Ecommerce Google Ads audits: what I check before recommending changes

An ecommerce Google Ads audit is useful when it turns a noisy account into a short list of decisions. I do not begin by collecting every setting or recommending a rebuild because the account looks untidy. I start with business goals, measurement, product data, and the relationship between campaign structure and demand.

Most audits fail in one of two directions. Either they are a settings inventory, a hundred observations with no ranking, which leaves the reader knowing more and deciding nothing. Or they are a sales document built to make the current setup look negligent so the next engagement looks necessary. A useful audit is neither. It answers what is happening, what it is costing, and what deserves attention first.

My broader guide to what ecommerce Google Ads management should include covers the operating foundation. This audit sequence focuses on the order of questions.

Start with the business question

The same account can require different work depending on the question. A brand trying to protect demand needs a different review from a brand trying to grow non-brand revenue, improve margin, recover product eligibility, or make reporting more trustworthy.

• What products, categories, or customer types deserve growth?
• Which outcomes count as primary conversions and why?
• What constraints matter now, such as stock, margin, promotions, or cash flow?

This step is skipped more often than any other, and skipping it is why so many audits recommend the same generic list regardless of the account. An account running at target on a constrained catalog does not need aggressive expansion advice. A brand carrying stock it needs to clear does not need a lecture on brand-term efficiency. Without the commercial context, a finding is just a deviation from a template.

Verify measurement before reading performance

I check that purchase events fire once, transaction values are sensible, revenue is assigned to the right order, and primary conversion actions reflect the business outcome. I also compare platform data with the store or analytics source closely enough to identify a tracking problem before it becomes a bidding problem.

This has to come first because everything downstream inherits it. If a purchase event double-fires, every ROAS figure in the account is inflated, the bidding algorithm has been optimizing toward a distorted signal for however long it has been broken, and any structural recommendation built on that data is built on sand. I have seen accounts described as high performing that were counting the order confirmation page on refresh.

• Does the purchase event fire exactly once per order, including on refresh and back navigation?
• Do transaction values match the store, and do they include or exclude shipping and tax consistently?
• Are secondary actions such as newsletter signups being counted as primary conversions?
• Is there a consent or tag-blocking issue suppressing a share of conversions?
• Does the attribution setting match how the business actually thinks about credit?

If the measurement layer is uncertain, use the ecommerce PPC tracking guide before making confident account changes.

Review products and campaign eligibility

Shopping and Performance Max depend on product inputs. I review active products, disapprovals, feed freshness, price and availability agreement, landing pages, country settings, and whether important products are actually eligible for the campaigns intended to sell them.

Look for commercial gaps, not only diagnostic counts

A small overall disapproval percentage can still hide a serious problem if the affected products are best sellers or high-margin categories. I compare account diagnostics with the commercial priority list so the audit reflects business impact.

The reverse also holds. A large disapproval count concentrated in discontinued lines or a variant explosion nobody intends to advertise is close to noise. Counting problems is easy and tells you almost nothing. Weighting them by the revenue they block is the part that takes account knowledge.

The feed health checklist covers the product-data checks in detail, and they are worth running before any structural recommendation.

Map campaign roles and query mix

Next I map brand, non-brand, category, product, competitor, Shopping, and automated campaign roles. I want to see where budgets are assigned, which queries each campaign captures, how branded demand is reported, and whether the account can answer basic questions without exporting every table.

• Is brand demand separated enough to read new-demand performance?
• Do Search campaigns match the landing pages and intent they target?
• Does Performance Max have a clear role and useful product grouping?
• Are budgets aligned with inventory, margin, and business priorities?

Branded demand is the single most common distortion I find. When brand traffic is absorbed into Performance Max or a broad non-brand campaign, the reported return looks strong because the account is being credited for demand the business already had. The account appears healthy and the non-brand engine underneath it can be quietly unprofitable for months.

That pattern is common enough to deserve its own treatment, which is why I wrote about separating branded demand from Performance Max.

Read the search terms against the catalog

Query review is where waste becomes visible, but the useful version is not a list of individual bad terms. I look for repeating patterns: themes the account keeps paying for that the catalog cannot serve, informational queries arriving on product pages, competitor names absorbing budget without converting, and terms whose intent does not match the products they trigger.

The output should be a small number of governance rules rather than a long negative list, because a rule keeps working as the catalog changes and a one-time negative list starts decaying the day it is applied.

End with a prioritized action plan

A strong audit ends with a sequence, not a long inventory of observations. I separate urgent measurement or eligibility work from structural improvements and controlled tests. Every recommendation should have an owner, a reason, a way to verify it, and a note about what could be affected.

Sequence matters as much as content. Measurement fixes come before structural work, because restructuring on untrustworthy data means you cannot tell whether the restructure helped. Product eligibility comes before bidding, because bidding cannot buy inventory that is not eligible to serve. An audit that lists ten correct recommendations in the wrong order can still waste a quarter.

What an honest audit is allowed to conclude

Sometimes the account is broadly fine. The structure is reasonable, measurement holds up, the feed is healthy, and performance is limited by price, margin, product, or market rather than by campaign management. That conclusion is legitimate and it is more useful than a manufactured problem list, though it is rarely what the person running the audit is incentivized to say.

Be cautious with any audit that opens with a large opportunity figure. If the number cannot be traced to arithmetic on data already in the account, it is a forecast wearing the costume of a finding, and it is usually the first thing that turns out to be wrong.

For ongoing visibility after an audit, Cardinal keeps reporting and reviewed next steps together. If you want this run on your own account, the paid ecommerce Google Ads audit covers all eighteen areas above and is credited against your first month if you go on to work with me.