PDP Grader Walkthrough: The 18-Point Product Page Audit We Run Before CRO Work

The audit that sits behind every CRO engagement we take

Before any A/B test, before any new hero image, before any price test, we grade the PDP. Across 60 plus D2C engagements, the pattern is consistent. Pages scoring 12 or below on our 18-point rubric do not need testing. They need fixing. Pages scoring 15 plus are in testing territory where learning velocity matters more than feature adds. The audit is how we sort the two.

TL;DR ▸ 18 points across five zones: above-the-fold, proof, merchandising, trust, and technical ▸ Score a PDP in 20 minutes, compare against the top 10 revenue pages, fix patterns not individual pages ▸ Below 14 out of 18, fix before testing. Above 14, test variants to squeeze more lift ▸ The framework is binary per point so there is no fuzzy scoring

How the rubric works

Each of the 18 points is binary. You either pass or you fail. No half-credit. We standardize scoring this way because subjective weighting destroys comparability across pages and across auditors. The tradeoff is that borderline-pass items look identical to clear-pass items in the final score, but we would rather lose that nuance than spend a meeting debating it.

The five zones break down as: above-the-fold (5 points), proof (4 points), merchandising (4 points), trust (3 points), technical (2 points). The weighting is deliberate. Above-the-fold and proof together are half the score because they drive the majority of add-to-cart decisions. Technical is intentionally small because fundamentals like Core Web Vitals belong in a separate performance optimization workstream.

Zone one: above-the-fold (5 points)

Point 1: Primary product image shows the hero use case within 0.5 seconds. Not a 360 view. Not a swatch selector. The single image a customer needs to understand what the product is and why they would buy it. Fail conditions: low-contrast product on white background, overly stylized shot where the product is 20 percent of the frame, gallery carousel that loads images after 1 second.

Point 2: Product title is specific enough to disambiguate. "Merino Crew" fails. "Merino 200 Crew in Charcoal, Men's" passes. A buyer should not have to scroll to confirm they are looking at the right product.

Point 3: Price, variant state, and add-to-cart are visible without scrolling on mobile at 390 pixels wide. The most common fail. Stores pack too much into the hero area and push add-to-cart below the fold on mobile. Measure this on a real device, not DevTools emulation.

Point 4: One-sentence value prop exists above the fold. Not a paragraph. Not bullet points. A single sentence that answers "why this product." "Lightweight merino base layer engineered for multi-day use without odor." Fail if you rely on the product title to do the value-prop work.

Point 5: Shipping threshold, return window, or guarantee is visible above the fold. The single most common missing element. Buyers decide whether to scroll based on whether the purchase risk feels manageable. "Free shipping over $75 and 60-day returns" in a visible strip passes. A policy footer link does not.

Zone two: proof (4 points)

Point 6: Star rating and review count visible above or immediately at the fold. Not in a sidebar tab. Not in a section that loads lazily. The stars must be rendered in the first paint.

Point 7: At least three recent reviews visible without a click. A scrollable horizontal list is fine. An accordion that requires a tap is not. Freshness matters: reviews from the last 90 days rotate to the top.

Point 8: At least one user-generated photo or video in the review section. Text reviews alone convert, but UGC photo reviews convert substantially better. If your review app does not support photo uploads, that is the fix. The product page CRO patterns post covers the UGC layout patterns we see work.

Point 9: Third-party trust signal relevant to the category. For supplements, this is FDA registration or third-party testing. For apparel, it is material certifications. For electronics, it is warranty terms. Generic trust badges (Verified Reviews, Secure Checkout) do not count because buyers pattern-match past them.

Zone three: merchandising (4 points)

Point 10: Variant selector shows in-stock availability per variant without a click. Buyers should not have to click size six to learn it is sold out. Disabled styling or a "Low stock" badge counts. A silent dropdown that reveals availability only on selection fails.

Point 11: At least one cross-sell or bundle option is relevant to the primary product. Relevance is the key. A cart-level "you might also like" carousel does not count here. On-PDP bundling with pricing visible does count. The bundling UX has to be frictionless because most attempts tank AOV rather than lift it.

Point 12: Product description has at least two distinct content blocks beyond a marketing paragraph. Examples: ingredients list, size chart, care instructions, spec table, fit notes. One big paragraph fails even if it is well-written. Buyers scan by information type.

Point 13: Imagery includes one scale or context shot. Product on a model, product in a hand, product in a room. A buyer needs to calibrate size and use case visually. Pure product-on-white galleries fail here even if they are gorgeous.

Zone four: trust (3 points)

Point 14: Return policy is explicit and linked from the PDP. Not buried in the footer. Not hidden behind a generic FAQ link. A one-line mention with a link to the detail page.

Point 15: Shipping cutoff time and delivery window are stated for the user's region. "Order in the next 3 hours for delivery by Friday" is the best case. "Standard shipping 3 to 5 business days" is acceptable. "Shipping calculated at checkout" fails because the buyer cannot price the decision.

Point 16: Brand story or credential is one click away from the PDP. A linked About section, a Why We Started block, a founder bio. Not because the PDP itself needs to tell the story, but because brand context reduces return rates and increases repurchase.

Zone five: technical (2 points)

Point 17: Largest Contentful Paint is under 2.5 seconds on mobile. Tested on real-world throttled connections, not Lighthouse lab scores alone. PDPs that fail this convert measurably worse because the hero image loads after the buyer has already started scrolling.

Point 18: Product and BreadcrumbList schema are valid with no duplicates. Covered in depth in the product schema for Shopify and Shopify breadcrumb schema posts. A PDP that fails Rich Results Test fails here regardless of visible stars.

The scoring rubric

The rubric is unforgiving on purpose. Stores that come to us with plateaued conversion rates usually score in the 10 to 13 range. They are testing the wrong things. The fix is not more tests. The fix is closing the gaps that show up in the audit first.

How to run the audit in practice

Print the 18-point checklist. Open your top revenue PDP on mobile in an incognito window. Walk through the points in order, scoring each as pass or fail. Do not argue with yourself on borderline items. The rubric is designed to be strict, so err toward fail when unsure.

Do this for your top 10 revenue PDPs and top 3 traffic PDPs. Tally a heatmap of which points fail most often. Those are your patterns. Fix the patterns at the template level, not page by page. A single fix to the variant selector logic in the theme propagates across the whole catalog.

Our CRO engagements begin with this audit and a theme-level fix list. The CRO services scope typically covers the template work in week one and then moves to variant testing in week two. The ecommerce CRO checklist for 2026 covers the broader template and funnel patterns if you want to extend the audit beyond PDPs.

What this audit deliberately does not cover

Three zones the 18-point audit skips. Cart and checkout friction, covered by the checkout friction audit. Post-purchase upsell flows, which deserve their own rubric. And SEO content depth, which is Shopify PDP SEO guide territory. Keeping the rubric PDP-specific is what makes it fast and comparable across pages.

The grader also skips aesthetic judgment. Two pages can score identically on all 18 points and feel radically different visually. Aesthetic quality matters, but it is not auditable on a binary checklist. That is brand work, not CRO work, and it belongs in a different conversation.

When a low score is actually fine

One exception to the rubric. Pages selling a single hero product at high AOV can legitimately score lower on merchandising points because they are not trying to cross-sell. A $400 single-SKU DTC mattress PDP does not need cross-sells. In that case, exclude points 10 and 11, adjust the denominator to 16, and rescore. The rubric is a tool, not a religion. Use judgment for the 5 percent of cases where the structure does not fit.

What to do this week

▸ Print the 18-point rubric and walk through your top 5 revenue PDPs on mobile in incognito ▸ Build a heatmap of which points fail across those 5 pages to identify template-level patterns ▸ Pick the top 3 failing points and scope a theme-level fix rather than page-level patches ▸ Validate fixes against the Rich Results Test and Core Web Vitals before declaring them done ▸ Re-score the PDPs after fixes and confirm they hit 14 plus before starting variant tests ▸ Share the heatmap with merchandising and brand teams so content production prioritizes the right blocks