In shortA Google Ads audit checklist in the order that matters: what to look at, what bad looks like and why each check matters for real enquiries.
A Google Ads audit should check, in order: who owns the account, whether conversions record real enquiries, whether won customers feed back in, campaign settings, search terms and negatives, match types and structure, ads and assets, landing pages, budget and bidding, and whether reporting shows enquiries or clicks. Tracking comes first because every other judgement depends on it.
That order is the point. Many audits open with ad copy because it is easy to comment on. But if the conversion data is wrong, every verdict on a campaign is unreliable, and Google's automated bidding is learning from the same wrong numbers. I run every free Google Ads account audit myself, and this is the sequence I follow. You can use it on your own account with read-only access.
Key Takeaways - Check ownership and tracking before judging any campaign. - A conversion should be a real enquiry: a completed form or a call long enough to be a conversation, not a page view or a tap on a number. - A useful audit ends with a prioritised list of fixes, and can honestly conclude "do not spend yet".
1. Who owns the account, and who can get in?
What to look at. Admin, then Access and security. List every user and their access level, and check who controls billing. The account should sit under a Google login the business controls, paid by the business's own card. An agency or freelancer should be linked in or given a user login, not hold the account itself. Google's guide to managing account access explains the levels.
What bad looks like. The account was created under an agency's email. A former employee or previous agency still has admin access. Nobody at the business can log in. Ad spend arrives bundled into an agency invoice, so the business never sees what Google charged.
Why it matters. If you do not own the account, you do not own its history, and years of conversion data can leave when a contract ends. That is why I only work in the client's own account, in the client's name.
2. Do the conversion actions record real enquiries?
What to look at. Goals, then Conversions, then Summary. For each action, note the source (Google Ads tag, Analytics import or calls), whether it is primary or secondary, its status and its counting setting. Then test it: complete a genuine enquiry with Google Tag Assistant running and confirm the conversion fires on a successful submission, not on page load or a button press.
What bad looks like.
- The primary conversion is a page view, a scroll or a tap on the phone number.
- The same form is counted twice, by the Google Ads tag and by an imported Analytics event, both primary.
- Lead actions count "every" conversion, so a reloaded thank-you page becomes a second lead. For enquiries, "one" per click is usually right.
- Calls count on the tap rather than by duration. Calls from ads should run through a Google forwarding number with a minimum call length that separates conversations from wrong numbers.
- The status shows no recent conversions, and nobody noticed.
Why it matters. Primary conversions are what automated bidding optimises towards. Count the wrong thing and the account gets very good at producing the wrong thing. Google's overview of conversion measurement covers the settings. Tracking also breaks silently: a site rebuild, a new form plugin or a consent banner change can stop a tag firing while the ads keep spending.
3. Does the account know which enquiries became customers?
What to look at. Whether enhanced conversions for leads are on, and whether anything goes back to Google after the enquiry: a qualified lead, a won job, a sale value. Check for imported conversion actions and when each last received data.
What bad looks like. The account's idea of success stops at the form. A spam submission and a signed contract count exactly the same.
Why it matters. Enhanced conversions use hashed details from the form, such as an email address, so a lead can later be matched back to the ad click. Offline conversion imports then tell Google which of those leads became customers. Bidding then learns which searches produce customers, not just form fills. This uses personal data, so the privacy notice should cover it; the ICO's UK GDPR guidance is the place to start.
4. Are the campaign settings spending where you expect?
What to look at. Each campaign's location options, networks, languages and ad schedule. Then open Recommendations and check what is set to auto-apply.
What bad looks like.
- Location targeting on "presence or interest", the option Google marks as recommended, which can show ads to people outside your area who have only shown interest in it. A local service business usually wants "presence": people in, or regularly in, the area. The locations report shows where clicks came from.
- A Search campaign also opted into the Display Network or third-party search sites, with nobody having chosen that.
- Auto-applied recommendations that add keywords, broaden match types or change the bid strategy without a person deciding.
- A Performance Max campaign free to send traffic to any page, with URL expansion and brand settings never reviewed. Google's Performance Max overview explains what it can reach.
Why it matters. These are defaults and switches rather than strategy, which is why they get missed. Each can move spend away from the searches and places the business serves.
5. Are search terms reviewed and negatives maintained?
What to look at. The search terms report, which shows the searches that actually triggered ads rather than the keywords you bid on. Sort by cost over a meaningful period, then review negative keywords and shared negative lists.
What bad looks like. Spend on job seekers, DIY and "how to" searches, competitors by name, or places and services you do not cover. No negative lists at all. Or negatives so broad they block searches that would have converted. Negative keywords do not match close variants, so plurals and misspellings may need their own entries.
Why it matters. This is the most direct lever on waste, and it is weekly work rather than a set-up task. Some terms are hidden from the report for privacy, so treat it as a large sample of your traffic rather than all of it.
6. Do the match types and structure make sense?
What to look at. Match types by campaign, the bid strategy each runs on, how many keywords sit in each ad group and whether brand and non-brand searches are separated.
What bad looks like. Broad match on manual bidding, or on automated bidding fed by the broken conversions from item 2. Ad groups holding dozens of loosely related keywords that share one ad and one page. The same keyword in several campaigns, competing with itself. Brand searches mixed with generic ones, so cheap clicks from people who already know you flatter the numbers.
Why it matters. Structure decides which ad and which page a search gets. Broad match relies on Google's reading of intent and on good conversion data; without that data it wanders. Keeping brand separate is the only way to see what new customers really cost.
7. Are the ads and assets relevant and working?
What to look at. Ad status and policy issues, the text of each responsive search ad, pinning, and assets such as sitelinks, callouts, call and location assets.
What bad looks like. Disapproved or limited ads nobody has opened. Generic headlines that could belong to any competitor. Every headline pinned, removing the variation responsive ads depend on. Sitelinks to pages that no longer exist. Claims about prices, response times or areas that the business does not stand behind or the page does not repeat.
Why it matters. Ad relevance and expected click-through rate are part of how Google assesses ad quality. Ad strength is a guide to how varied your text is, not a measure of enquiries, so do not chase it at the expense of accuracy.
8. Does each landing page match the ad and load quickly?
What to look at. The final URL of every active ad, opened on a phone. Read the search term, the ad and the page together.
What bad looks like. Every ad pointing at the homepage. A town-specific ad landing on a page that never mentions the town. A slow mobile page, a form that fails, or a redirect that strips the click identifier so conversions cannot be tied back to ads.
Why it matters. You pay for the click either way. Google's landing page experience guidance covers relevance and usability, and the Google Ads landing page checklist goes through the page itself in detail.
9. Do budget and bidding fit the conversion volume?
What to look at. Campaign status (for example "Limited by budget"), search impression share lost to budget versus lost to rank, and each bid strategy against the number of real conversions its campaign records.
What bad looks like. Maximise conversions or a target cost per acquisition on a campaign with only a handful of genuine conversions, too few to learn from. A target so tight the campaign barely spends. Bid strategies changed every few weeks, restarting the learning. Budget spread so thinly that no campaign gets enough traffic to judge.
Why it matters. Budget problems and rank problems need opposite fixes, and more money does not help a campaign that is losing auctions on relevance. On one Yorkshire campaign I run, the finding was rank-limited rather than budget-limited, so the fix was splitting it into single-city ad groups, not raising the budget.
10. Does reporting show enquiries and sales, or just clicks?
What to look at. The last few reports the business received, and whether their numbers can be reconciled with the enquiries and sales the business actually recorded.
What bad looks like. Reports led by impressions, clicks and click-through rate. A conversions figure that includes secondary actions or duplicates. No cost per enquiry, no cost per customer, and nobody comparing the platform's figures with the phone log, CRM or invoices.
Why it matters. Platform figures are a model of what happened; the business's own records are what happened. Procon 24/7, a Yorkshire concrete supplier, is measured off its own revenue reports, not the Ads dashboard. When a site rebuild silently stopped the conversion tag firing, that outside record kept the numbers trustworthy until the fault was caught. From 2022 to September 2026 those reports show £1,337,693 of tracked revenue on £150,443 of ad spend, an 8.9x return and 2,055 new customers. That is one client's result, not a forecast. It also carries a lesson I own: call tracking should have gone in during month one.
What should you get at the end of an audit?
A prioritised list, not a score. Fixes that stop wasted spend or broken tracking come first; copy and structure come after. Each finding should say what was found, where and what to change.
A good audit is also allowed to say "do not spend yet". If tracking cannot see enquiries, the site does not convert, or a job's value cannot support the cost of the clicks, more budget only makes the loss bigger. Is Google Ads worth it for a small business? walks through that arithmetic.
If you would rather someone else worked through this list, the Google Ads audit is free, done by me personally, and can end with advice not to spend. If you then want the account run for you, Google Ads management is a flat monthly fee, never a percentage of ad spend, given on the audit call.
