A paid media audit should tell you what to trust, what to change, and what to stop doing. If it only produces screenshots and a longer task list, it has missed the point.
Advertising accounts accumulate decisions. A conversion action is added for a short-term campaign and quietly becomes primary. A brand campaign props up blended return. Performance Max absorbs more products. Budgets move, landing pages change, and the reporting still looks familiar enough that nobody stops to question the system underneath it.
A quarterly review creates that stop. The cadence is not magic: high-spend accounts may need lighter monthly checks, while a major site migration, CRM change, new market, or sudden performance break should trigger an immediate review. But once a quarter is a useful minimum for stepping outside daily optimization and asking whether the account still represents the business.
Prepare the evidence before opening the ad platform
Start with a defined period. Ninety days is usually long enough to reduce daily noise, but compare it with the prior period and the same season last year when seasonality matters. Note promotions, stockouts, price changes, major creative launches, site releases, and attribution changes. An unexplained spike often becomes obvious once the operating context is on the same page.
Gather these inputs before the audit:
- Your business source of truth for orders, revenue, qualified leads, or closed sales.
- Channel reports at campaign, ad group or asset group, ad, product, and search-term level where available.
- Change history for the review period.
- Current margin, allowable customer acquisition cost, payback, or qualified-pipeline targets.
- Analytics, tag manager, Merchant Center, CRM, and landing-page access relevant to the buying system.
Write the business question at the top of the audit. “Improve ROAS” is too vague. Better questions include: Can we acquire more new customers without exceeding a $75 CAC? Which campaigns create qualified showroom visits? Is the account under-spending because demand is limited, because bids are constrained, or because the structure misses eligible searches?
| Audit area | The question it must answer |
|---|---|
| Measurement | Can we reconcile the signals used for bidding with a real business outcome? |
| Economics | What can we afford to pay, and does the target change by customer, product, or lead quality? |
| Structure | Can each campaign’s role, budget, and bidding decision be explained? |
| Demand | Which searches and products are captured, excluded, or hidden inside blended reporting? |
| Creative and path | Is the message still earning response, and can a user complete the promised action? |
| Action plan | Who changes what, in which order, and how will the team know it worked? |
1. Verify conversions before evaluating campaigns
Campaign analysis is downstream of measurement. If the conversion set is wrong, a precise-looking CPA is still wrong. List every conversion action and mark whether it is primary, secondary, included in bidding, assigned a value, and still used by the business.
Then test the important paths yourself. Place a test order where possible. Submit a lead. Move a test record through the relevant CRM stages. Confirm that one real action does not produce duplicate browser, server, analytics, and imported conversions. Check that currency, tax, shipping, refunds, and subscription events are handled consistently with the number finance uses.
Do not expect the ad platform and the order system to match exactly. Attribution windows, reporting time zones, modeled conversions, consent, and return timing create legitimate differences. The goal is to explain the gap and decide which number governs which decision. An unexplained difference is the problem.
For a longer sales cycle, a raw lead is rarely enough. Review whether qualified stages or closed outcomes return to the platform as offline conversions. If a showroom visit is worth more than an unqualified inquiry, the bidding system needs a way to distinguish them.
Measurement checklist
- Remove obsolete or accidental primary goals from bidding.
- Confirm purchase or lead events fire once and carry the intended value.
- Compare platform conversions with the business source of truth by day or week.
- Document the attribution window and lag before comparing recent periods.
- Check enhanced conversions or offline imports where they are part of the setup.
- Record every unexplained discrepancy as a finding, not a footnote.
2. Rebuild the target from unit economics
Before labelling a campaign efficient or inefficient, define the economic threshold. Revenue is not contribution. Start with product margin, discounts, shipping subsidy, payment fees, expected returns, fulfillment costs, and any repeat-purchase value the business can support with evidence.
Separate new and returning customers where the data allows. A campaign can show an attractive blended ROAS because existing customers are searching for the brand. That may still be useful revenue, but it does not answer what the company pays for incremental acquisition. For lead generation, replace the generic lead target with the expected value of a qualified stage or closed sale.
If the team has not built this model, use the paid media budgeting guide to calculate an allowable CAC and a spend range the business can fund. Write the approved threshold into the audit. Otherwise the account team will optimize toward whichever platform number happens to look strongest.
3. Separate brand demand from acquisition
Pull search terms and campaign results for brand, non-brand category, competitor, and shopping activity. The purpose is not to dismiss brand. It is to stop one type of demand from concealing another.
Check whether brand traffic sits inside Performance Max or broad non-brand campaigns, whether exclusions are intentional, and whether branded impression share aligns with the account’s stated job. Review partner traffic and queries with ambiguous intent. A blended total should be the last layer of reporting, not the only one.
One athletic-apparel Google Ads audit illustrates the issue. The account was meeting headline ROAS benchmarks, yet branded campaigns supported a blended $33 CPA and branded impression share was 72%. Its strongest PMAX campaign had only 32% impression share while another sat at a $185 CPA. Those were audit findings that informed a proposed rebuild; the public record does not include implementation or post-change results.
A blended metric is a summary. It is not permission to stop asking what produced it.
4. Make every campaign justify its structure
For every campaign, write one sentence describing its role. Who or what does it target? Which conversion does it optimize? Why does it have a separate budget? What decision becomes possible because it exists?
Look for campaigns that split too little data across too many budgets, duplicate the same demand, or combine products with incompatible economics. Also look for the opposite problem: one campaign doing so many jobs that nobody can tell which product, audience, or query caused the result.
Review bidding settings against available volume and the business target. Inspect change history around performance breaks. Check shared budgets, location settings, networks, exclusions, schedules, and automatically applied recommendations. Platform recommendations can surface useful ideas, but each change should earn its place in the account rather than arrive by default.
Structure checklist
- Give every campaign a documented purpose and primary outcome.
- Identify overlap in queries, products, audiences, locations, and conversion goals.
- Flag budgets that prevent a proven campaign from entering eligible auctions.
- Flag low-volume fragmentation that prevents a strategy from learning.
- Review automated settings and recent changes with an owner and rationale.
- Prefer the simplest structure that still exposes the decisions the team must make.
5. Audit Shopping, Performance Max, and the feed together
Shopping performance begins before the campaign. Review Merchant Center diagnostics, disapprovals, pricing and availability mismatches, identifiers, product types, titles, images, and custom labels. A bidding strategy cannot recover demand for products that are ineligible, poorly described, or grouped without regard to margin and inventory.
Inside Performance Max, inspect product coverage and exclusions, asset-group organization, creative inputs, search-term and category insights, audience signals, and the campaign’s relationship with Search and standard Shopping. Ask what PMAX is responsible for and which parts of that responsibility remain observable.
Do not add a new campaign merely because a format is absent. Document the demand or control the new structure is meant to provide. The audit should connect each proposed Search, Dynamic Search, Shopping, or PMAX change to a specific finding.
6. Diagnose creative decline as a pattern
For paid social and asset-led campaigns, compare spend, reach, frequency, click-through rate, conversion rate, and acquisition cost over consistent windows. Creative fatigue rarely appears in one metric on one day. Look for response weakening across several related signals while delivery conditions remain reasonably stable.
Group ads by concept, not only by file name. A new crop or headline may still be the same idea to the audience. Record which hooks, proof points, offers, formats, and visual treatments have actually been tested. If response returns when a meaningful variation launches into the same campaign conditions, the diagnosis becomes stronger.
For a fuller diagnostic sequence, use the guide to spot creative fatigue before performance collapses.
7. Follow the click through the conversion path
Open the live ads and behave like a customer. Test desktop and mobile. Confirm the destination matches the message and the intended product or offer is available. Check page speed, navigation, forms, checkout, promo codes, confirmation pages, tracking parameters, and the handoff into email or CRM systems.
For lead generation, follow a record beyond the form. Does it reach the correct team? Is the source retained? Can sales mark qualification and closure in a way that reporting and bidding can use? A campaign can produce inexpensive leads while the operating system loses the information required to judge them.
Record conversion-path failures separately from media findings. A weak landing page should not automatically trigger a new bid strategy, and a tracking failure should not be described as a demand problem.
8. Judge budget by marginal performance
Average ROAS or CPA describes the spend already made. The budget decision concerns the next dollar. Compare periods in which spend rose or fell and observe what happened to new-customer volume, contribution, qualified pipeline, impression share, and efficiency at the margin.
Check whether campaigns are budget-limited for useful reasons, whether inventory and service capacity can absorb more demand, and whether cash flow can tolerate the payback period. A campaign with room to enter more auctions is not automatically entitled to more budget. The account needs evidence that the additional demand is valuable and operationally supportable.
9. Turn findings into a 90-day action register
Finish with a short, ordered register—not a bag of optimizations. Each item should include the evidence, business risk, proposed action, owner, dependency, verification metric, and review date.
| Priority | Use it for |
|---|---|
| Critical | Broken or misleading measurement, policy risk, uncontrolled spend, or a conversion path that cannot complete. |
| High | Structural problems that materially distort acquisition economics or block proven demand. |
| Medium | Tests with a clear hypothesis and enough expected volume to produce a readable result. |
| Monitor | Conditions worth watching that do not yet justify an account change. |
Sequence measurement repairs before performance judgments that depend on them. Avoid launching several structural changes at once when that would make the effect impossible to read. Capture a baseline, annotate the change, and decide in advance how long the system needs before review.
The one-page quarterly checklist
- Define the business question and review window.
- Reconcile primary conversions with the business source of truth.
- Set the CAC, contribution, payback, or qualified-pipeline threshold.
- Separate brand, non-brand, returning-customer, and prospecting economics.
- Make each campaign justify its role, budget, bid strategy, and conversion goal.
- Review queries, products, feed health, Shopping, and PMAX coverage together.
- Diagnose creative with a pattern of delivery and response signals.
- Test the live path from ad click through order, qualification, or closed sale.
- Evaluate the next dollar with marginal performance and operating capacity.
- Assign an owner, metric, and review date to every approved action.
Know when the internal audit needs a second set of eyes
A capable team can run this checklist. Outside review becomes useful when the team cannot reconcile conversion data, when brand and non-brand demand remain inseparable, when PMAX or Shopping obscures product-level decisions, when a major budget increase is being considered, or when the same people who built the structure need an independent challenge to its assumptions.
Channelshift’s Google Ads audit service reviews the evidence and produces a prioritized rebuild plan. The linked athletic-apparel audit case shows the level of structural detail such a review can uncover—and clearly separates proposed improvements from measured results. You can also see the broader set of paid media services if the constraint extends beyond the account itself.
The value of the quarterly audit is not that it creates more work. It is that it forces the account, the data, and the business objective back into the same conversation. At the end, the team should know what is true, what remains uncertain, and which decision comes next.