Short version: A Shopping audit is not a Search audit with products attached. The feed is the campaign, so most of the findings live in Merchant Center rather than Google Ads. Work through feed health, product coverage, structure, bidding and search terms in that order, because a problem upstream makes everything downstream unreadable.
Shopping accounts fail differently from Search accounts. A Search account usually degrades slowly through accumulated waste. A Shopping account tends to look fine in the aggregate while a specific subset of the catalogue quietly does all the work and the rest burns budget. The aggregate ROAS hides it, and nobody notices until growth stalls.
This checklist works for standard Shopping and, with the noted differences, for the Shopping side of Performance Max. It assumes read access to both Google Ads and Merchant Center. Without Merchant Center you are auditing half the account.
1. Feed health: what is actually eligible to serve?
Open Merchant Center and look at the product count that is approved and active, then compare it to the number of products you sell. The gap is the first finding. Disapprovals, pending review items and products excluded by a feed rule all sit in that gap, and every one of them is a product that cannot appear no matter how much you bid.
Common causes in order of frequency: missing or invalid GTINs, price mismatches between the feed and the landing page, missing availability data, image quality failures, and policy disapprovals on restricted categories. Price mismatch is the most damaging because it can disapprove products intermittently as the site updates, which makes it hard to spot in a single snapshot.
Check the diagnostics tab for warnings as well as errors. Warnings do not block serving but they suppress it. A product missing brand, colour, or size attributes competes worse than an identical product that has them.
2. Product coverage: how much of the catalogue actually sells?
Pull a report of impressions by product for the last 90 days. In most unaudited accounts a small fraction of the catalogue takes the overwhelming majority of impressions. Some of that is correct, since some products genuinely have more demand. Much of it is structural: those products won early, accumulated performance data, and now get preferentially served while the rest never gets the impressions it needs to prove itself.
The finding to write down is the number of products with zero impressions in 90 days. Those products are in the feed, approved, and invisible. Whether that matters depends on their margin and stock position, which is a conversation with the business rather than something the account data answers.
3. Product titles: the field that decides what you match
Shopping has no keywords. The title is the primary matching signal, and most titles are whatever the store platform exported, which is usually the product name and nothing else.
A title that reads "Classic Tee" matches almost nothing. The same product titled "Organic Cotton Crew Neck T Shirt, Navy, Mens Medium" matches the way people actually search. The general shape that works is brand, then product type, then the attributes a buyer would specify: colour, size, material, model number where relevant.
Titles truncate visually around 70 characters in most placements but the full 150 is used for matching, so write to the full length and put the most important words first. Our guide to product feed optimisation covers the attribute work in more detail.
4. Campaign structure: can you control anything?
A single Shopping campaign containing the entire catalogue gives you exactly one bid lever for every product you sell. That is fine for a very small store and unmanageable for anything else, because a high margin product and a loss leader get treated identically.
The structural question is whether the account separates products by something that matters commercially. Margin band is usually the most useful split, implemented through custom labels in the feed. Product type, brand and price band all work too. What matters is that the split lets you set different targets for products with genuinely different economics.
Also check whether brand searches are separated from non brand. In Shopping this is done with negative keywords and campaign priority rather than keyword targeting, and when it is not done, brand conversions inflate the apparent performance of cold traffic.
5. Bidding: is the target based on anything real?
Find the target ROAS on each campaign and ask where the number came from. If nobody can answer, it was chosen because it sounded reasonable, which is the norm.
The number should derive from contribution margin, not gross margin, and it should differ across product groups with different margins. A single account wide target applied to products ranging from 20 to 70 percent margin guarantees that some of that spend is unprofitable and some profitable demand goes unbought. The arithmetic is in our break-even ROAS guide.
Check also whether campaigns are budget constrained. A campaign hitting its daily budget every day is being throttled, and its performance data is a sample of the cheapest available traffic rather than a representative read.
6. Search terms: what are you actually paying for?
Shopping search terms are where the waste is most visible. Because you cannot select positive keywords, the only steering available is negatives, and on most accounts nobody has added any in months.
Filter to clicks greater than zero with no conversions, sort by cost, and read down. Look particularly for research intent, competitor names, wrong product categories, and the accessory trap, where a store selling machines gets traffic for spare parts it does not stock. The method is in how to add negative keywords.
7. Performance Max: what changes
If Shopping runs through Performance Max, several checks change. Asset group structure replaces product group structure as the main organisational lever, and listing groups within asset groups control which products serve where. The same margin based logic applies, implemented differently.
Check whether the account uses a brand exclusion list, since without one PMax will absorb brand traffic and report it as its own success. Check whether search themes and audience signals have been set, since an asset group with no signals starts colder than it needs to. And check the asset quality ratings, because a PMax asset group short on assets is capped regardless of budget.
The comparison between campaign types is covered in PMax versus Shopping versus Demand Gen.
8. Tracking: do the numbers survive checking?
Compare Google Ads conversions and conversion value against actual orders and revenue in the store backend for the same window. They will not match exactly, since attribution windows differ, but they should be in the same territory. A gap of two or three times means double counting somewhere, and every optimisation decision made on those numbers has been wrong.
Check specifically whether conversion value includes tax and shipping, whether refunds are excluded, and whether the value passed is revenue or profit. Each of those quietly shifts what a ROAS target means.
9. Landing pages and stock
The last check is the one furthest from the account and often the most valuable. Click through to the landing pages for your highest spending products. Confirm the price matches the feed, the product is in stock, and the page loads quickly on a phone.
Out of stock products still serving is a surprisingly common finding, particularly where the feed updates on a slower schedule than the store. You are paying for clicks on something nobody can buy.
Seasonality and stock: reading the account in context
Shopping performance moves with the calendar in ways that make month on month comparison misleading. A November figure compared against October tells you almost nothing useful in retail, and a competent audit compares against the same period last year wherever enough history exists.
Stock position matters just as much and is invisible in the ads interface. A campaign whose performance collapsed may have simply run out of its best selling products. Before diagnosing a bidding or structural problem, check whether the products that drove the previous period's revenue are still available. This single check has explained more sudden Shopping declines than any setting in the account.
Promotional history is the third piece of context. If the comparison period included a sale, the conversion rate and average order value from that period are not a fair baseline, and targets set against them will strangle the account once the promotion ends.
Merchant Center settings worth checking
A few Merchant Center configurations cause problems disproportionate to how obscure they are.
Shipping settings that do not reflect reality will suppress products or misrepresent the total price in the listing. If shipping is configured as a flat rate in Merchant Center but calculated at checkout, the mismatch can trigger disapprovals and it certainly damages click through, because the price shown is wrong.
Automatic item updates, which let Google correct price and availability from your structured data, are worth enabling on most stores. They prevent the intermittent disapprovals that come from a feed refreshing more slowly than the site changes.
Check the feed refresh schedule against how often prices move. A daily feed on a store that reprices hourly will spend part of every day out of sync, and price mismatch is one of the fastest routes to disapproval at scale.
Turning the findings into a plan
Order the findings by cost of inaction rather than by ease. Broken tracking comes first, because it invalidates everything else. Then feed disapprovals, since they represent products that cannot sell at all. Then wasted spend, then structure, then bidding refinement.
Resist the urge to change everything in one week. Shopping campaigns and Performance Max both relearn after significant changes, and a simultaneous overhaul of structure, bids and feed makes it impossible to attribute what worked.
Frequently asked questions
How often should I audit a Shopping account?
A full audit twice a year is reasonable for a stable account. Feed diagnostics and search terms should be checked monthly, since both change continuously as the catalogue and the auction move.
Can I audit Shopping without Merchant Center access?
Partially, but you will miss most of it. Disapprovals, feed attribute quality and product level diagnostics all live in Merchant Center. An audit without it can assess structure and bidding but cannot explain why products are not serving.
What is the single most common Shopping problem?
Product titles left as the platform exported them. It costs nothing to fix, affects every impression the account is eligible for, and is skipped in the large majority of accounts we look at.
Does a bigger catalogue need a different approach?
Yes. Beyond a few thousand products, manual product group management stops being viable and the work shifts to feed rules and custom labels that segment automatically. The audit questions stay the same, the implementation changes.
Should Shopping and Performance Max run at the same time?
They can, though PMax generally takes priority for the same products, so running both on an identical catalogue usually means Shopping serves very little. Where both run, it is normally with a deliberate product split between them.
Where to start
Start with tracking and feed diagnostics, in that order. Between them they explain most stalled Shopping accounts, and neither requires changing anything to investigate.
If you want an independent read on a Shopping account, request a free audit and we will run this checklist against it and send you what we find.
About the author
This guide is written by Vasant Chaudhary, a Google Ads specialist with more than five years of experience managing over 50 e-commerce accounts across the US, UK, and India. He focuses on Google Shopping, Performance Max, and product feed management. Get in touch or start with a free audit.