A paid media audit example should not start with clicks, impressions or a platform-generated optimisation score. It should start with a harder question: after product cost, fulfilment, returns and advertising, is each channel creating profitable revenue?
For an established eCommerce brand spending £10,000 or £100,000 a month, small structural mistakes compound quickly. A Shopping campaign can look efficient while it funds low-margin bestsellers. Meta can report strong return on ad spend while its prospecting activity is largely taking credit for sales that would have happened anyway. An audit exists to separate reported performance from commercial performance, then show exactly where budget should move.
The scenario: a £40,000 monthly paid media account
Consider a direct-to-consumer retailer with a £40,000 monthly paid media budget. It sells home products online, has a blended gross margin of 58%, and an average order value of £72. After payment fees, pick-and-pack costs, delivery subsidies and an expected return rate, the business has calculated that it can afford a maximum cost of sale of 28% on new-customer acquisition. That is the number that matters.
Its account is split across Google Shopping, Performance Max, Search and Meta Ads. The previous agency reports £180,000 in tracked revenue at a 4.5x ROAS. On the surface, that is a success. The founder is less convinced: cash is tight, discounts are increasing, and profitable growth has stalled.
A proper audit does not immediately rebuild campaigns. First, it tests the numbers against the business model, the tracking setup and the way users actually buy.
Step one: establish the real performance benchmark
The account’s reported ROAS of 4.5x equates to a 22.2% advertising cost of sale. That sits beneath the 28% target, but the blended view hides material differences:
- Google Shopping and Performance Max report a 6.1x ROAS.
- Brand Search reports 12x ROAS.
- Non-brand Search reports 2.4x ROAS.
- Meta prospecting reports 2.1x ROAS.
- Meta retargeting reports 7.8x ROAS.
The first finding is not that Meta should be switched off or non-brand Search should be cut. It is that one blended target is too blunt to manage a mixed funnel. Brand Search and retargeting are harvesting existing demand. They have a role, but they cannot be used as proof that acquisition is working.
The audit should also compare platform revenue with the retailer’s analytics platform, total orders, new versus returning customer split and post-purchase data. Attribution will never line up perfectly. The issue is whether discrepancies are understood and consistent. If Meta claims £35,000 in revenue from a £10,000 spend while the store’s overall revenue barely moves, the account is not ready for budget expansion.
Paid media audit example: where the spend is leaking
In this example, four leaks are responsible for most of the lost profit. None is unusual. Together, they explain why the reported 4.5x ROAS is not translating into the commercial result the business expects.
1. Performance Max is spending against weak products
Performance Max generates £68,000 in attributed revenue from £11,000 spend. But a product-level review reveals that 44% of its cost is assigned to products with below-average margin, high return rates or frequent stock issues.
One range of oversized products is particularly damaging. It has a healthy conversion rate, but delivery costs and returns make its true break-even ROAS 5.8x. The campaign delivers 4.2x. Platform reporting calls this acceptable because it sees revenue, not contribution margin.
The recommendation is not to exclude every lower-margin product automatically. Some products introduce customers to the brand and lead to repeat purchases. The immediate action is to segment the feed by commercial value: high-margin, proven products; products that need controlled testing; and products that cannot support paid acquisition at the current price. Campaign structure and budget should follow that segmentation.
2. Brand demand is masking acquisition performance
Brand Search has an exceptional ROAS, as expected. Yet its impression share is already high, and the campaign is using broad matching with weak exclusions. It is serving on competitor terms, product research queries and searches that would be better handled elsewhere.
More importantly, the previous reporting includes brand revenue in the acquisition success story. That makes the prospecting channels appear more scalable than they are. The audit separates brand, non-brand, Shopping and retention audiences so the business can see the cost of creating demand rather than simply collecting it.
A sensible brand campaign protects the terms that matter, controls spend and prevents competitors from taking easy demand. It should not become a convenient place to hide an underperforming acquisition strategy.
3. Meta is optimised for the wrong conversion signal
Meta prospecting is running broad audiences, lookalikes and interest sets, but most ad sets are optimised towards Purchase with a seven-day click, one-day view attribution model. That can work, but only if server-side tracking, purchase values and event quality are reliable.
Here, the implementation is incomplete. Purchase events are duplicated for a portion of users, and the product catalogue includes discontinued variants. The reported return is inflated, while dynamic product ads occasionally direct users to unavailable options. This is not a creative problem. It is a data and feed problem that creative testing cannot solve.
The audit recommends fixing event deduplication, validating value and currency parameters, and removing unavailable products before making major budget calls. Then it restructures campaigns around a clear purpose: prospecting to create new demand, retargeting to close high-intent users, and existing-customer activity only where repeat purchase economics support it.
4. Search terms are wasting budget quietly
Non-brand Search looks poor at 2.4x ROAS, but the answer is not necessarily to abandon it. Search can capture valuable, high-intent customers before they know the retailer by name. The search-term report exposes the actual issue: nearly 30% of spend is going to informational, bargain-led and irrelevant queries.
Queries containing “ideas”, “free”, “used”, “wholesale” and competitor names have been allowed to accrue cost without a deliberate strategy. Match types are too open, negatives have not been maintained, and expensive generic terms are treated the same as specific product searches.
The fix is disciplined query management. Exclude irrelevant intent, isolate proven query themes, set budgets based on margin rather than volume, and keep exploratory search activity separate from dependable revenue drivers. Search needs room to learn, but it does not deserve unlimited tolerance.
What changes after the audit
The first 30 days should prioritise measurement and waste reduction, not dramatic scaling. In this example, the retailer pauses unprofitable product groups, rebuilds feed labels around margin and stock status, tightens Search queries, and repairs Meta event tracking. Brand reporting is separated from prospecting reporting.
Budget is then reallocated gradually. High-margin Shopping inventory receives more investment, while Performance Max is prevented from absorbing spend into weak ranges. Meta prospecting is judged on new-customer efficiency and blended revenue movement, not its most flattering attribution view. Non-brand Search retains a controlled test budget because some terms have strong contribution potential once waste is removed.
After six weeks, reported ROAS may even fall slightly as brand demand is stripped out of headline reporting. That is not failure. The more useful measures are contribution after advertising, new-customer revenue, stock-adjusted profitability and the relationship between paid spend and total revenue. Honest measurement can look less impressive before it becomes more profitable.
What a serious audit should give you
A useful audit is not a document full of screenshots and generic best-practice comments. It should identify the commercial constraint, quantify the biggest opportunities, explain the risk behind each recommendation and set priorities in the right order.
For a mature eCommerce business, that means reviewing product margins, feeds, tracking, search queries, audience overlap, campaign structure, attribution and budget allocation together. Looking at any one in isolation produces partial answers. A technically tidy account can still lose money if it pushes the wrong products. Equally, a strong product range cannot compensate for broken measurement or unchecked query waste.
The goal is not a prettier account. It is a paid media operation that can spend more only when the economics justify it. If an audit cannot show where profit is leaking, what must change first and how success will be measured, it has not gone far enough.