Google Ads Audit: Evidence, Priorities and Release Checks

A useful Google Ads audit establishes whether the account is optimizing for the right business outcome before recommending changes to bids or budgets. Start with goals and measurement, then review search intent, targeting, ads and landing pages. Finish with a short, evidence-backed change plan that names an owner and a retest for each finding.

Jafrin Akter AloLast updated 12 min read
On this page

Audit coverage

Keep the evidence period, campaign scope and business objective visible throughout the review.

Audit coverage
Audit areaEvidenceDecisionOwner
Goals and conversion actionsAction settings and campaign goal selectionsCorrect optimization outcomeMarketing and analytics
Event implementationControlled success and failure casesReliable trigger, identity and valueEngineering and analytics
Search intentSearch terms, keywords and exclusionsRelevant demand and defensible exclusionsPaid media
Targeting and deliveryLocations, audiences, schedules and statusIntended reach and eligibilityPaid media
Ads and landing pagesCurrent assets and completed customer journeyConsistent promise and usable destinationMarketing and web team
Budget and biddingSpend, mature outcomes and strategy settingsAffordable, measurable allocationMarketing and finance
Change controlHistory, proposed edits and acceptance criteriaReviewable releaseAccount owner

This Google Ads audit checklist is a workflow for producing findings, not a claim that every account has the same defects. Begin with read-only exports and settings records. Preserve the original evidence before making changes so that the team can separate the condition found from the condition after a repair.

When conversion reliability is uncertain, analytics and tracking support can connect the account goal, website event and business record. A campaign may have well-written ads while its bidding signal counts the wrong action, so measurement deserves an early decision gate.

A white C-shaped inspection clamp holds a lime card beneath a pink pad and cobalt handle.

Audit scope and evidence

  1. Record the account, currency, time zone, campaign types and review dates.
  2. Write the business outcome for each campaign group, such as accepted purchases or qualified sales opportunities.
  3. Export performance with the conversion columns and segments needed to test that outcome.
  4. Save the relevant goal, bidding, targeting and automation settings.
  5. Review change history around the period where performance changed.
  6. Record known business changes, including offers, inventory, pricing and website releases.

Choose a comparison period that suits the conversion delay and seasonality. A short period may reveal a broken landing page, but it may be too immature to judge acquisition economics. Keep operational defects separate from performance hypotheses that need more observations.

Google's change history can show account, campaign and ad-group edits from the past two years, including budget changes and the actor associated with changes. Use that timeline to identify plausible causes, while recognizing that timing alone does not prove causation. Google Ads change history.

The initial evidence pack should be small enough for another person to review. Include the export, relevant settings and a concise description of the concern. Avoid collecting unrelated customer information or turning the audit into an unrestricted download of every available table.

Conversion goals

Map each campaign to the goals it actually uses. Account-default goals affect campaigns using those defaults; campaigns with other goal combinations need separate inspection. Changing a shared action or goal can therefore affect more than the campaign currently on screen. Conversion goal management.

Check the action's business definition, source, value, counting setting and role in optimization. A page view, form start, accepted lead and closed sale can all describe different stages. Their names should make the distinction clear enough that a new account owner can understand the report.

Primary actions are used in the Conversions column and for bidding when their relevant standard goal is selected. Secondary actions are generally observational, but custom goals are an important exception: actions inside a selected custom goal can be used for bidding regardless of their primary or secondary label. Primary and secondary actions.

That exception belongs in the audit. Marking a diagnostic action secondary is not sufficient evidence that every campaign ignores it. Inspect custom goals and campaign selections before approving the optimization map.

For purchases, identify whether native website tracking, an Analytics import or another source represents the accepted outcome. More than one source can be useful for comparison, but do not assume two similarly named purchase actions should both influence the same goal. Record the intended measurement design before changing either one.

For leads, trace a sample from submission through the stage the business considers valuable. A form completion may be legitimate but commercially unqualified. Decide whether the audit is evaluating lead capture, lead quality or revenue, then select evidence that matches that question.

Conversion implementation

Run controlled cases that distinguish a successful business action from a nearby interaction. A purchase test should include completed payment, failed payment, confirmation-page reload and a second distinct order. A lead test should distinguish a click on Submit from an accepted submission and any subsequent duplicate attempt.

Google provides Tag Assistant for investigating unverified, inactive or attention-needed conversion actions. Use it to inspect the relevant implementation and retain the result with the event's business evidence. A tag firing is not proof that the business action occurred or that the platform attributed a sale. Tag Assistant diagnostics.

For native coded website actions, the conversion label guide explains the destination and action identifiers. The audit should inspect those identifiers together with the trigger, monetary fields and transaction identity rather than copying an identifier into a tag and declaring the action repaired.

Google documents duplicate handling for the same conversion action and transaction ID. Each distinct purchase needs its own ID, while a repeat delivery for that same purchase should retain it. A constant ID can undercount different purchases; a new random ID on every reload can undermine duplicate prevention. Transaction ID guidance.

Check consent behavior against the approved implementation. Basic and advanced Consent Mode have different loading and transmission behavior, so a declined choice is not evaluated through a universal assumption that no request should exist. Record the intended default, update and resulting tag behavior. Consent Mode overview.

Where enhanced conversions are enabled, inspect their diagnostics and the source of user-provided data. Google's current guidance describes a unified setting accepting data from website tags, Data Manager and API connections. Review the actual connected routes and relevant diagnostics rather than following an old checklist that assumes mutually exclusive setup methods. Enhanced conversions diagnostics.

For deeper measurement failures, use Google Ads conversion diagnostics and the GTM container audit skill. Keep account attribution, tag execution and accepted business records as separate evidence layers.

Search intent and exclusions

Review actual search terms alongside keywords, ad groups, landing pages and the intended customer. A term with spend and no recorded conversion deserves investigation, but that observation alone does not establish irrelevance. Check the available volume, conversion maturity and whether the measurement is trustworthy.

The search terms report is not a complete list of every query. Google omits some terms with insufficient activity for privacy reasons; search terms insights can aggregate otherwise hidden queries into themes. State that coverage limit when presenting a query audit. Search terms reporting.

Group the visible evidence into practical categories: clearly irrelevant intent, relevant intent with weak landing-page alignment, relevant intent with insufficient evidence, and strong candidates for further investment. Each category implies a different action. A long list of automatic exclusions can remove useful demand if these distinctions are skipped.

Negative keywords need match-type review. Google states that negative keywords do not match close variants or other expansions, so a single excluded form does not necessarily cover related variants. Also inspect where the exclusion is applied and whether it conflicts with valuable queries elsewhere. Negative keyword behavior.

Use the negative keyword workflow skill to prepare a reviewable candidate list. Preserve the query, proposed negative, match type, scope and reason. Have the campaign owner check the exclusions before application, then monitor whether intended searches remain eligible.

Targeting and customer journey

Inspect included and excluded locations together with the advanced location option. Google's default targeting can include physical presence and location interest; a campaign restricted to people in or regularly in an area uses a different option. The right choice depends on the offer and customer journey. Advanced location options.

A local service might need a different reach definition from a destination business serving people planning travel. Record the intended audience first, then compare settings and observed outcomes. Do not label an out-of-area interaction as waste until the business explains whether that person could be a valid customer.

Review networks, audience settings, exclusions, schedules and device performance within the campaign type being audited. Preserve intentional differences. A setting appropriate to one campaign should not be copied across the account merely to make the structure look uniform.

Follow the ad promise through the landing page and conversion path. Confirm the destination, offer, availability, currency, form behavior and success state. Include the relevant mobile layout and any redirects. A strong click-through rate is of limited value when a visitor reaches a mismatched offer or cannot complete the task.

The impressions and clicks guide helps separate visibility from engagement. Use the CTR calculator to check the arithmetic, then assess the quality of the resulting traffic with business outcomes. Treat a higher CTR as a diagnostic signal rather than an automatic business improvement.

Budget and bidding

Review budget decisions only after the measurement basis is accepted or its limitations are clearly recorded. Check the campaign objective, selected strategy, constraints, spend pattern and mature conversion value. Compare ratios with absolute contribution and the amount of useful demand available.

Search impression share can help explain eligibility and delivery constraints. Google defines it as received impressions divided by estimated eligible impressions; it is not a share of every search in the market. Changes to targeting or auction competitiveness can alter that estimated opportunity set. Impression share.

Use search impression share to distinguish budget and rank limitations before proposing a budget increase. More eligible impressions are only useful if the resulting customer acquisition remains acceptable. A delivery constraint does not establish the economic value of removing it.

For revenue-led campaigns, good ROAS targets connects the bidding discussion to margins and reporting definitions. A target that exceeds the account's realistic opportunity may reduce volume, while a lower target can increase spend without satisfying the business's required contribution. Document the intended tradeoff instead of recommending a universal setting.

Finding priorities

Prioritize confirmed measurement and customer-journey failures before speculative optimization ideas. Within each group, consider the breadth of the affected scope, business consequence, confidence in the diagnosis and effort needed for a safe repair.

Finding priorities
Illustrative findingEvidenceRiskOwnerRetest
Submit click treated as an accepted leadFailed-form case still fires the actionBidding signal includes failuresAnalytics and engineeringValid and invalid submissions
Repeated order with a new transaction IDReload produces a different IDDuplicate measurementEngineeringSame order twice, distinct order once
Irrelevant query familyQuery intent conflicts with the offerUnproductive trafficPaid mediaExclusion scope and eligible queries
Expired landing-page promotionAd and destination disagreeCustomer confusionMarketing and web teamFull ad-to-conversion journey
Low recent reported ROASIncomplete conversion maturityPremature budget decisionPaid media and analyticsSame cohort after agreed delay

These are illustrative audit cases, not findings from an inspected account. Preserve that distinction in an audit template until the reviewer has evidence for each row.

Affected-spend example

Suppose a hypothetical review covers four non-overlapping campaigns during the same month and currency. Campaign A spent $2,000 and has an untrusted lead trigger. Campaign D spent $500 and has untrusted purchase duplication. Campaigns B and C spent $1,500 and $1,000 respectively, with no identified measurement defect in this limited review.

The reviewed spend is $5,000. The spend associated with the two affected campaigns is $2,500, or 50%. This is a measure of exposure to an unreliable decision signal. It is not evidence that $2,500 was wasted, that every conversion is wrong, or that repairing tracking will recover that amount as savings.

If a campaign has both defects, count its spend once when calculating the combined exposure. Keep the finding-level rows for remediation, but use unique campaign identifiers for the aggregate. This avoids inflating the apparent scale of the problem by adding overlapping findings.

Repair and release workflow

Repair and release workflow
Business outcome and scope lead to goal and measurement checks. Failed decision signals require goal/tracking repair and repeat the outcome definition. Accepted signals permit intent, customer-journey and bidding review; rejected proposals repeat that downstream assessment before controlled release.
Read diagram as text
  • Business outcome and evidence scope to Goal and measurement checks.
  • Goal and measurement checks to Accepted decision signal.
  • Accepted decision signal to Goal and tracking repair: No.
  • Goal and tracking repair to Business outcome and evidence scope.
  • Accepted decision signal to Intent and customer-journey review: Yes.
  • Intent and customer-journey review to Budget and bidding assessment.
  • Budget and bidding assessment to Prioritized change plan.
  • Prioritized change plan to Accepted scope and retest.
  • Accepted scope and retest to Proposal revision: No.
  • Proposal revision to Intent and customer-journey review.
  • Accepted scope and retest to Controlled release and follow-up: Yes
Open full diagram

A measurement repair returns to the business outcome because changing the trigger or goal may change what the account optimizes. A proposal revision repeats the downstream assessment so that a new exclusion, landing-page plan or budget assumption is considered in context.

Write the proposed change as a before-and-after behavior, not merely a setting name. For a form repair, that might be: an invalid submission currently fires the lead action; after the change, only an accepted submission should do so. Include the failure case, the success case and the evidence required to accept the repair.

Keep the change record with its rollback method and review date. Separate implementation acceptance from later performance evaluation. The former asks whether the change behaves as specified; the latter asks whether it improves the outcome over a suitable period.

Keep reading

Browse all Tracking & Attribution guides

Questions people ask

Should a Google Ads audit start with negative keywords?

Usually it should start with business goals and measurement. Negative keywords are useful, but their priority depends on the quality of the outcome data and the relevance of the queries. Confirm what the account is optimizing for, then review exclusions alongside landing-page alignment, campaign scope and the maturity of the observed results.

Can a secondary conversion action still affect bidding?

Yes, a secondary action can be used for bidding when it belongs to a custom goal selected for the campaign. The label alone therefore does not settle the audit question. Inspect campaign goal selection and custom-goal contents together, and document the intended role of each action before recommending a change.

Does the search terms report contain every search that triggered an ad?

No, some low-activity terms are omitted for privacy reasons. The visible report remains useful for intent review, but its rows should not be described as the full query universe. Use available aggregated insights where relevant and state the coverage limit when estimating the scale of a query-related finding.

Is spend on a campaign with broken tracking automatically wasted spend?

No, broken tracking makes the decision signal unreliable, but it does not prove that every click lacked business value. Report the associated spend as affected exposure and investigate actual customer outcomes separately. Avoid presenting that exposure as recoverable savings or claiming that a measurement repair will produce an equivalent cash return.

Should all audit recommendations be applied in one release?

No, sequence changes according to their dependencies and acceptance criteria. Repairing a conversion definition may need to precede a bidding decision, while an unrelated landing-page error can have its own release. Keep the before-and-after evidence, rollback method and review date so the team can distinguish implementation correctness from later performance effects.

Sources

  1. Google Ads change history
  2. Conversion goal management
  3. Primary and secondary actions
  4. Tag Assistant diagnostics
  5. Transaction ID guidance
  6. Consent Mode overview
  7. Enhanced conversions diagnostics
  8. Search terms reporting
  9. Negative keyword behavior
  10. Advanced location options
  11. Impression share
Portrait of Jafrin Akter Alo

Written by

Jafrin Akter Alo

Data Engineer and Co-founder at Martechdevs and Growth and Analytics Partners

Jafrin writes practical guides to marketing technology, customer data, analytics and AI workflows. Martechdevs is a Growth and Analytics Partners company.

View LinkedIn profile

Book a Free Call

Thirty minutes on your stack and what it needs. Free, and no pitch.

Thirty minutes, and you leave with a plan

We look at what you have, what it is costing you and what to do about it first.

  • 30 minutes
  • Video call, link on booking
  • No obligation

What we'll cover

  • Your current stack and where it hurts
  • Quick wins we can ship first
  • An integration roadmap
  • Timeline, scope and next steps

Calendar not loading? Open it in a new tab or email hello@martechdevs.com.