default, read after a 7 or 35 day maturity window.
The brand refresh strategy set the bar for a call: an 80 to 85%+ chance to beat the default, with a tight interval and no drop in the guardrails. A result near 50% is a coin flip. 90%+ is a call.
The scoreboard
25 homepage experiments, July 2025 to October 2026. Where a charter was ambiguous, the more cautious label is used.
What we learned
Mid 2025 to early 2026. New headlines, proof and messages in the hero, and how little a hero swap moves when the default is already doing its job.
We swapped the default hero for copy and visuals about the Syft integration, hoping to lift plan mix and ARPU. AU and US were flat. The UK was not.
1177-default.png in src/assets/xero1177-variant.png in src/assets/xeroCopy, colour and a second CTA were each tried against the default. None beat it. The primary CTA stays "Compare plans and pricing".
[TODO: add click-through detail from the 1293 charter]
1293-default.png in src/assets/xero1293-variant-a.png in src/assets/xero1293-variant-b.png in src/assets/xeroMoving plan mix and building trust earlier on the page.
Two tests (1252, 1282) pointed the same way: visitors who were nudged to talk to sales picked higher plans, though the results did not clear the bar to ship.
1282-default.png in src/assets/xero1282-variant-a.png in src/assets/xero1282-variant-b.png in src/assets/xero1359-default.png in src/assets/xero1359-variant.png in src/assets/xeroA reskin across the top traffic pages, then a close look at where it hurt. The biggest project on the homepage, and the one the rollout plan depends on.
A reskin, not a redesign: new imagery, updated copy and theme variations. Two weeks of A/B testing across the 3x3 markets. Goal: confirm the refresh causes no statistical drop in CMR, account activations, paying orders or average CMR.
Isolated tests on hero messaging, trust bar, Making Tax Digital framing, pricing, switch messaging and customer stories, Jul to Oct. Goal: find the best component treatments inside the new brand.
1377-default.png in src/assets/xero1377-variant.png in src/assets/xeroThe refresh held revenue steady and lifted activations, passing the safety check Phase 1 was designed for. The homepage itself varied sharply by region, which fed directly into how the rollout was sequenced.
The pricing component is where homepage visitors decide. Three tests asked whether to show it differently, recommend a plan, or aim higher.
Most visitors do not scroll far enough to see Plans Compact. We tested removing it, replacing it with the Plan Recommendation tool, defaulting to "Get one month free", or combining the last two.
1415-default.png in src/assets/xero1415-variant-b.png in src/assets/xero35d CMR. Bars capped at half scale.
Why it matters: the winning claim is a UK result, not a global one. Any rollout needs to be region-aware, or rerun in the US first.
Where the homepage is now
The hero primary CTA stays "Compare plans and pricing", every alternative lost. The brand refresh is rolling out across tested page groups, with the US homepage getting its own follow-up tests. The Plan Recommendation tool is a UK win, with a cleaner rerun recommended before going wider. Five tests were running or waiting on results at the time of writing.
[TODO before publishing: confirm what has shipped to production and when, results for 1173, 1372 round one, 1395, 1402 and 1405, and any total revenue, activation or ARPU impact across shipped changes.]