Scope: what this audit covers and what it does not
The scorecard covers one Google Ads account and the settings inside it: how conversions are recorded and imported, how campaigns and ad groups are organised, which bid strategiesThe rule an ad platform follows to set the bid in each auction. are used and what they are fed, how keywords and search terms are managed, the quality and coverage of ads and assets, whether landing pages fit the traffic, and how budget is spread across campaigns. It works for lead generation and ecommerce accounts alike, and for search, shopping and broader campaign types where the same principles apply.
It deliberately does not teach the review. The step-by-step method for opening each report and reading it sits in the Google Ads account audit guide, and the pass-or-fail list to work through while fixing is the Google Ads account audit checklist. It also stops at the edge of the account. Whether the analytics property agrees with Google Ads, whether the website is technically healthy, and whether the offer is competitive are separate questions with their own scorecards in the audit library. Creative strategy and market sizing are out of scope too; the scorecard can say an ad group lacks ad variety, not what the message should be.
Before you score: access and evidence
Ask for standard access at minimum, because read-only users cannot see some conversion and billing settings. Agree a fixed evidence windowThe fixed period of data an audit scores, agreed before scoring, long enough to cover a full cycle for the business and kept the same at every rescore. before you start, long enough to cover a full buying cycle for the business and recent enough to reflect the current build, and use the same window at every rescore. Collect four things before rating anything: the list of conversion actions with their status, category, counting method and whether each is primary; the change history for the window; the search terms reportThe Google Ads report of the actual searches that triggered an account's ads.; and the business targets the account is meant to hit, written as a target CPACost of each conversion. or target ROASRevenue generated per unit of ad spend..
Write down the business targets in the brief rather than reading them off the bid strategy, because a target set in the platform is often a historical accident. If you need to sanity-check a target, the CPA calculator and the ROAS calculator turn margin and order value into a ceiling the account can be measured against. Finally, note anything you cannot see. An unscoreable criterion is recorded as missing evidence, not quietly skipped.
Criteria and scoring
Rate each row on the three-level scale. Rows marked as gates decide the band on their own if they score missing. Score from evidence in the account, not from what the previous manager says was set up.
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Conversion actions (gate) | Each action maps to a real business outcome, fires once per outcome and is verified by a test | Missing: no working primary action / Partial: works but duplicated or untested / Solid: verified and documented |
| Primary versus secondary goals (gate) | Only actions the business values are primary; micro-actions are secondary | Missing: everything primary / Partial: mostly right with stray micro-actions / Solid: deliberate split |
| Conversion values | Values reflect revenue or a lead value agreed with sales | Missing: no values / Partial: static guesses / Solid: dynamic or agreed values |
| Campaign structure | Campaigns split by goal, budget need or geography, not by habit | Missing: one catch-all or sprawl / Partial: logic unclear in places / Solid: each campaign has a stated purpose |
| Ad group themes | Tight themes so ads and pages match the keywords they serve | Missing: mixed intents / Partial: some broad groups / Solid: clear themes throughout |
| Bid strategy fit (gate) | Strategy matches the goal and has enough conversion data to learn | Missing: smart bidding on broken data / Partial: right type, poor targets / Solid: fit and fed |
| Bid targets | Targets come from margins and agreed thresholds | Missing: none or arbitrary / Partial: set once, never reviewed / Solid: tied to business targets |
| Keyword coverage | Core terms covered with match types chosen on purpose | Missing: obvious gaps / Partial: coverage but careless match types / Solid: deliberate coverage |
| Search terms and negatives | Search terms reviewed on a rhythm; shared negative lists maintained | Missing: never reviewed / Partial: occasional, ad hoc / Solid: regular review with lists |
| Ad relevance and quality | Ads mirror the theme; relevance diagnostics show no persistent weak spots | Missing: generic ads / Partial: patchy relevance / Solid: relevant ads across the account |
| Assets coverage | Sitelinks, callouts and other assets set at the right level and current | Missing: none / Partial: out of date or thin / Solid: full and current |
| Landing page fit | Each ad group sends to a page that matches intent and loads well on mobile | Missing: homepage for everything / Partial: some mismatches / Solid: matched pages |
| Budget allocation | Budget follows return; limited campaigns are the ones that deserve more | Missing: spread evenly or by history / Partial: some misallocation / Solid: allocation follows results |
| Lost impression share | Lost share from budget and from rank is understood and acted on | Missing: not looked at / Partial: noted, no action / Solid: reviewed with a decision |
| Naming and change discipline | Consistent names; change history shows deliberate, logged changes | Missing: chaos / Partial: partial convention / Solid: convention followed |
Two rows need judgement. Ad relevance is read through the diagnostics the Google Ads Help centre documents, including the components behind Quality ScoreGoogle's 1–10 keyword diagnostic.; score the pattern across the account, not a single keyword. Lost Impression ShareShare of eligible impressions you received. is only a problem when the campaign losing it is profitable, so read it alongside the targets you wrote down.
How to score and prioritise
Score the gates first. If conversion tracking is missing, everything that depends on it, from bid strategy to budget allocation, is being judged on faulty numbers, and the honest move is to mark those rows provisional. The conversion tracking set-up guide is where the repair starts. When the gate scores partial because actions are duplicated or the wrong ones are primary, the conversion action clean-up tutorial repairs the existing set rather than rebuilding it. Then score the remaining rows in the order money flows: structure, bidding, keywords, ads, pages and budget.
Prioritise fixes by two questions: how much spend does the weakness touch, and how hard is it to fix? A partial score on search terms in a campaign that takes most of the budget outranks a missing asset in a small brand campaign. Weaknesses that touch most of the spend and can be fixed quickly go to the top. Large structural rebuilds go on the list but are scheduled, not rushed, because restructuring resets learning. Where bidding is the weak area, the smart bidding strategies guide explains which strategy fits which goal and data volume.
Reading the result: what to do with each band
Red means a gate scored missing. Freeze structural changes, stop raising budgets, and fix the tracking or bidding foundation first. Any target changes made now would be tuned to bad data. Rescore the gates once the repair has recorded a full cycle of conversions.
Amber means the gates hold but a large share of other rows are partial or missing. Build a fix list of a handful of items ranked by spend touched, assign an owner to each, and work through them in order before starting any new tests. Avoid the temptation to fix everything at once; overlapping changes make it impossible to tell what helped.
Green means the gates are solid and most other rows are too. The account is not the bottleneck. Keep a light quarterly rescore, and move effort to experiments on offers, landing pages and audiences.
From audit to action plan
Turn every partial or missing row into a line in the plan with four fields: the finding in one sentence, the fix, the owner and the date it will be rescored. Link each fix to the guide that explains it so whoever picks it up does not need the auditor in the room. Share the band and the top items with the budget holder, not the full table; the table is the working document.
An inherited account scored amber
A regional plumbing firm hands its Google Ads account to a new freelancer. The account spends £4,000 a month across six campaigns. Conversion tracking scores partial: calls and forms both record, but form submissions count twice because a thank-you page and a button click are both primary. Bidding scores partial, with a target CPA of £35 that nobody can explain. Search terms score missing; a review finds roughly a fifth of spend going to job-seeker and DIY searches. The band is amber. The plan puts the duplicate conversion first, the negative keyword lists second and the CPA target third, recalculated from a £120 average job margin. The account is rescored after eight weeks and moves to green on every gate.
Common mistakes
- Scoring from what you were told. The previous manager's notes describe intentions. Score what the account and its change history show.
- Treating every finding as equal. A missing sitelink and double-counted conversions are not the same size of problem. Gates exist so the second one cannot hide among the first.
- Changing the evidence window between audits. A different window makes the rescore incomparable. Fix it once and keep it.
- Optimising bids on a red account. Tuning targets on broken conversion data trains the system to find more of the wrong outcome.
- Rewriting the walkthrough into the scorecard. Keep method in the guide and the tick-box pass in the checklist; the scorecard only rates and ranks.
