Before you start
- Read-only or standard access to the account, and to the manager account if there is one, with change history visible. Admin access is not needed to audit and is safer not to have.
- The business context: what the account is for, the agreed KPIA metric tied directly to an objective. set and targets, the margin or lead valueWhat a lead is worth at a pipeline stage: what a closed deal contributes times the share of leads there that close; a planning figure, never revenue. behind them, seasonality, and anything the owner already knows is broken. The account structure guide and the measurement hub are the prerequisites because the audit judges against them.
- A period: a stable recent window long enough to include a full business cycle, and a comparison period a year earlier if the account is old enough. Avoid periods with migrations or outages.
- Access to the site's tag container and analytics, so measurement can be checked at source rather than inferred from reports.
- A findings sheet with columns for area, finding, evidence, affected spend, impact, effort and priority. The scoring table at the end fills it.
Set scope and goals
Write down what the audit is for before opening a report. An audit to decide whether to keep an agency asks different questions from one to prepare for a new market or to find why cost per acquisition (CPACost of each conversion.) rose. Confirm the campaign types in scope, the period, the targets the account is judged against and who will receive the findings. Agree what a finding looks like: a specific problem with evidence, affected spend and a recommended action. Without this step the audit becomes a list of preferences, and the owner cannot tell which items matter. Where affiliates also advertise on the brand's terms, the tutorial on auditing an affiliate programme covers the partner side of the same auctions.
Verify measurement integrity
Nothing after this step is trustworthy until it passes. List every conversion action with its source, category, counting, value, window, attribution model and primary or secondary flag. Check which actions are primary and whether they represent the business outcome. Confirm the tag fires correctly on the site and that the actions show a recording status, following the checks in the conversion tracking tutorial. Look for duplicates between native tags and imports, for "every" counting on leads, for auto-tagging switched off and for conversion goals assigned per campaign. Compare conversions with back-office outcomes for the period. A measurement failure is always the first item in the backlog, and the rest of the audit is written with the caveat that figures may be wrong until it is fixed.
Review account settings and structure
At account level check currency, time zone, auto-tagging, linked accounts, access and billing, shared negative lists and shared budgets. Then map the structure: campaigns by type, what each exists to control, and whether the split follows a control need or a habit. Count campaigns with too few conversions for their bid strategy to learn, campaigns limited by budget, ad groups with a single keyword or with none, and duplicate keywords across campaigns competing with each other. Judge the structure against the principles in the account structure guide rather than against a preferred template. Record structural findings as proposals, since restructuring is a project rather than a fix.
Check campaign settings
Open each campaign's settings and read them as a list, because defaults cause more waste than deliberate choices. Networks: Search campaigns opted into the Display Network or search partners without intent. Locations: the location option set to presence or interest where presence was meant, and targets that include places the business cannot serve. Languages, ad schedules, start and end dates, ad rotation, campaign-level goals, and any URL options or tracking templates. Note every setting that differs from what the campaign's purpose implies, with the spend affected. This step is fast and usually produces several cheap wins.
Examine keywords and match types
For Search campaigns, review keyword coverage against the business: products or services with no keywords, keywords for things the business does not sell, and head terms absent while long tail is present. Review match types by spend share and by performance, and whether broad match is paired with Smart Bidding and a negative strategy or left to run on manual bids. Read Quality Score (Quality ScoreGoogle's 1–10 keyword diagnostic.) and its components by keyword where spend concentrates, as a pointer to relevance and landing page problems rather than a target. Flag keywords with status issues: low search volume, rarely shown, below first-page bid under manual bidding, and conflicts with negatives. The guide to Google Ads status messages reads these alongside the campaign, ad, bid strategy and conversion statuses and names the page that owns each fix.
Read search terms and negatives
The search terms report shows what the account actually bought. Sort by spend without conversions, by conversions with high cost, and look for themes: competitors, job seekers, informational queries, locations outside the service area, products not sold. Check which negatives exist at account, list, campaign and ad group level, whether shared lists are applied to the campaigns they should be, and whether any negative blocks a converting term. The negative keywords guide covers the method; the audit's job is to quantify the waste, propose the list additions and note where a negative is doing harm. Repeat the same reading for Performance Max search terms insights where available.
Assess ads and assets
For each ad group with meaningful spend, check that responsive search ads have enough distinct headlines and descriptions, that ad strength is not uniformly poor, that pinning is used deliberately rather than everywhere, and that headlines speak to the keywords in the group. Review asset-level reporting for assets rated low, and the campaign and account-level assets: sitelinks, callouts, structured snippets, images, calls, lead forms, prices and promotions, their relevance and whether any are expired or disapproved. For Display, video and Performance Max, check asset coverage and formats against what the campaign type can use. Note ads that have not been refreshed in a long time and ad groups running one ad.
Evaluate bidding and budgets
For each campaign, record the bid strategy, its target, the conversions it had in the period, whether it is still in learning, and whether it is limited by budget. Compare targets with what the business can afford using the CPA calculator and the ROAS calculator, and compare actual return on ad spend (ROASRevenue generated per unit of ad spend.) or cost per acquisition with those targets. Check portfolio strategies and shared budgets for campaigns that should or should not share. Read the bid strategy report for recent target changes and their effect. The Smart Bidding strategies guide explains what each strategy needs; the finding is where a strategy is starved of data, mismatched to the goal, or given a target nobody can justify.
Review audiences and exclusions
List the audience segments applied to each campaign and whether they are set to observation or targeting. Check remarketing lists for size, source and whether the tag that populates them still fires. Read the audience report for segments with bid adjustments that no longer make sense under automated bidding. Review exclusions: placements and content on Display and video, demographic exclusions, customer lists used for exclusion, and brand exclusions on Performance Max. Look for campaigns that target and observe the same list, and for a complete absence of first-party audiences on an account that has customer data.
Audit Shopping and Performance Max
Where Shopping or Performance Max runs, start in Merchant Center: disapprovals, items pending, feed freshness, and the share of the catalogue eligible to serve. Then in the account: campaign priorities and product filters that overlap, product groups with a single "everything else" bucket, asset groups with thin creative or no audience signals, listing groups that exclude the products the business most wants to sell, and the brand traffic question, meaning whether Performance Max is taking brand searches that a Search campaign should own. Compare performance by product label or category against margin where the owner can supply it; a campaign that looks efficient may be selling the wrong products. For a fuller review of the catalogue side, the product feed and Performance Max audit scorecard rates the feed, labels, structure, values and reconciliation and ranks the fixes.
Inspect landing pages
Pull the final URLs by spend and open the top ones on a phone. Check speed, that the page answers the query the ad group targets, that the conversion action is reachable in one or two actions, and that the page returns a success status rather than a redirect chain. Compare conversion rate by landing page within the same ad group where several are used. Confirm the tag and consent banner behave on these pages specifically. Landing page findings are usually the highest-impact items the account team cannot fix alone, so write them for the web team with URLs and the spend they carry.
Prioritise and report
Score every finding using the sheet below, order by score, and write the top items as actions: what to change, where, the evidence, the expected effect in direction rather than in invented figures, and how to confirm it worked. Group findings into quick fixes, projects and questions that need data. Present measurement first regardless of score, because it changes how every other number should be read. End with a date for the next audit and the three things to do this week. A good audit report can be handed to someone who was not involved and acted on in order.
Verify the result
A complete audit has: a scoped brief; a measurement verdict with the actions, flags and tag checks listed; a structure map with proposed changes; a settings list with deviations; keyword, search term and negative findings with waste quantified; an ads and assets review by ad group; a bidding table with strategy, target, data and verdict per campaign; an audiences and exclusions list; Shopping and Performance Max findings anchored in Merchant Center; landing page findings with URLs; and a scored backlog with the first three actions named. Each finding carries evidence, affected spend and a verification step. A quick test of quality is to hand one finding to someone outside the audit and ask them to reproduce it from the evidence column alone; if they cannot find the report, the filter or the period, the finding is not finished. The second test is whether the owner can say, for each of the top items, what will be checked and when to confirm the fix worked.
Audit scoring sheet
Score each finding on impact and effort, then order the backlog by impact over effort, with measurement findings always first. Plotting the scored findings on an impact-effort matrix shows at a glance which are quick wins and which are larger projects to plan before the order is fixed.
| Column | What goes in it | Scale |
|---|---|---|
| Area | Measurement, structure, settings, keywords, search terms, ads, bidding, audiences, Shopping, landing pages | Fixed list |
| Finding | One specific problem, stated so it can be verified | Text |
| Evidence | The report, filter and period that show it | Text or link |
| Affected spend | Share of the period's spend that the finding touches | Low, medium, high |
| Impact | Expected effect on the primary KPI if fixed | Low, medium, high |
| Effort | Hours, who, and whether a third party is needed | Low, medium, high |
| Type | Quick fix, project, or question needing data | Fixed list |
| Priority | Impact and affected spend against effort, measurement first | Ordered |
| Verification | The report and the date on which the fix will be confirmed | Text |
The top of a backlog for a B2B software account
First, measurement: the account's only primary action is a "contact page view" that counts every visit, so reported conversions of 1,900 a month overstate real enquiries of around 140 by more than tenfold; the fix is a form-success action as primary with the page view secondary, and every bidding target reset against the honest baseline. Second, settings: four Search campaigns are opted into the Display Network and spend about £2,300 a month there with two enquiries between them. Third, search terms: roughly £4,000 a month goes to job-seeker and training queries that a shared negative list of 180 terms would stop. Fourth, structure: brand and generic share one campaign, so the reported £38 cost per enquiry hides a generic figure closer to £210. Fifth, landing pages: the highest-spend ad group lands on a page that redirects twice on mobile. The first three are quick fixes scheduled for this week; the restructure is a project for next quarter; the landing page goes to the web team with the URLs and the spend attached.
When you want the review turned into a working document, the Google Ads account audit worksheet sets out each check in order so findings can be recorded, prioritised and handed over.
An audit is also the quickest way to find what a change of agency left behind; the diagnosis of results that dropped after changing agency uses the same measurement, bidding and exclusion checks against the handover timeline.
Common mistakes
- Starting with keywords. If measurement is wrong, the keyword verdicts are wrong too.
- Reporting everything. Fifty findings of equal weight is a list, not an audit.
- Findings without evidence. "Ads could be better" cannot be acted on or verified.
- Auditing against a template. The question is whether the structure serves this business, not whether it matches a diagram.
- Treating the recommendations tab as the audit. It is one input, generated from the account's own data and Google's priorities.
- Inventing expected uplifts. State direction and reasoning; let the verification step supply the figure.
- Skipping landing pages because the account team does not own them. They often carry the largest finding.
- Never scheduling the next audit. Accounts drift; the audit is a routine, not an event.
The author's campaign management expertise page describes how this sequence is run on client accounts and how the backlog feeds the monthly management routine.
