Paid Ads Waste After the 2025 Changes: A Self-Check

Why this article changed

This page was first published as "Paid Advertising in 2025: Why Businesses Waste Money". That year is over, but the changes it brought are now how the platforms work. In 2025 Google launched AI Max for Search and added controls to Performance Max, including campaign-level negative keywords and channel reporting (Google). Meta broadened and renamed its automated sales campaigns and rolled its Andromeda retrieval engine across its apps. Consent and server-side measurement became the norm rather than an extra.

The result: automation now makes most spending decisions, and waste has moved from bad bids to bad inputs. The self-check below finds those inputs.

The self-check

Assessment

Is your paid advertising leaking budget?

Answer for the account or accounts with the most spend. Yes means it is true and you could show evidence; partly means true for some campaigns or not checked recently; no means not true or not known. Score 2 for yes, 1 for partly, 0 for no.

1. 1. Every campaign bids towards a real business outcome (a sale or a qualified lead), with micro-actions such as page views set as secondary.

   Automated bidding finds whatever you count. Counting cheap actions buys cheap actions.

   YesPartlyNo
2. 2. Lead quality or order value is fed back to each platform from your CRM or shop (for example enhanced conversions for leads or a Conversions API).

   Without feedback, the platforms cannot tell a good lead from a bad one and optimise for volume.

   YesPartlyNo
3. 3. For UK and EEA visitors, a consent banner sends valid consent signals to Google and Meta, and ad pixels only fire where consent allows.

   Missing consent signals shrink measurement and remarketing audiences, and firing pixels without consent is a compliance risk.

   YesPartlyNo
4. 4. Someone reviews search terms at least weekly, including Performance Max search term insights, and maintains negative keyword lists.

   Broad match, AI Max and Performance Max all extend matching beyond your keywords.

   YesPartlyNo
5. 5. Brand searches are separated from prospecting, so non-brand and Performance Max results are not flattered by people already looking for you.

   Brand traffic converts well almost regardless of the ad, which hides weak prospecting.

   YesPartlyNo
6. 6. Automatic features (auto-applied recommendations, final URL expansion, generative creative enhancements) were switched on or off deliberately, and are reviewed.

   Defaults can widen targeting, raise budgets, send traffic to weak pages or alter creative without anyone deciding.

   YesPartlyNo
7. 7. New creative concepts with genuinely different angles go live at least monthly, and AI-generated imagery is checked for accuracy.

   Meta's delivery relies on creative variety, and tired or misleading creative raises cost and risk.

   YesPartlyNo
8. 8. Each month, platform-reported results are reconciled with revenue or qualified pipeline in your own systems, by channel.

   Every platform credits itself generously; only your own data shows what the spend really produced.

   YesPartlyNo

Answered 0 of 8. Your result appears when every question is answered.

What each score range means

• 0 to 6: Leaking. Automation is spending your budget on goals and traffic nobody chose. Reported results are probably unreliable. Next step: Fix measurement first: questions 1, 2 and 3. Pause or reduce spend on campaigns you cannot measure until they are fixed.
• 7 to 11: Partly controlled. The foundations exist but some automation runs unchecked, or reporting is not reconciled. Next step: Work through the questions you scored 0 or 1 in order of the spend behind them, starting with search terms and brand separation.
• 12 to 16: Controlled. The account is set up to spend on outcomes and to show you what happened. Next step: Shift effort to creative testing and target setting, and repeat this check each quarter as features change.

What each question costs you when it is a no

Q1, wrong goal. The platform hits the target you set, so a campaign optimising for form starts or add-to-baskets will report good results while revenue stays flat. Fix: one primary conversion per goal, everything else secondary.

Q2, no quality feedback. A B2B account that counts every form fill as equal will drift towards whoever fills in forms most readily. Fix: import qualified stages from the CRM through enhanced conversions for leads and the Conversions API. Upload methods change, so check Google Ads Help for the method Google currently supports and confirm that older integrations still send data.

Q3, consent gaps. Google requires consent signals for measurement and remarketing for EEA traffic (Tag Manager Help), and the ICO's guidance on storage and access technologies covers tracking pixels as well as cookies (ICO). Fix: a consent platform that integrates with Google's consent mode and your other tags, tested on a live page.

Q4, unread search terms. Every expansion of matching buys some searches you would never choose. Fix: a weekly review with a shared negative list, now usable in Performance Max at campaign level.

Q5, brand mixed in. A campaign that includes brand searches can report a return that is mostly people who would have found you anyway. Fix: a separate brand campaign, and brand exclusions in Performance Max and broad prospecting.

Q6, unchecked automation. Auto-applied recommendations and final URL expansion are convenient and occasionally expensive. Fix: decide for each whether it is on, write down why, and look at its effect monthly.

Q7, stale creative. When broad audiences do the targeting, the ad decides who responds. Fix: a monthly slot for new concepts; see the PPC manager's guide to AI ad creative.

Q8, unreconciled reporting. If Google and Meta together claim more sales than your shop recorded, at least one is over-crediting itself. Fix: a monthly table of spend, platform-claimed results and your own results by channel. The ROAS calculator turns that into break-even and contribution.

What this check does not tell you

It checks set-up, not strategy. An account can score 16 and still sell an offer the market does not want, or aim at the wrong audience. For what is changing next, read the 2026 trends. If you want an outside view, the free written breakdown works from public information and any figures you choose to share.

FAQ

How long does it take to fix a Leaking score?

It depends on your tracking set-up and who controls the website and CRM. The measurement questions often need a developer and whoever owns the CRM, so plan them as a small project rather than a same-day change.

Should I pause campaigns while I fix measurement?

Not necessarily. Reduce spend on campaigns you cannot measure at all, and keep proven ones running at a steady level so you do not lose their learning while tracking is repaired.

Is Performance Max still a source of waste?

It can be, mainly through brand cannibalisation and low-quality placements or leads. The 2025 controls (campaign-level negatives, search term insights and channel reporting) make that visible and fixable, which was harder before.

How often should I repeat this self-check?

Quarterly, and after any major platform change or new campaign type. The platforms add and rename automated features frequently, and new defaults can undo earlier decisions.