How to Find Wasted Ad Spend: A 6-Step Audit Framework for 2026

To find wasted ad spend, audit six things in order: working media rate, invalid traffic, search term relevance, inventory you never chose, frequency decay, and conversion reconciliation. The ANA's Q1 2026 benchmark found only 43.3% of programmatic dollars reach a fraud-free, viewable, measurable impression — the other 56.7% is where the waste hides.

Most advertisers audit their media the way people audit their gym membership: once a year, guiltily, after the money is gone.

That was survivable when media was simple. It is not survivable now. Automation decides more of your delivery every quarter, and the places it sends your money are the places you have the least visibility into. Waste no longer looks like a bad keyword you can spot in a report. It looks like a campaign that appears to be performing while a third of the budget never reaches a human.

This is a six-step audit you can run in an afternoon with reports you already have access to. Each step comes with a benchmark, so you are not guessing whether your number is bad — you are comparing it against published 2026 data.

Before you start: pull these five reports

Do not start clicking around. Export these first, for the last 90 days, and put them in one place:

1. Spend by channel, campaign and placement — every platform, one sheet
2. The Google Ads search terms report, filtered to the last 30 days
3. Your placement or domain report from each DSP and from Performance Max
4. Frequency and reach by campaign, wherever your platform reports it
5. The conversion counts your business actually recorded — from your CRM or order system, not from the ad platform

That fifth one is the report most teams skip, and it is the one that makes the other four honest.

Step 1: Measure your true working media rate

Working media rate is the share of every dollar that becomes an impression a real person could have seen. It is the number that governs everything downstream, and almost nobody calculates it.

The ANA's Q1 2026 Programmatic Transparency Benchmark, published on 27 May 2026 with TAG TrustNet and Fiducia, gives you the yardstick. Its market-level TrueAdSpend Index — the share of programmatic investment delivering fraud-free, measurable, viewable, MFA-free impressions — sat at 43.3% in Q1 2026.

The spread underneath that average is the actionable part. Higher-performing advertisers converted 54.0% of programmatic spend into qualified impressions. Lower-performing advertisers converted 32.1%. That 21.9-point gap is the widest the benchmark has recorded.

Here is the finding that should reorganise your priorities. The gap between the two cohorts was driven almost entirely by media quality, not by fees. Transaction costs differed by just 2.4 percentage points between cohorts. Media productivity losses differed by 19.4 points. The top cohort lost 19.0% of spend to quality issues; the bottom cohort lost 38.4%.

Translate that into price and the case closes itself. Adjusted for quality, the higher-performing cohort paid a TrueCPM of $7.46 per thousand qualified impressions against $19.04 for the lower-performing cohort. A $1.95 difference in raw CPM became an $11.58 difference in real cost.

So if you have spent this year negotiating your agency fee down by a point, you optimised the wrong variable. Compare your raw CPMs against our 2026 programmatic cost benchmarks for CPM, CPC and CPV, then divide by your viewable, measurable, fraud-free rate to get the number that actually matters.

Step 2: Quantify invalid traffic, platform by platform

Invalid traffic is the cleanest waste to find and the least often measured. A Lunio analysis of 2.7 billion clicks across six platforms, eight industries and ten countries — reported by MediaPost in January 2026 — put the global invalid traffic rate at 8.51%, or roughly one click in twelve, amounting to $63 billion in wasted global ad spend.

The averages conceal enormous platform variance:

By sector, gaming topped the table at 18.49%, followed by education at 14.41%, telecoms and utilities at 14.26%, and real estate at 13.61%. Financial services and insurance came in at 10.12%; retail sat lowest at 6.03%. Lead-generation businesses experienced IVT rates 32.07% higher than ecommerce brands — which distorts funnel maths precisely where most B2B advertisers are least equipped to notice.

Retail's 6.03% looks benign until you set it against retail margins of 5% to 8%. At those margins, a 6% invalid traffic rate eats a meaningful share of profit on every transaction.

Almost nobody is measuring this. A Lunio survey of 131 marketers reported in July 2026 found that 75.6% believe they are losing budget to bots, while only 5.3% run a dedicated invalid traffic platform. Half said their bidding now optimises toward automated traffic — which is the genuinely expensive version of the problem, because Smart Bidding learns from whatever converts in the data it is given.

Step 3: Audit search terms and match type allocation

Open the search terms report. Sort by cost descending. Read the top 50 queries and mark each one: would you have bid on this deliberately?

The waste here is structural, not accidental. Optmyzr analysed 30,000 Google Ads accounts for February 2026 and found exact match has lost roughly 9.5 percentage points of spend share since 2022, with broad match now the dominant match type by budget. Across non-branded campaigns, exact match still led on efficiency across the board.

Three findings from that study are worth acting on directly:

• Broad match is a bet on the algorithm, and the conditions matter. It works when Smart Bidding has rich signals — conversion value in ecommerce, strong conversion volume in lead gen. Strip those conditions away and the efficiency gap becomes hard to ignore.
• Phrase match is the most consistent overperformer. It carries the largest share of non-brand spend and punches above its weight on conversions. In ecommerce accounts it posted the highest conversion rate of any match type in the entire study.
• Branded broad match deserves more scrutiny than it gets. On brand terms, exact match led on CTR, conversion share and ROAS, while broad match produced more spend and fewer conversions. Phrase match converted 36% better than broad on brand terms.

Broad match is not the enemy. Broad match running without a maintained negative keyword list and without conversion signal is. If you want the full account-level version of this pass, our 10-step Google Ads account audit checklist covers the settings this step assumes you have already got right.

Step 4: Cut the inventory you never chose

Every automated campaign type ships with inventory switched on by default. You did not select it. You are paying for it.

Performance Max is the clearest case. Until recently, the Search Partner Network and the Google Display Network were compulsory. In July 2026 Google began a limited alpha letting selected advertisers deselect both — two checkboxes, enabled by default, that can now be unticked. Industry research cited by PPC Land put Search Partner Network return on ad spend 37% below Google Search proper.

Access to that alpha is being rationed rather than announced; several agencies reported having requests refused. Check your own campaign settings, because some buyers found the checkboxes appear with no notification at all. Separately, account-level placement exclusions arrived in January 2026 as a single unified block list spanning Performance Max, Demand Gen, YouTube and Display — use it.

Made-for-advertising inventory is the same problem in programmatic. The ANA put MFA exposure at 1.1% in Q1 2026, up from a range of 0.4% to 0.6% across 2025, and flagged AI slop as an emerging subtype needing ongoing mitigation. Integral Ad Science research published in July 2026 found MFA traffic rose 5% over Christmas 2025 while MFA impressions climbed 219% between 2 November and 5 December — the divergence widens exactly when your budgets peak.

The structural fix is supply concentration. The ANA found higher-performing advertisers operated with significantly more concentrated supply footprints, held a 13.3-point advantage in measurable inventory and a 6.7-point advantage in viewability, and converted more spend into working media while paying lower average CPMs. Fewer, better paths. Our comparison of open exchange versus private marketplace performance benchmarks shows where that concentration pays off and where it does not.

If you have run steps one through four and the numbers came back worse than the benchmarks above, that is not a reason to stop — it is the point at which an outside pass is cheapest. Our free paid media audit runs this same six-step framework across your accounts and returns the recoverable spend figure in writing.

Step 5: Check frequency and creative decay

Frequency waste is the most expensive line item nobody reports on, because over-delivery looks identical to delivery in a dashboard. Impressions went up. Reach did not.

Run this check per campaign: plot reach against frequency over a fixed budget. Every impression delivered above your effective cap is money spent re-reaching a household that had already decided. In connected TV especially, completion rates stay high because the format is non-skippable, which means the metric most buyers watch cannot tell you when the ad stopped working. We covered the workable ceiling in our 2026 guide to CTV frequency capping.

Creative decay compounds it. Pull CTR and conversion rate by creative by week since launch, and find the week where the curve bends. That week is your rotation interval. Most teams discover it is considerably shorter than their production schedule, which is a planning problem rather than a media problem — and worth solving in creative strategy rather than by buying more impressions against tired assets.

Step 6: Reconcile platform conversions against the business record

This is the step that catches the waste the other five miss, and it takes ten minutes.

Put the conversion count each ad platform reports next to the count your CRM or order system recorded for the same period. Then explain the difference. Sum your platforms and you will usually exceed the business number, sometimes by a wide margin, because platforms claim overlapping credit under their own attribution windows.

Two failure modes hide here, and they need opposite fixes. Either real conversions are firing and not reaching the team who should act on them — a routing failure that is costing you revenue, not just measurement accuracy — or the platform is counting interactions that were never completed conversions, which means your bidding has been optimising toward a phantom. Automated form and click tracking is a common culprit for the second.

Until you can explain the gap, every efficiency number above it is decorative. Our breakdown of last-click, data-driven and media mix modelling attribution covers how to choose a model once the underlying counts are trustworthy, and average ROAS benchmarks by industry and channel gives you something credible to measure the reconciled figure against.

The six-step waste audit scorecard

Screenshot this. Run it quarterly.

One caveat on the scorecard. These benchmarks describe different universes — the ANA figures cover programmatic display and CTV, Lunio's cover paid click traffic, Optmyzr's cover Google Search. Do not average them into a single site-wide waste percentage. Use each one against the channel it was measured in.

Frequently asked questions

How much of a typical ad budget is actually wasted?

It depends entirely on the channel, which is why a single headline figure is misleading. In programmatic, the ANA found higher-performing advertisers lost 19.0% of spend to media quality issues while lower-performing advertisers lost 38.4%. On paid click traffic, Lunio measured an 8.51% invalid traffic rate globally. Audit each channel against its own benchmark rather than applying one number across the account.

How often should I run a media waste audit?

Quarterly for the full six steps, monthly for steps 2, 3 and 6. Invalid traffic rates, search term mixes and conversion reconciliation all move fast enough that a quarterly cadence lets a problem run for eleven weeks before you see it. Steps 1 and 4 are structural and change more slowly.

Does broad match waste money?

Not inherently. Optmyzr's analysis of 30,000 accounts found broad match works when Smart Bidding has strong signals to learn from — conversion value in ecommerce, conversion volume in lead gen. Broad match running without a maintained negative keyword list, or without enough conversion data to guide bidding, is where the waste appears. The accounts getting the most from match types use all three deliberately.

Can I exclude Search Partners and Display from Performance Max?

Only if you have alpha access. Google began testing two campaign-level checkboxes in mid-2026 that let selected advertisers deselect the Search Partner Network and the Google Display Network, both enabled by default. It remains in limited alpha with no confirmed general release date, and several agencies have had requests declined. Check your campaign settings directly — some buyers found the option had appeared with no notification.

Is invalid traffic the same as click fraud?

Not quite. Invalid traffic covers everything that is not a real user with genuine purchase intent, including duplicate clicks, bot patterns and misleading publisher behaviour. Some is deliberate fraud; much of it is not. AI-powered and agentic bots are an increasingly difficult category, since they navigate sites and mimic human browsing well enough to pass most naive filters.

Methodology and sources

Figures in this playbook come from four sources retrieved in September 2026. The ANA Q1 2026 Programmatic Transparency Benchmark, published 27 May 2026 in partnership with TAG TrustNet and Fiducia, covering 86 participating marketers. Lunio's invalid traffic analysis of 2.7 billion clicks recorded between 1 August 2024 and 31 July 2025 across six platforms, eight industries and ten countries, reported by MediaPost in January 2026, plus a Lunio survey of 131 marketers reported in July 2026. Optmyzr's match type study covering 30,000 Google Ads Search accounts for February 2026. PPC Land's July 2026 reporting on Performance Max network controls, including Integral Ad Science research on made-for-advertising traffic.

Start with one step this week

Six steps is a quarterly exercise, not a Wednesday afternoon. If you only run one, run step six — reconcile what the platforms claim against what your business recorded. It is the cheapest, and it tells you whether the other five are measuring anything real.

If you would rather have someone else do the pass: book a 30-minute intro call to talk through what your reports are showing, or request a free paid media audit and we will run all six steps across your accounts and send back the recoverable number. We do this across programmatic and paid search for Canadian and US advertisers every week — and you can see who you would actually be working with before you book anything.