Verified against ChatGPT · 2026-08-11
Audit a product page for the specific reason it's leaking buyers, not a generic best-practices checklist
Runs a structured conversion audit on one product detail page, diagnosing the most likely specific reason it underperforms against the given funnel data instead of returning a generic PDP best-practices checklist.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Audit one product detail page for conversion issues. The output must diagnose the most likely specific cause of underperformance given the actual funnel numbers below — not a generic 'add trust badges, add urgency, improve photos' checklist that would apply to any PDP.
PRODUCT PAGE CONTENT
Title, 4 product photos (no size-comparison shot), price shown without mention of shipping cost until checkout, no size guide link, 12 reviews averaging 4.2 stars.
FUNNEL DATA
View-to-add-to-cart rate is 8.5% (roughly in line with category average), but add-to-cart-to-purchase is only 19% versus a 34% site average.
WHAT'S ALREADY BEEN TRIED
Added a free-shipping banner site-wide two months ago; add-to-cart-to-purchase rate didn't move.
TRAFFIC SOURCE
70% from a paid Instagram campaign showing the product on a model, cold audience, first-time visitors.
STEP 1 — LOCATE THE ACTUAL DROP-OFF
Using the funnel data, identify exactly where in the page (or the flow) the biggest relative drop happens — view-to-add-to-cart, or add-to-cart-to-purchase — since the fix for one is usually different from the fix for the other, and a generic audit that addresses both equally usually fixes neither well.
STEP 2 — DIAGNOSE, DON'T JUST LIST SYMPTOMS
Given where the drop-off is and the actual page content, name the one or two most probable specific causes — reference actual content on the page (a specific missing spec, an unclear price break, a shipping cost surprise) rather than generic categories like "trust" or "clarity." Cross-check against what's already been tried — do not recommend something already attempted unless you have a specific reason the prior attempt likely failed and a different approach would work.
STEP 3 — TRAFFIC-SOURCE CONTEXT
Consider whether the traffic source changes the diagnosis — a page converting poorly from paid social cold traffic often has a different root cause (unmet expectation set by the ad) than the same page converting poorly from returning-customer email traffic.
STEP 4 — RECOMMENDATION
Give exactly the top 2 fixes to test first, each with the specific change and the specific reason it addresses the diagnosed cause — not a long list of every possible PDP improvement.
WHAT NOT TO DO
Do not include generic PDP best practices ("add more reviews," "use high-quality images," "add scarcity messaging") unless the funnel data and page content actually point to that specific gap.
OUTPUT FORMAT
Four labeled steps as specified, ending with the top 2 prioritized fixes.Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Requiring the audit to locate the specific drop-off stage before diagnosing anything counters GPT-5.1's default behavior on CRO requests, which is to return a comprehensive best-practices checklist spanning the entire page (photos, trust signals, urgency, shipping) regardless of where the actual leak is — that checklist is defensible in the abstract since every item is plausibly good practice, but it wastes testing effort on stages that aren't actually the problem, and a 19% add-to-cart-to-purchase rate against a 34% site average is specifically a checkout-stage or price-surprise problem, not a browse-stage photo problem, which the funnel data makes diagnosable if the model is forced to actually use it rather than treat it as color commentary. Cross-checking against what's already been tried, with the requirement to explain why a repeated fix might now work differently rather than just re-suggesting it, prevents the model from recommending the free-shipping banner again in slightly different words — a generic CRO checklist has no memory of what's already failed, so without this explicit instruction the model will happily re-suggest an already-disproven fix because it's a category-correct suggestion, even though it's specifically known not to work here. Factoring in the traffic source addresses a real and specific CRO pattern: a page underperforming specifically for cold paid-social traffic is frequently an expectation-mismatch problem between what the ad promised and what the PDP delivers, which is a different root cause than the same conversion metrics would suggest for warm, already-familiar traffic, and conflating the two produces a diagnosis that only fits one segment.
What you get back
Step 1: the leak is add-to-cart-to-purchase (19% vs 34% average), not view-to-cart. Step 2: most likely cause is the shipping cost surprise at checkout — the free-shipping banner change didn't address it because shipping cost isn't disclosed until the final step regardless of the banner's presence elsewhere on the site. Top fix: surface actual shipping cost or free-shipping threshold directly on the PDP near the price, not just in a site-wide banner.
Verified against
ChatGPT GPT-5.1 · 2026-08-11
Changelog
- 2026-08-11 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — Google Ads management, built and maintained for you — that's Scult's day job.
EXPLORE GOOGLE ADS MANAGEMENT
