Wasting money on Google Ads? 7 places it's actually leaking

Most advertisers who feel like they're wasting money on Google Ads are right — but the leak is almost never where the dashboard is pointing. The Recommendations tab flags a 12% "optimisation score" issue and you chase that. Meanwhile the real waste — the stuff burning 25–40% of the budget — sits in structural choices nobody's looked at in months.

We audit roughly 40 Google Ads accounts a year. The leaks repeat. These seven show up in most accounts spending under £20k a month, and they compound.

1. Broad match bleeding into irrelevant queries

Broad match in 2026 is not what it was in 2018. Google's matching is smarter — but it's also aggressive, and it optimises to impressions and clicks rather than what actually converts for you.

Open the Search Terms report on any broad-match campaign and sort by spend, descending. You'll almost certainly find queries that made no business sense: too generic, wrong intent, wrong geography. Each click on those is budget that should have gone to a converting query.

The fix isn't to turn broad match off. It's to run broad match alongside a proper negative-keyword list, review search terms weekly, and add the obvious losers as negatives the moment they appear. If nobody's looked at search terms in the last 30 days, that's your first leak — and it's usually the biggest.

2. Performance Max stealing from Search

This is the quiet budget-killer of the last two years. Performance Max is allowed to bid on Search by default. Without exclusions, it will happily bid on the same high-intent queries your Search campaigns already cover — including brand terms.

What that looks like in reporting: Search campaign impressions and clicks drop. Performance Max reporting shows strong performance. Account-level conversions are flat or down, because you're now paying more for the same conversions you already had.

The two-line fix: add your brand terms as negative keywords applied to Performance Max (now possible directly in the UI), and segment branded queries into their own dedicated Search campaign with a higher position target so Pmax can't undercut it.

3. Smart bidding set to the wrong goal

Target CPA and Target ROAS are only as smart as the conversion signal you feed them. If your conversion action includes every form submission — including spam, low-intent enquiries, and freebie-downloaders — the algorithm is going to optimise for the cheapest of those, not the ones that actually close.

This is the quietest of the seven leaks because the numbers look fine. CPA is hitting target, form submissions are up. Sales pipeline is not.

The fix: separate conversions by value. Enhanced Conversions for Leads to pass offline sales data back. Conversion Value Rules to weight high-value leads higher than low-value. Then retarget smart bidding at the weighted value, not the count. Once the algorithm is optimising toward real revenue, it stops chasing cheap junk.

4. Ad groups with 40 keywords and one generic ad

We still see this in 2026. A single ad group with dozens of loosely-related keywords, and one or two RSAs written for the ad group rather than the query clusters inside it.

Google now auto-generates responsive search ad combinations, but it can only generate within the asset pool you gave it. If your headlines and descriptions are too generic to speak to any specific intent, quality score drops, CPCs rise, and click-through rate collapses. You end up paying more for worse positioning.

The fix: tightly-themed ad groups (5–15 keywords maximum with a single intent), and at least one RSA where every headline matches the intent of that cluster. This is boring structural work. It is also the biggest free lever for lowering CPCs most accounts have.

5. Location targeting set to "Presence or interest"

Buried in campaign settings is a targeting option most advertisers never touched. The default for Google Ads campaigns in 2026 is still "presence or interest in your targeted locations" — which means your UK campaign can serve to someone in France who once searched for something UK-related.

For service businesses especially, this is a meaningful leak. If you're an accountant in Manchester, a click from someone in Dubai browsing about UK accountants is not a lead.

The fix takes 30 seconds per campaign: set to "Presence: People in or regularly in your targeted locations." Everywhere. On every campaign. It won't fix a broken account but it stops a steady trickle of irrelevant spend.

6. No negative-keyword list maintained at account level

Most accounts have some negatives. Most accounts do not have a properly-maintained, account-level negative keyword list that gets added to weekly.

What should be on it: competitor brand terms you don't want to bid on, irrelevant industries, generic words that always pull the wrong traffic ("free", "jobs", "salary", "download" depending on your business), long-tail queries that historically don't convert.

Without this list, every new campaign starts from zero, and the same lessons get re-learned (paid-for) in each new campaign. Build it once, apply account-wide, update weekly. The effort is an hour a week. The saving is typically 10–15% of spend within a month.

7. Attribution still on last-click

Last-click attribution pretends every conversion came from the final ad click. It didn't. It came from a journey — a YouTube view three weeks ago, a Discovery impression, a Search click on a brand term, and then the final click that got the credit.

With last-click, any campaign that plays a supporting role (YouTube, Display, Discovery, broad-match top-of-funnel) looks unprofitable, gets its budget cut, and the account quietly collapses inward to brand and branded-non-brand Search. Conversion volume looks stable for a month, then starts declining, because the top of the funnel has been switched off.

The fix: move to data-driven attribution (default for new accounts, still not enabled on many older ones), and report on both the ad-platform number and the blended CAC at the business level so you can see when the funnel's contracting.

How to work through these

Don't fix all seven at once. In order of impact, and how long each takes:

• Search terms cleanup + negative list (hours, big impact)
• Performance Max exclusions (minutes, big impact if you run both)
• Location targeting (minutes, small-medium impact)
• Ad group restructuring (days, compounds over months)
• Conversion value rules + offline sales feedback (days, changes bidding quality permanently)
• Attribution model change (minutes to switch, weeks to re-evaluate)

If you do nothing else this month, look at your search terms and your Performance Max exclusions. Between them they're where most of the waste lives.

When it's time to get another set of eyes on the account

The account has been running for over a year, and the person who set it up is no longer involved. Your CPA has been drifting upward for three months and nobody can explain why. You spend £5k+ a month and haven't had a structured review since the last algorithm change.

In any of those cases, a second set of eyes catches things the day-to-day operator stops seeing. Book a free Google Ads audit — we'll run through these seven leaks on your account specifically and send back what we find. No slide deck, no sales pitch — just the numbers and what to do about them.