Turn a paid landing page's own funnel numbers into a ranked CRO audit, not a generic checklist

You are auditing the conversion path for ONE specific landing page that is currently receiving paid traffic — not doing a general "best practices" pass, but diagnosing this specific page against its own numbers.

CAMPAIGN CONTEXT
Google Search, exact-match on "project management software for agencies", $80/day, running 3 weeks.

LANDING PAGE AND OFFER
/agencies-landing — gated 14-day trial signup form, no credit card required.

FUNNEL NUMBERS SO FAR
2,400 clicks -> 2,100 landing views -> 340 form starts -> 96 submits (14% landing-to-start, 28% start-to-submit).

TRAFFIC SOURCE BEHAVIOR
Avg session 38 seconds, 71% mobile, 62% bounce on landing view.

SUSPECTED FRICTION POINT
Team suspects the pricing mention in the hero contradicts the ad's "free trial" framing.

HOW TO AUDIT
Start from the numbers, not the page layout — if funnel_numbers shows a specific step where visitors drop (landing view to form-start, versus form-start to submit), diagnose that step first and say explicitly why the other steps are lower priority given the data, rather than auditing top-to-bottom by page section regardless of where the actual leak is. Distinguish problems caused by a mismatch between the ad's promise and the landing page's message — traffic arriving with the wrong expectation — from problems caused by the page failing visitors who arrived with the right expectation; these need different fixes, and conflating them wastes a test cycle on the wrong lever. For every issue you flag, state your confidence based on the numbers given: call out plainly when something is a plausible hypothesis versus something the funnel data actually supports, since a few hundred paid clicks cannot support ten confident claims at once. Do not recommend a redesign as the first move for a problem a single element change (headline, form field count, CTA copy) could plausibly fix — propose a full redesign only when the numbers show broad, non-localized underperformance across every step rather than one clear leak point.

WHAT NOT TO DO
Do not produce a generic CRO checklist (page speed, mobile responsiveness, above-the-fold CTA) unless something in the context actually suggests that item is the issue here — every recommendation must trace back to a specific number or observation given above. Do not recommend more than one first test; prioritization is the entire point of this audit, not a menu of options.

OUTPUT FORMAT
1. The single highest-priority fix, with the funnel number that justifies it and the mechanism for why it should move that number.
2. Two to four secondary issues, each tagged [ad-message-mismatch] or [page-execution], ranked by likely impact.
3. One explicit "not enough data to say" item — a common suspect you're declining to diagnose because the numbers given don't support a verdict either way.
4. A single next A/B test recommendation naming the exact variable to change and the metric that would confirm the fix worked.