Shopping Ads Audit Checklist: 32-Point Diagnostic

How to use this checklist

Work the layers in order. Each layer compounds onto the one below — you cannot bid-optimize your way out of a broken feed, and you cannot structure-optimize your way out of a tracking leak. Fix upstream first.

1. Conversion tracking — you cannot optimize what you cannot measure.
2. Feed health — you cannot sell what Google cannot show.
3. Campaign structure — you cannot segment what is not split.
4. Bidding and audiences — strategy fits data maturity.
5. Competitive position — price reality vs. category.
6. Conversion funnel — post-click experience.

Layer 1 — conversion tracking audit

1. Native purchase tag as primary

The native Google Ads purchase tag attributes 10–15% more conversions to Google Ads than a GA4 import does. If the account uses GA4 as primary, Target ROAS is bidding against a deflated signal and performance trails. Switch native to primary, keep GA4 as secondary for cross-channel comparison.

2. Product ID parameter

Verify the product ID parameter is passing: Tools → Data Manager → Google tag → View Stats. Look for product or product_id events. If missing, dynamic remarketing shows random products instead of the ones the visitor actually viewed. The revenue lift from fixing this alone is usually 8–15% of remarketing spend.

3. Attribution model

Data-driven attribution enabled for every purchase conversion. Last-click is still the Google Ads default in some accounts and it systematically undervalues upper-funnel touches, making Smart Bidding cautious on acquisition campaigns.

4. Auto-tagging

Auto-tagging ON in Account Settings. Without it, GCLID does not persist and offline conversion imports break. Free fix, instant effect.

5. Conversion window aligned to cycle

6. Enhanced conversions live

Enhanced conversions recover 5–10% of signal lost to cookie restrictions. Enable via the Google tag or via the Google Ads conversion action directly. Measure lift by watching the weekly conversion count after enabling.

Layer 2 — feed health

7. Disapproval rate under 3%

Merchant Center disapprovals above 3% drag the whole account. Review the diagnostics tab, fix by theme (image, policy, missing GTIN), resubmit. Perfoads flags revenue-at-risk — the total revenue tied up in products currently disapproved or in a policy warning state.

8. Title formula

Brand + Product Type + Key Attributes (size / color / material). Generic titles (“Blue Running Shoes”) lose in auctions against structured ones (“Nike Air Max 270 Men's Running Shoes — Navy Size 10”). Top-200 products by impression deserve hand-written titles; the rest can use rule-based generation from structured data.

9. Image quality

• 800×800 minimum resolution.
• White or neutral background on the main image.
• No promotional overlays, badges, or watermarks (Google strips impressions for these).
• Lifestyle images allowed as additional, not primary.

10. Google product categories correct

Wrong taxonomy costs impressions. Use the official Google product taxonomy, match as deep as possible (level 4 or 5), and custom labels for your own internal segmentation.

11. Custom labels for segmentation

At minimum, label products by margin tier (high / mid / low) and velocity (winner / scale / fix / cut). These labels drive the campaign structure in Layer 3.

Layer 3 — campaign structure

12. Shopping vs PMax split logic

PMax catches most traffic in most accounts now, but Standard Shopping still earns a role for brand defense and for SKUs that need manual-bid precision. A healthy account has: PMax on the main catalog with brand exclusions, a Standard Shopping brand campaign capturing branded demand, and Standard Shopping on the top-margin SKUs if PMax is under-bidding them.

13. PMax brand exclusions enabled

Settings → Brand Exclusions. Without this, PMax will cannibalize Search brand campaigns and report the conversions as PMax wins. Not 100% effective, so audit PMax search-term insights monthly for brand leakage.

14. PMax asset groups segmented by category

One asset group per major product category. Avoid the single-asset-group-for-everything pattern — it is the Shopping equivalent of the mixed-brand non-brand campaign. Separate creative per group, separate audience signals per group.

15. Audience signals per asset group

PMax audience signals are guidance, not gates, but richer signals learn faster. Add: customer match list, cart abandoners, product viewers, in-market for the category, custom segment if you have competitor data.

16. New-customer acquisition settings configured

The PMax “new customer” setting can be: bid higher for new customers only, or both. Match to the goal. Acquisition-focused accounts should bid higher for new customers explicitly, not leave the setting neutral.

17. Standard Shopping priority-layering

High priority on non-brand Shopping with brand terms as negatives. Low priority on brand Shopping. Queries fall through cleanly. This is Shopping's equivalent of Search brand/non-brand separation and the same logic applies.

18. Search network disabled in Shopping

Shopping campaigns occasionally have Search Network partners enabled by default. Disable unless you have measured and it earns its spend. Usually it burns budget on lower-intent traffic.

19. Data fragmentation check

Warning signs: 15+ campaigns splitting a limited budget; campaigns with under 15 conversions per month; duplicate targeting across campaigns; keyword cannibalization. Smart Bidding needs conversions per campaign to optimize — the threshold is around 30/month per campaign for Target ROAS to work reliably.

Layer 4 — bidding and audiences

20. Bid strategy matches data maturity

21. Target ROAS aligned with margin

Target ROAS without margin awareness produces the classic trap: the algorithm hits the ROAS number by bidding up on high-AOV low-margin SKUs, inflating revenue while margin falls. Fix by sending margin-adjusted conversion value (AOV times margin %) instead of raw AOV.

22. Dynamic remarketing live

Product-level remarketing with the product ID parameter. Excluded: recent purchasers (0 to 30 days depending on category). Included: cart abandoners with tighter recency bands.

23. Customer Match uploaded

Email list of purchasers uploaded. Match rate should be 30–70% — lower suggests incorrect email formatting or bot-heavy lists. Similar audiences enabled for lookalike expansion.

24. Audience segmentation

• Cart abandoners (0–7 days): high priority, aggressive bids.
• Product viewers (0–30 days): medium priority, moderate bids.
• Past purchasers (30–180 days): cross-sell / upsell campaigns.
• High-value customers: VIP treatment, exclusive offers.

25. Lost impression share

• Lost IS (Budget) >20%: increase budget on that campaign.
• Lost IS (Rank) >30%: improve ads, bids, or ad strength.

Layer 5 — competitive position

26. Price benchmarking against category

Pull the Price Competitiveness report (Merchant Center → Reports). Identify products flagged as “more expensive than competitors” — these are usually the auction-losers driving low impression share on otherwise healthy SKUs. Also identify under-priced products: revenue opportunity by raising price without losing position.

27. Overpriced / underpriced quadrant

Build the quadrant: high volume overpriced (fix pricing or reduce bids), high volume underpriced (raise price, watch impression share), low volume overpriced (cut from campaigns), low volume underpriced (scale).

28. Shipping and return policy signals

Free shipping and easy returns show as badges in Shopping listings. Both increase CTR measurably. If offered, they should be in the feed.

Layer 6 — conversion funnel

29. Mobile Core Web Vitals

LCP under 2.5s, INP under 200ms, CLS under 0.1 on the top 20 product pages by impression. A 4-second mobile LCP will blow out conversion rate no matter how good the feed is. Run PageSpeed Insights for the top 10 landing pages.

30. Cart and checkout friction

Guest checkout available. Shipping cost shown before payment step. Minimum form fields. Payment methods match the market (BLIK in PL, Klarna in DE, Apple Pay everywhere).

31. Landing page to ad match

For each top-spend product in Shopping, verify the landing page matches the ad promise — price, availability, promotional claims. Mismatches produce abandonment that looks like feed or bid problems but is really post-click.

32. Cross-device tracking

Users research on mobile, purchase on desktop. Enhanced conversions + proper login flow stitch these touchpoints. Without stitching, mobile looks unprofitable and desktop looks organic. Budget gets misallocated.

Priority fixes — the order you should work in

Most audits surface too many findings to fix at once. Work in this order:

1. Tracking first. Any item from Layer 1 that is broken. Without accurate signal, every downstream optimization is noise.
2. Feed disapprovals next. Revenue literally tied up and unshowable.
3. Brand exclusions and PMax structure. Stops cannibalization that inflates PMax and deflates Search.
4. Margin-adjusted bidding. Biggest long-run lever once data is clean.
5. Audience layering. High-ROI retargeting adds, easy to ship.
6. Post-click friction. CWV and checkout. Compounds with the above.

The five most common fixes we surface across ecommerce audits — present in roughly two-thirds of accounts we see — are: GA4-as-primary tracking, missing brand exclusions on PMax, no margin adjustment in conversion value, dynamic remarketing not firing on the right audience, and mobile LCP over 4s on top products.

Key takeaways

• Upstream first: tracking, feed, structure — fix these before you touch bids.
• Native Google Ads purchase tag as primary; GA4 as secondary. 10–15% of attribution is at stake.
• PMax without brand exclusions cannibalizes Search brand and distorts channel attribution.
• Target ROAS without margin adjustment optimizes the wrong number.
• Audience layering (cart abandoners, past purchasers, Customer Match) is the highest-ROI addition on most accounts.
• Post-click experience is part of the Shopping audit, not a separate conversation.