Everything on this page is instrumented. I write the experiment code that runs on your live revenue pages — flicker-free, QA'd across browsers, and it reverts clean.
01 // selected work
experiments · builds · writeups
shipped to production
Selected experiments showing the kinds of problems I solve — SPA-aware UI, server-load mitigation, enterprise platform quirks, and accessibility-compliant variants.
Large UK home improvement retailer
Vertical search overlay had low engagement. F-pattern reading suggested a landscape layout could lift SRP progression.
Built a darkened focus-state overlay, repositioned the search element to a landscape layout, and added search-term-driven subcategory chips for Variant 2. Maintained existing product-suggestion logic to isolate layout as the variable. Two variants shipped sitewide on desktop.
Read the case study →Why your GA4 split never matches the platform
GA4 shows 60:40 when the testing platform shows 50:50. Random, unstable, never matches — and the experiment code is rarely the actual cause.
Walks through the four root causes of GA4/platform Sample
Ratio Mismatch on SPAs — data layer timing, beacon loss, virtual pageview
context, bot traffic — and the Tealium/GTM tracking pattern (with
utag.link intercept and payload mirroring) that fixes most of
them.
abtestrig — a CRO-specific build tool, open source on GitHub
Every CRO dev rebuilds the same Gulp/webpack rig. Hardcoded loader-script IDs, 8–15s rebuilds, page reloads on every CSS tweak, and no way to test interactions between two live experiments before shipping.
Built abtestrig around esbuild — sub-200ms builds, a loader
that auto-discovers the active experiment and port (no script edits between
experiments), CSS hot-swap via a stable <style> slot so
the activated state is preserved, stack-mode that runs multiple experiments
in deterministic order for conflict testing, and a shared @lib
helper library so pollerLite/waitForElement stop
getting copy-pasted between projects.
UK ground engineering / residential subsidence
Geobear runs a HubSpot CMS estate of 50+ pages across several brand variants, all powered by the same theme. The brief: a new residential landing page with a different layout and modules — without putting the 50+ live pages at risk.
Cloned the live theme into a parallel build via
hs cms fetch / hs cms upload, then shipped nine custom
modules from scratch — sticky nav, hero with background video, case carousel,
get-advice flow, logos, testimonials, video gallery, CTA banner, and a Leadoo
form wrapper. Deferred Swiper init via IntersectionObserver, a
ResizeObserver shim for sticky-nav padding under HubSpot's
overflow: hidden wrappers, CSS custom properties for the design
system, and inline schema.org JSON-LD (HowTo,
ItemList, Review) that passed the Rich Results
Test.
Large UK home improvement retailer
Google Shopping landers were exiting PDPs and re-entering through Shopping multiple times when the specific product didn't fit. Top-50 power-tool SKUs by click volume were the priority.
Built a sticky "follow-me" similar-products carousel
scoped to the product's category. Scrapes the breadcrumb-derived PLP at a
price-bound API call, caches results in sessionStorage so
subsequent PDPs in the same category render instantly. V2 layers a brand
filter for brand-loyal users.
UK builders' merchant
Trade prices were rendered separately from retail prices on PLPs and PDPs, so non-account users couldn't easily compare the two and weren't motivated to create a trade account. The hypothesis: making trade prices visually dominant alongside retail prices would lift trade account creation.
Replaced the host's separate trade-price block with a unified price banner showing retail and trade side-by-side, gated to logged-out users via cookie check. Banner click opens an embedded login modal using the host's own identity SDK, with a separate "I don't have an account" CTA routing to signup.
Read the case study →UK home insurance provider
Internal data showed bundled "Gas Combo" products had lower year-on-year renewal rates than single-category cover, dragging long-term profitability. Stakeholders wanted to surface them on a separate page so single-category products got cleaner focus.
Removed bundle items from three comparison PLPs by URL match, restructured the grid from 3-up to 2-up, destroyed and re-initialised the Swiper instance to match the new slide count, and dynamically injected a new nav link to the dedicated bundle page once a user had visited a comparison PLP.
Read the case study →Global cosmetics retailer (Romania)
Off-boarding from the UK site cut the planned social-proof test runtime short. The team needed a market that could deliver significance inside a tight timeframe.
Built a 19-product bestsellers carousel with desktop
dropdown and mobile bottom-sheet variant pickers, integrated with Avon's
internal product API and CartServiceModule for inline add-to-bag,
plus IntersectionObserver-driven impression tracking on both control and
variant.
UK luxury flooring retailer (Magento)
Customer gallery sat on a separate page where most homepage visitors never reached it — a high-trust social proof asset locked behind a navigation step.
Live-fetched the customer gallery into the homepage, built a product modal that scrapes form data from PDPs to power inline AJAX sample ordering, and surfaced Magento's native success/error notifications by cloning the host's existing UI.
Read the case study →Large UK home improvement retailer
Plain PLP exit rate of 14.4% across organic and paid channels — users were leaving without finding a product that matched their price or criteria. Returning users exited at 15.1%.
Built a custom ExitIntent detection class
(mouse trajectory, inactivity, tab-switch heuristics), an accessible modal
with focus trap, and a V2 carousel that fetches category-relevant clearance
products at price ranges derived from the breadcrumb and SKU scrape.
02 // how i work
hand-written test code
QA'd · accessible · reverts clean
Comfortable with Tealium, GA4, Adobe Analytics and most consent-management providers. Familiar with Magento, Shopify Plus, Next.js, and AEM host stacks. Also build custom HubSpot CMS modules (HubL) where the brief calls for it.
03 // clients & partners
retailers · insurers · DTC
direct + via agencies
Experiments and builds shipped for retailers, insurers, charities, and DTC brands — both directly and through agencies and CRO partners.
// experiments shipped for
// agency & direct partners
04 // field notes
deep dives on the bugs
that come up on real builds
Writeups from real client work on the specific failures that come up building experiments on modern stacks, and how I fix them.
05 // get in touch
open to new projects
UK + US business hours
For project enquiries, overflow capacity, or to discuss a specific test idea. I respond within one UK business day.