A small-business owner should be able to leave an advertising review knowing what is broken, what is uncertain and what will happen next. A long list of generic recommendations does not provide that clarity.
This checklist organizes the review around evidence and decisions. Use the Google Ads audit template or Meta Ads audit template as an editable working copy. Both are free CSV downloads with prepared checks, evidence fields and space for owners.
Define the business outcome first
Write what the business needs advertising to produce. For ecommerce, that may be new customers with enough retained contribution to support acquisition costs. For a service business, it may be qualified inquiries in an area the team can actually serve.
Avoid beginning with a universal “good ROAS” or “good cost per lead.” The acceptable result depends on margins, close rates, capacity and the definition of the measured outcome. Use the allowable CAC calculator when first-order economics are part of the decision.
Record the account, currency, time zone, included campaigns and reporting dates. Decide how long outcomes need to mature. A lead created yesterday may not have a meaningful sales outcome yet, and a recent purchase report may still change as refunds or delayed events arrive.
Audit the outcome path before the dashboard
Trace the path from the ad to the destination and then to the business system. Ask whether the landing page works, the visitor can complete the intended action, and the business receives that action once.
A browser tag firing is useful technical evidence. It is not the same as an accepted inquiry or completed order. Likewise, a form button click can occur even when validation fails or the server rejects the request.
For an authorized test, record the destination, expected event, accepted business-system outcome and any matching or duplicate-handling rule. Keep personal details out of a widely shared checklist. An internal evidence reference is usually enough for the review document.
If the measurement path is broken, prioritize that repair before treating reported efficiency as a reliable budget signal.
Inspect the goals the campaign actually uses
On Google Ads, review the campaign's selected goals and the actions included in them. Google distinguishes primary and secondary conversion actions; custom goals are an important exception because included secondary actions can still be used for bidding. Read the actual campaign configuration instead of assuming the account-level label tells the full story. Google's conversion-action documentation.
For other platforms, inspect the selected objective, outcome event and event source in the current interface. Write down the intended customer action and compare it with what the campaign is configured to pursue.
An audit finding might be “the campaign optimizes toward a shallow action that does not match our current objective.” That is a concrete configuration question. It is stronger than saying “the algorithm is bad” without identifying the inputs it was given.
Check demand and eligibility
Review who can see or respond to the advertising and whether the offer serves those people. For search campaigns, inspect representative search-term groups by intent, not only individual words. For a local service, compare location assumptions with the areas the business can support.
Check exclusions as carefully as inclusions. An old negative list can block useful demand; a missing exclusion can admit irrelevant demand. Before editing, record the evidence and consider what else shares the setting.
Keep brand demand visible in the analysis. A campaign that receives credit for people already looking for the business answers a different question from one reaching prospective new customers. Neither result alone proves incremental impact.
Review creative and the destination as one promise
Open the ads and the actual pages they point to. Check whether the price, offer, eligibility, product and next step agree. A polished ad can still lead to a confusing page, a stale promotion or a form that is difficult to use on a phone.
Group creative by concept and offer before comparing results. A set of near-identical variants can make the account look busy without testing a materially different reason to buy.
Also inspect delivery. An asset with very little spend has limited evidence behind its observed rate. Do not call it a proven loser just because another asset received most of the budget. Use the creative testing template to turn a proposed improvement into a question the account can answer.
Reconcile spend with the approved allowance
Compare current spend with the budget for the same period. Note planned promotions, pauses and changes in operating capacity. Use the budget pacing calculator to make the arithmetic explicit.
Then review recent strategy or target changes alongside outcome maturity. Several edits made together can make a performance swing difficult to interpret. The audit should identify the next decision, not encourage changes merely to demonstrate activity.
For an ecommerce review, distinguish gross attributed revenue from retained revenue and contribution. For a service business, distinguish raw inquiries from qualified opportunities and completed sales. The paid-media report template keeps those definitions beside the numbers.
Bring sales evidence into a lead-generation audit
Ask how many inquiries were accepted, reviewed, qualified and converted. Keep invalid submissions, legitimate poor-fit inquiries, unreachable prospects and unreviewed leads separate.
A campaign with a large review backlog may appear to have worse quality simply because its outcomes are incomplete. Show review coverage before comparing qualification rates. The lead quality scorecard provides a stage dictionary and cohort fields for this review.
If the business uses offline outcomes in its advertising measurement, inspect delivery and definitions as a separate technical dependency. Do not assume that a successful import proves the qualification process is consistent or that every lead has had time to reach the relevant stage.
Use an evidence-based triage table
| Finding type | Example | Useful next step | Completion evidence |
|---|---|---|---|
| Confirmed broken path | The inquiry form returns an error | Repair and verify the intended submission flow | Accepted controlled test and correct error handling |
| Measurement mismatch | A report counts clicks as well as inquiries | Reconcile definitions and inspect goal selection | Documented configuration and event checks |
| Commercial mismatch | Leads request a service outside the offer | Review promise, eligibility and campaign scope | Agreed change plus a mature follow-up cohort |
| Performance hypothesis | A clearer demonstration may improve response | Plan a bounded creative test | Design, result and uncertainty |
| Missing evidence | Sales has not reviewed the recent cohort | Complete the review process | Comparable mature counts |
This is an editorial triage framework, not a provider scoring system. Assign urgency based on the business consequence and confidence in the finding. A confirmed broken page can justify a direct repair; an uncertain creative theory needs a different process.
Work through one illustrative finding
Suppose a local service spends $3,000 and receives 100 accepted inquiries. The dashboard's cost per inquiry is $30. Sales has reviewed 60 inquiries and qualified 18, while 40 remain pending.
The observed qualified share among reviewed inquiries is 30%. It would be misleading to treat all 82 inquiries not yet qualified as rejected. It would also be premature to call the campaign successful based on the $30 inquiry cost alone.
The next actions are to complete a comparable review window, inspect the reasons for poor fit, and calculate economics using the intended qualified or converted outcome. If the form is also broken on one mobile path, repair that technical issue independently. Do not force every finding into one explanation.
End with a short action register
Choose a small set of actions the team can actually complete. For each, record the finding, evidence reference, owner, decision authority, implementation step and verification date.
Keep “implemented” and “improved performance” as different statuses. A configuration readback proves the setting changed. A later report may show outcomes changed, but other factors and the design determine what causal claim is justified.
Download the appropriate audit checklist, save the working scope and start with the outcome path. The value of an audit is the quality of the next decisions it supports, not how many boxes it turns green.