Prelander Funnels Built for the Click and the Sale

A prelander is judged on two numbers multiplied together, and almost everyone reports the first one alone. Click-through to the offer rising 40 percent while conversion on the offer falls 45 percent is a page that has cost you money and looked like a win in the weekly deck. I build funnels against the product of the two, measured end to end through your tracker, and I will show you the losing version of that pair from my own work if you want to see what it looks like before you commit. The build: Hand-written HTML and CSS, deployed as static files behind Cloudflare or on your own VPS. No page builder. Unbounce, Instapage and the drag-and-drop templates in the affiliate tooling all add a redirect hop and 300 to 900 kilobytes of script before your copy renders, and on the in-app and push traffic these funnels exist to serve, that is where the click was lost. Target is a largest contentful paint under 1.8 seconds on throttled 4G on a mid-range Android device, total page weight under 180 kilobytes including images, system font stack, no web fonts, no tag manager. Images as AVIF with WebP fallback, sized to the breakpoint rather than to the desktop. Angles, not decoration: Three to five funnels, each built on a different reason the visitor is not clicking — price disbelief, an unfamiliar brand, a subscription they think they cannot cancel, a form they assume is long. The copy answers the objection in the first screen. I write it; I do not brief it out. For app funnels the page has to survive the store review that follows, so claims on the prelander match the listing text and the paywall screenshot matches the paywall. Compliance, since it is where these get expensive: EU and Irish traffic means consent before any non-essential script fires, and the funnel is built so nothing needs one. Claims go through the ASAI code for Ireland and the CAP code for the UK, and comparative pricing has to be evidenced on the page or it comes off. Health and supplement claims I refer out rather than write. What I will not build: Fabricated countdown timers, invented stock counters, comment sections populated with people who do not exist, fake news layouts, celebrity endorsements that were never given, or any page whose second version differs from the one the network reviewed. Not because they fail to lift click-through in the short term, but because they raise refund and chargeback rates on the offer behind them, and the advertiser eventually reads that data and cuts the account that supplied it. Handover: Source files in a repository you own, with the tracker parameters, postback wiring and the split-test configuration documented in the same repo. A short recorded walkthrough for whoever maintains it. No hosting lock-in, no licence, no per-page fee afterwards. Who this suits and who it does not: It suits a team with traffic already flowing and a conversion rate they can no longer explain. It does not suit anyone who needs the page next Tuesday — a set of angles that were argued about takes two to three weeks, and the first week is spent reading your session recordings and refund reasons rather than designing. It also does not suit a brand whose legal team requires sign-off on each headline variant; the test cadence this work depends on cannot survive a two-week approval cycle, and I would be charging you to sit still.