Finisterre is a surf and outdoor clothing brand founded in St Agnes, Cornwall, in 2003, built around sustainability and life in and around the ocean. Its optimisation programme with Conversio was already well established on commercial questions: what improves conversion, add-to-bag, revenue per visit. Its environmental commitments sat in a separate conversation, one with no equivalent evidence base to draw on.

Neither Conversio nor Finisterre set out with a ready answer. Testing, as a discipline, is built to answer one kind of question: does this change convert better. Nobody could say with confidence whether the same discipline could answer a different one: what does this change cost or save the planet, measured at the level of a single test rather than the site as a whole. So that became the challenge we set ourselves, together. Whole-site carbon tools already existed and could describe Finisterre's overall footprint, but only as a snapshot. They couldn't account for the fact that a website's carbon footprint moves constantly: new content, new products, shifting traffic and changing page structures all change the number, every minute, on every site. None of them could isolate the environmental cost of one specific change inside a live experiment, or tell you whether that cost had gone up or down between control and variant. Before deciding what to build, we first had to find out whether anything reliable could be built at that level at all.

Starting With a Site-Wide Audit: We began by auditing Finisterre's full website to see where carbon impact actually concentrated, rather than guessing. Product listing pages stood out immediately, accounting for around 64% of site carbon emissions from just 28% of page views, caused largely by the image carousel shown on every product card on mobile. That gave the programme its first real hypothesis to test.

Finding Out If It Could Be Measured at All: This was the heart of the challenge. When a specific part of a page changes, can you isolate how much lighter that change makes it for the customer, and turn that into a carbon figure worth trusting? Working with open, peer-reviewed carbon measurement tools from the Green Web Foundation, we built an approach that looks only at what a test actually changes for the customer, not the page around it. That meant we could test a genuine hypothesis: reduce what a customer's device has to download, and you reduce the carbon cost of that visit, in a way that can be measured, compared and forecast rather than estimated.

Proving the Concept, Test by Test: We then applied that approach to three different kinds of change: a mobile product listing page test, a product page video test, and a mobile navigation test. Each one needed its own bespoke tracking, built specifically for that experience. But each one returned a carbon figure Finisterre could trust and forecast against a full year of traffic. The point wasn't that one tool could measure everything automatically. It was that the underlying approach held up wherever we pointed it, which is what makes it worth building into something bigger.

From Audit Line Item to Trusted Metric: Carbon can now be measured per test, compared between control and variant, and forecast across a year of traffic, the same way conversion or revenue can. That's a capability Finisterre didn't have before this programme, and can now use on any future test.

Optimisation as a Business Decision Tool, Not Just a Conversion Lever: Across all three tests, reducing carbon never came at the expense of commercial performance. The product listing page test increased click-through to product pages and reduced bounce rate. The navigation test held conversion and add-to-bag steady. That evidence matters because it reframes what testing is for: a way to make business decisions accountable to more than one measure at once, environmental impact included, rather than a trade-off between doing well and doing good.

Looking Ahead: Each test so far has needed its own bespoke tracking, built by hand for that specific change. The next stage of this work is turning that into an always-on capability that tracks carbon across the whole site automatically, taking into account where relevant A/B test changes are made.

Impact: The mobile product listing page test is now live on Finisterre's site, forecast to save 14,467kg of carbon over 12 months with no negative impact on conversion or add-to-bag. That number is welcome, but it's evidence, not the headline. The real result is a repeatable way to measure, trust and forecast the environmental impact of a business decision before it's made.