Pages that turn traffic into revenue.
Landing systems, CRO, and funnels — hypothesis-driven testing on Shopify, WordPress, or custom stacks. AI-assisted variant generation, Core Web Vitals discipline, and the friction work that lifts revenue without lifting spend.
The market reality in 2026
The Core Web Vitals goalposts moved — INP (Interaction to Next Paint) replaced FID as a Core Web Vital in 2024, and Google now weighs it heavily for both ranking and ad-quality scoring. Mobile commerce is the majority of traffic on most sites, and the friction tolerance on a phone is approximately zero. One-click checkout (Shop Pay, Apple Pay, Google Pay) and persistent-session experiences are table stakes, not differentiators.
On the testing side, AI-assisted variant generation has changed the economics of CRO: variants that used to take a week of design + copy can be drafted in an hour, and the testing constraint has shifted from “can we generate enough variants” to “can we ship them without breaking statistical hygiene.” Edge personalization (Vercel Edge, Cloudflare Workers) has become accessible to mid-market teams, opening up routing experiments that used to require a six-figure investment. The brands compounding conversion are running more tests, faster, against cleaner data — not running fewer tests harder.
How I deliver conversion work
-
01
Audit the funnel, end to end
Session recordings, heatmaps, scroll depth, INP / LCP / CLS measurements, abandonment patterns by step. The audit produces a friction list — every drop-off ranked by lost-revenue impact, not by how cool the fix would be to ship.
-
02
Hypothesize against predicted lift
Each friction item gets a written hypothesis with a predicted lift range, a measurement plan, and a minimum sample size for statistical significance. Variant ideas are scored on impact × effort and tied to a primary metric that maps to revenue, not engagement vanity.
-
03
Test with statistical hygiene
A/B framework with significance built in — no “this variant is winning after 2 days” calls. Multivariate where traffic volume allows. Bayesian or frequentist depending on the test design; whichever, the methodology is locked before the test ships. Every variant tracked, every interaction quantified.
-
04
Compound the wins
Winning variants become the new control. Losing variants get a written autopsy so the next iteration learns from them. The conversion roadmap is updated quarterly. The Core Web Vitals score is monitored continuously — a regression there silently undoes ten test wins.
The stack
Shopify · WordPress · Webflow · custom Next.js / Astro builds · Headless commerce (Shopify Hydrogen, Saleor)
GrowthBook (open-source) · VWO · Optimizely · AB Tasty · Vercel Edge Config for routing · Cloudflare Workers for edge personalization
Hotjar · Microsoft Clarity · FullStory · Mouseflow · user-testing platforms · on-site survey tools
Figma for design variants · Adobe Firefly / Midjourney for visual variants · Claude / ChatGPT for copy drafts (human-edited) · PageSpeed Insights + WebPageTest for the speed check
What gets measured
- Conversion rateoverall + per funnel step
- Revenue per visitor (RPV)the metric that matches up with media-spend math
- Session valuetracked against landing-page variant + traffic source
- INP / LCP / CLSCore Web Vitals on real traffic (CrUX), not lab averages
- Test win rate% of tests that produce a statistically significant lift
- Time-to-decision per testdays from launch to call — tracks methodology health
Proof from the work
A boutique resort property
Booking-flow CRO + mobile-first funnel rebuild + WhatsApp follow-up shortened the booking-decision window and reduced OTA dependency.
3.4× direct bookingsA premium D2C tea brand
Shopify CRO + checkout-flow optimization + first-purchase economics tuned for retention. AOV climbed alongside conversion rate.
2.9× revenue liftA real-estate & lifestyle brand
Landing systems + form-friction reduction + lead-quality scoring closed the gap between paid traffic and signed walk-in visits.
3.2× qualified inquiriesCommon questions
How much traffic do I need for A/B testing to work?
Approximately 1,000 conversions per month per variant for clean test cycles in 2–4 weeks. Below that, testing still works — the cycles just stretch (or we move to MVT, sequential testing, or Bayesian approaches with informative priors). For pre-product-market-fit brands, the right move is often research first, testing second — qualitative session reviews and on-site surveys produce better learnings than under-powered A/B tests.
What about AI-generated landing pages?
Useful for variant scale, dangerous as the source of truth. AI is great at producing the tenth headline variant or the seventh hero-image option for a test that’s already designed. It’s not great at writing the actual offer page on which the test runs — that’s a positioning problem, not a generation problem. We use AI heavily in the variant production pipeline, not at the strategy layer.
Do you build the pages or just optimize them?
Both. Engagements typically include strategy + design direction + build (in Shopify, WordPress, Webflow, or via your in-house dev team). For brands with established product teams, I’ll work with your developers and own the strategy / measurement / testing layer. For brands without that capacity, the build is part of scope.
What’s a realistic conversion lift?
Varies wildly by starting point. Brands with mature funnels see 10–30% lift per quarter compounding. Brands with broken funnels can see 2–5× in the first 90 days just from fixing the obvious leaks. The honest answer in a first conversation: I’ll give you a real range after the audit, not before.
Ready to plug the leaks in your funnel?
First 30 minutes are on me — you’ll leave with a clearer picture of where conversion is leaking and what the highest-leverage first test would be.