The 11-second rule: what people actually read above the fold
Eye-tracking on twelve client sites, and the pattern that held every time.
All servicesHover to preview
Case studies by disciplineHover to preview
13 case studies, with numbers reconciled against each client's own reporting.
Open the case indexThirty articles, each written by whoever actually ran the work — with the numbers, the tests that failed, and the parts we would do differently.
The archive
Eye-tracking on twelve client sites, and the pattern that held every time.
You do not need forty components. You need the eleven you keep rebuilding.
We removed it on six sites. Five got better. Here is what separated them.
Every page gets a job description before it gets a layout.
Passing AA while keeping the palette the founder actually approved.
Nineteen questions we ask before touching a single pricing table.
Twelve months of ranking data from accounts that gained and accounts that lost.
No new content, no outreach, no budget. Just the graph you already have.
1,200 pages, one template, and the three rules that kept them indexed.
The redirect map exists before the staging site does. Here is the process.
Where the model works, where it is cargo cult, and how to tell early.
INP replaced FID two years ago. Most audits still have not noticed.
Six feed changes that moved return further than any bidding experiment.
How we proved an account was reporting revenue it did not cause.
A fixed test budget, a retirement rule, and no sentimentality about winners.
What the channel report hides, and the three queries that reveal it.
Teaching the bidder which leads actually closed, in about an afternoon.
Marginal CAC is the number. Almost nobody watches it.
The cost comparison we run before recommending either, with the maths shown.
The scoring sheet: query cost, update cadence, asset weight, real bespoke need.
Patterns your marketing team can ship without opening a support ticket.
Eleven sites on one codebase is wonderful. Three sites is usually a mistake.
A container, a subdomain, and the conversion signal your bidder deserves.
Twenty minutes that changes the shape of most rebuild proposals.
Across $40M of traffic — and the three best practices that lost every test.
What happened to completion rate, and the two fields we had to put back.
Sort by margin, not by newest. The uplift is larger than most redesigns.
Pause, swap and dunning. Get these right and retention takes care of itself.
Nine tactics, plotted against the engineering hours each one actually costs.
The revenue threshold where the complexity starts paying for itself.
Next step
Pick the audit you actually need. Within 48 hours you'll get a teardown of that channel — free, specific, and occasionally uncomfortable.
Average reply time: 3 hours · No obligation, no retargeting, promise.