Google Ads Audit: 20 Red Flags Often Seen in UK Accounts

Your dashboard says return is healthy, yet the bank balance and CRM tell a different story. Leads look plentiful in Google Ads, sales staff say quality is patchy, and every bid change seems to make results noisier. That's usually not a bidding problem. It's a measurement problem feeding every later decision, which is why a Google Ads audit should rank faults by how far they distort judgement, not by how easy they are to fix.

Triage Table for 20 Red Flags Before You Change Anything

Accounts with meaningful monthly spend often show a handful of these flags rather than all twenty. Triage them by how far they can distort decisions. Start with measurement: if conversion data is wrong, changes to structure, keywords, bids or budgets can send spend in the wrong direction.

Before changing settings, pull a 90-day export of campaigns, search terms and conversions. Gather access to conversion settings and change history, then match the same dates against CRM or order records. The wider market context shows why it pays to check spend carefully: IAB UK, working with Oliver Wyman reported that the UK digital advertising market exceeded £40 billion in 2025, growing 10% year on year. Search accounted for 44% of total digital advertising expenditure, with investment reaching £17.9 billion after 6% annual growth.

The order matters. First establish whether reported outcomes represent real sales or qualified leads. Then assess how campaigns divide spend, whether bidding targets reflect margin, and which queries and ads attract the traffic. A platform-reported return is a claim to check against business records, not proof that the account is profitable.

Use the table to triage, not to treat every flag as an automatic fault. Rank each one by spend exposure and its effect on decisions. Some findings need judgement: a campaign split may be deliberate, and a flagged term may still bring valuable customers. Confirm the likely impact before changing settings, then fix the highest-risk issue first.

Tracking Red Flags That Skew Every Decision

The surprise in many audits is simple. One sale fires twice. A browser tag records the purchase, then a server-side or imported action records it again, so Google Ads reports two conversions and Smart Bidding learns from an inflated value.

It usually looks like platform conversions sitting close to double the CRM count, often with near-identical timestamps. Place a test order, follow the event trail in Tag Assistant, and keep one authoritative action per real outcome. If you're rebuilding the setup, work through what conversion tracking should record and where it breaks before you touch bids.

Micro-events Set as Primary Goals

A page view, basket add or brochure download isn't worthless. It just isn't a sale or a qualified lead. When you set those actions as primary, bidding optimises towards cheap, early behaviour and cost per conversion looks excellent while revenue stays flat.

Move micro-events to secondary unless you can show they correlate tightly with revenue. Bid on completed purchases, qualified enquiries and imported offline outcomes that sales staff have verified. Keep the softer events for observation and audience building.

Platform Figures That Do Not Match the Business

Export Google Ads conversions, GA4 key events, CRM outcomes and order records for the same period. Reconcile them by campaign, device and date. Calculate variance as 1 minus platform conversions divided by CRM conversions, then investigate large gaps by browser, landing page and lead type.

Say a lead generation account reports strong form completions on mobile, but the CRM shows few contactable enquiries from the same campaigns. Don't raise bids because the interface looks efficient. Check for spam submissions, duplicate fires on thank-you page refreshes, and forms counted before validation.

Practical rule: Treat platform-reported return as a claim to verify against orders and qualified leads, not as the truth itself.

Consent sits underneath all of this. The ICO guidance on storage and access technologies states that cookies, tracking pixels, link decoration, local storage, and scripts/tags fall within UK rules on storage and access technologies, and for online advertising, including campaign-effectiveness measurement, consent is generally required where such technologies are used.

Enhanced Conversions or Consent Mode Missing

Check whether Consent Mode signals reach every tag, whether sitewide tagging is consistent, and whether Enhanced Conversions are active for each relevant conversion source. Document what happens when a user denies consent. A lower reported conversion rate after you tighten consent handling can mean cleaner governance, not weaker advertising.

Don't paper over gaps by importing every GA4 event you can find. I'd rather bid on fewer, verified conversions than scale an account on duplicated or legally unusable signals.

Structure, Bidding and Budget Red Flags That Waste Spend

A ROAS target is only as good as the margin behind it. Set a target because it looks respectable, and you can hit it every month while losing money on product cost, fulfilment and returns.

Structure That Fights Itself

Theme mixing is the first structural fault to clear. One campaign or ad group holds unrelated intents, for example emergency call-out searches beside DIY advice queries, or premium products beside clearance lines. One bid and one set of ads can't suit both, so you overpay for the weaker intent and underbid the stronger one. Split by intent and margin where the economics genuinely differ, and write copy for each job.

Check location options next. If targeting is set to people interested in your location rather than people present there, a UK service business can pay for clicks from abroad. Switch to presence for local and national UK delivery unless you have a deliberate reason to reach people researching from overseas.

Then look for the opposite problem, campaigns split too thinly. Separate campaigns by device, region and minor product variant can each gather too few conversions for bidding to learn. Consolidate splits that don't show a real difference in conversion rate or value. A split needs an economic reason, not just a tidy naming convention.

• Keep a split when margin, intent or conversion value differs clearly between groups.
• Merge a split when performance is near identical and each campaign is starved of conversions.
• Test a split with a defined budget and end date, rather than leaving it running indefinitely.

Targets With No Margin Maths

Break-even ROAS equals 1 divided by gross margin on ex-VAT revenue. Use ex-VAT figures, because VAT collected for HMRC isn't revenue you can spend on clicks.

For lead generation, do the same work backwards from close rate and average order or contract value to a defensible target CPA. If you need the mechanics, Google Ads bid management covers how targets and bid strategies interact. Don't copy a target from another campaign with different margins.

Budgets and Constant Editing

Sort campaigns by impression share lost to budget. If your strongest non-brand campaign is capped by midday while a weak campaign spends freely all month, move budget from the loser first. That's a cleaner test than raising total spend.

Finally, open change history. Weekly target changes, repeated strategy switches and campaigns running on thin conversion volume keep resetting learning. Change one variable, note the date, and let the learning period finish before you judge it. Some of these calls are judgement, not mathematics, but constant interference isn't optimisation.

Keyword and Search Term Red Flags to Fix First

Start with broad match running on unreliable conversion data. Broad keywords on Maximise Conversions or a target strategy will explore widely. If conversions are duplicated, too soft or too few, that exploration buys unrelated queries and calls it discovery.

Pull back to phrase or exact match in the affected campaigns, add negatives for the clear mismatches, and only widen again once the conversion signal is stable. I wouldn't ban broad match outright. I'd refuse to fund it blind.

Search Terms Nobody Has Read

Pull the last 90 days in the search terms report and sort by cost. Work down the spend-bearing terms with no conversions and ask one question of each: would you pay for this query again on purpose?

• Add negatives at campaign level for intent you never want, such as jobs, free, DIY or wholesale queries you don't serve.
• Use shared negative lists for account-wide exclusions, so a fix in one campaign doesn't leak elsewhere.
• Promote converting terms into exact match in the right ad group where you want tighter control of copy and landing page.

This is unglamorous work and it's where wasted spend usually sits. Good keyword research for PPC should be refreshed from real search terms, not just planner estimates.

Brand Return Hiding Weak Non-Brand Performance

Brand queries normally convert cheaply because the searcher already knows you. Blend them into generic campaigns and headline ROAS looks strong while non-brand prospecting slowly loses money without anyone noticing.

Segment brand versus non-brand and judge non-brand on its own cost per qualified lead or margin-based return. Decide separately how much brand defence you need, given organic visibility and competitor activity.

Keywords Competing With Themselves

Filter the keyword list for duplicates across ad groups and campaigns. The same keyword in several homes splits data, creates inconsistent ads and landing pages, and makes it unclear which version you intended to win. Consolidate each keyword into one home, then use negatives to route queries where campaigns genuinely overlap.

One caution before you declare a tracking failure. Google's search terms report guidance warns that conversion figures can differ between the main search terms report and search terms insights because conversion data is processed differently and may be affected by conversion lag, and store visits and store sales conversions are not compatible with the search terms report. Record the source, attribution setting, reporting surface and lag before you change bids.

Ad and Asset Red Flags That Lower Click Quality

Ad copy isn't just a creative preference. Weak or unsupported copy changes who clicks, what they expect and whether they convert once they land.

Thin responsive search ads are the common fault. One RSA per ad group, a handful of near-identical headlines, and heavy pinning leave Google few sensible combinations to test. The result isn't neutral. Your weakest message gets shown repeatedly because nothing better exists.

Build at least two RSAs per ad group with headlines covering the core benefit, a proof point, the offer and a clear action. Avoid pinning unless compliance or a genuine ordering issue requires it. Check asset ratings and replace headlines marked Low rather than adding more variations of the same claim.

Claims the Page Cannot Support

Words like best, cheapest, fastest and free delivery need evidence on the landing page, not just confidence in the ad. Compare every headline, description and sitelink claim against the destination page and remove or substantiate anything that doesn't hold up.

The UK Government review of online targeting states that UK paid search falls within the CAP Code, which covers non-broadcast advertising including online advertising, and the code can apply to claims on an advertiser's own website and other controlled, non-paid spaces. Your landing page isn't outside advertising compliance just because you own it.

If the page cannot prove the claim in the ad, the claim comes out, however well it converts.

Missing Assets

Check the assets tab for missing sitelinks, callouts, structured snippets and call assets. Without them your ad takes less space, gives searchers fewer routes to the right page and usually earns a weaker click-through rate.

Point each sitelink to a distinct, relevant page, not four versions of the same category. Use callouts for delivery, returns, warranty or opening hours only where they're accurate. Add a call asset where phone enquiries are a real, tracked outcome, and make sure call recording and qualification are handled properly in the CRM.

Landing Page and Performance Max Red Flags

A Performance Max campaign reporting excellent ROAS often turns out to be harvesting people who searched for the business by name. It looks like growth. Mostly, it's recaptured demand that Search was already winning cheaply.

Start, though, with the page every click reaches. Check final URLs by ad group. If an ad promises a specific service, product or offer and sends people to the homepage or a broad category page, the searcher has to do the work you should've done. Many won't.

Map each ad group to a page with a matching headline, offer and price, then test the page on a mid-range phone over a mobile connection. Slow mobile pages don't just frustrate visitors. They raise the effective cost of every conversion because you pay for clicks that never see the form or basket. Fix obvious culprits first, such as oversized images, heavy scripts and pop-ups that block the main action.

Brand Absorption in Performance Max

Check whether brand exclusions are applied and whether account-level negatives exist. Without them, Performance Max can absorb brand queries across Search and other placements, inflate its own conversions and starve your brand Search campaign of data.

Add the brand exclusion, then watch brand Search impression share and non-brand Performance Max results separately. Expect reported Performance Max return to fall. That's often the honest number appearing for the first time.

Blended ROAS With No Query Review

Google's Performance Max search terms report exposes the queries that triggered ads and helps identify search terms that generated conversions. Use it to find irrelevant queries, pull converting language into Search campaigns or new assets, and check whether landing page selection is sending commercially irrelevant traffic.

Apply sensible Performance Max best practices, but don't set budgets from blended ROAS alone. Segment brand from non-brand, review asset group performance, and for retailers, isolate low-margin or high-return products in the feed before you ask the campaign to scale.

Confirm Tracking Before You Change Bids

The visual below sets out the order. Measurement first, bids last.

Work through the account this week in this sequence:

1. Reconcile conversions against CRM or order records and remove duplicate and micro-event goals, so bidding learns from real outcomes.
2. Confirm consent settings, Consent Mode and Enhanced Conversions for each source, so the data you keep is usable.
3. Separate brand from non-brand, review spend-bearing search terms and add negatives.
4. Correct location settings and consolidate campaign splits too thin to learn.
5. Reset bid targets against margin and move budget away from weak campaigns.
6. Tighten RSAs, claims and assets so clicks match what the page delivers.
7. Review landing pages and Performance Max with brand exclusions and search terms checked.

The walkthrough below shows the checking mindset in practice.

Change one variable at a time and wait out the learning period before judging. If you'd like a second opinion, PPC Geeks offers a free Google Ads audit as one option.

If your account shows several of these flags and you don't have the time to unpick tracking, search terms and margin-based targets yourself, PPC Geeks can audit the account and set out the fixes in priority order. Visit PPC Geeks to request that review and get a clear view of where your spend is working and where it isn't.