Most marketing waste is not a spending decision anyone made. It is spend that no report in the building is structured to show: campaigns still live with no budget moving, line items nobody can attribute to a channel, naming conventions that quietly split one campaign into four. The money is not stolen, it is unaccounted for, and the reason it survives quarter after quarter is that the reports built to catch it are assembled from the same broken taxonomy that hides it. This is where the leaks actually are, why the standard reporting stack cannot surface them, and how to find yours in about a week.

This expands on a post I published on LinkedIn about what actually closed our deals last quarter.

Key Takeaways

  • The leak is measured, not hypothetical. The ANA's log-level study of 21 major advertisers found that only 36 cents of every dollar entering a demand-side platform reaches the consumer, which it frames as a $22 billion efficiency opportunity for marketers.
  • Budgets are flat, so the waste is the growth budget. Gartner found 59% of CMOs report insufficient budget to execute their strategy, with marketing budgets flatlined at 7.7% of company revenue.
  • Four leaks account for most of it: unattributable spend, zero-delivery campaigns still running, taxonomy drift that fragments one campaign into several, and duplicate spend across overlapping audiences.
  • Your dashboards cannot show you this. A report groups by the naming field, so a broken naming field produces a clean-looking report of the wrong thing. The error is invisible precisely because the report renders successfully.
  • Automated reporting is the floor now, not the edge. Getting analyst hours back is table stakes; the thing that changes decisions is the stack finding money you did not know you were losing.

What Changed My Mind About Why People Buy

I used to think we won deals because of the demo. Last quarter humbled me.

Looking back across our recent wins, the pattern was consistent and it was not the pitch. Every one of them started by killing manual reporting, which is work we have done for years and take real pride in. Analysts get their capacity back, the Monday morning scramble ends, and everyone is happy. But I no longer believe that is why anyone signs. It is the floor now, not the edge. Every serious vendor in this category can automate a report.

The turn always came later, inside the trial, and it always came from the customer's own data rather than from anything I said. The system would read their accounts and point at spend nobody in the room could attribute. Campaigns with zero delivery still sitting in an active state. A naming convention that had drifted so far that the same campaign appeared under four different labels across three platforms. Budget leaking in plain sight, in an account somebody owned and reviewed monthly.

The room goes quiet for a second, and then the conversation stops being about features. Nobody argues with their own numbers. That is the whole mechanism, and it is worth being honest that it has nothing to do with how well I present.

The Leak Is Measured, Not Hypothetical

It would be easy to treat this as anecdote. It is not, and the best evidence is not a survey asking marketers to estimate their own waste, because self-reported estimates measure how much waste people are aware of, which is precisely the thing in question here.

The stronger evidence comes from log-level data. The ANA's Programmatic Media Supply Chain Transparency Study, issued in December 2023, examined the actual transaction logs of 21 major advertisers whose campaigns ran across an average of 44,000 websites. Its central finding is that only 36 cents of every dollar entering a demand-side platform effectively reaches the consumer. The ANA sizes the gap at $22 billion in efficiency available to marketers, and its follow-up work on direct contracts concluded that following best practices could raise that 36 cents to 50 cents or more. Since early 2024 the ANA has run this as an ongoing quarterly benchmark rather than a one-off study, which is the part I would pay attention to: this is a maintained measurement, not a headline from one bad year.

Two caveats I want to state plainly, because the number gets stretched. This measures the programmatic supply chain specifically, not your entire marketing budget, and a good deal of that 64-cent gap is ad tech margin rather than pure waste. It is not a claim that two thirds of your marketing spend evaporates. It is a claim that the distance between "dollars spent" and "dollars working" is large, measurable, and mostly invisible from inside a normal reporting setup.

The reason it matters more this year than last is arithmetic. Gartner's 2025 CMO Spend Survey, conducted in February and March 2025 among 402 CMOs across North America, the UK, and Europe, found marketing budgets flat at 7.7% of company revenue for the second consecutive year, and 59% of CMOs reporting they have insufficient budget to execute their strategy. When nobody is getting a raise, recovered waste is the only new money on the table.

Where the Money Actually Leaks

In trials, the same four patterns surface again and again. None of them is exotic and none of them requires anyone to have been careless.

Spend nobody can attribute. A line item exists, money left the account, and no one can say which channel, campaign, or business unit it belongs to. This is usually a mapping gap rather than a mystery: the platform's own categorization does not survive the trip into the warehouse, so the spend lands in an "other" bucket that everyone has learned to scroll past.

Zero-delivery campaigns still running. Campaigns in an active state, sometimes still drawing budget, that are delivering effectively nothing. A disapproved ad, an audience that shrank below deliverable size, a landing page returning an error, a bid that stopped being competitive months ago. Nothing errors loudly, because from the platform's point of view nothing is wrong. The campaign is simply losing every auction, quietly.

Taxonomy drift. One campaign appears as four because three people named it three ways across two platforms and the fourth variant came from an agency. Every roll-up that groups by campaign name now under-reports each fragment, and the fragments individually look too small to investigate. This is the leak that most reliably survives audits, because each piece looks like a small campaign rather than a quarter of a large one. It is also the cheapest to fix, which is why campaign naming conventions and UTM naming conventions are worth more attention than their glamour suggests.

Duplicate spend across overlapping audiences. Two campaigns, often owned by different teams or a brand and an agency, bidding into substantially the same audience. You pay twice to reach the same person and you frequently pay more, because you are part of your own competition on the auction.

Talk to an Improvado expert about finding unattributable and zero-delivery spend across your ad accounts.

Why Your Current Reports Cannot Show You This

Here is the part that took me longest to understand, and it is structural rather than a failure of diligence.

A report groups by a field. If the field is wrong, the report is not wrong in a way you can see. It renders successfully, the totals foot, the chart looks reasonable, and it silently answers a slightly different question than the one you asked. A dashboard grouping by campaign name cannot tell you that a campaign name is fragmented, because the fragmentation is in the dimension it is grouping on. The instrument and the defect share a component.

The same holds for zero-delivery campaigns. Most dashboards are built to rank things: top campaigns by spend, best performers by ROAS. A campaign spending money and delivering nothing does not rank anywhere. It is not at the top, and it is filtered out of the bottom because the denominator is zero or missing. It is structurally invisible to a report designed around ranking, and it will stay invisible no matter how many times someone reviews that report carefully.

This is why "we already review spend monthly" and "the waste is still there" are both true statements in the same organization. The review is real. It is being conducted through an instrument that cannot register the defect. That is also why the fix is not a better dashboard on the same foundation. It is a governed data model underneath: one mapping table that covers every account and property, campaign identity that does not depend on a hand-typed string, and monitoring that alerts on absence rather than only ranking presence. We wrote up the same principle applied to budget pacing in per-location pacing and alerts, where a healthy network average hides locations that are badly over and badly under.

How to Find Yours in About a Week

You do not need a platform migration to run this. Four checks, in this order, because each one makes the next cheaper.

One, reconcile platform spend against your warehouse and your invoices for one month. Three numbers, one month, one channel to start. The gap between them is your unattributable bucket, and it is usually the largest single finding. If the three do not reconcile, stop and fix that before trusting anything downstream.

Two, list every campaign in an active state with spend above zero and delivery at or near zero. This is a single query and it is the fastest money you will find all quarter. Everything on that list is either a bug to fix or a campaign to pause today.

Three, count distinct campaign names and compare against the number of campaigns you believe you are running. If distinct names materially exceeds the real count, you have taxonomy drift, and every performance report you have read this year has been reading fragments.

Four, check audience overlap between your largest concurrent campaigns. Start with the pairs owned by different teams, which is where the overlap is least likely to have been deliberate.

Do these against one channel and one month first. The point of the narrow scope is that you get a defensible number quickly, and a defensible number is what earns the mandate to fix the taxonomy properly. Ongoing, this becomes continuous ad spend monitoring rather than a quarterly fire drill, and it belongs to a named owner in the same way accountability for AI-generated numbers does.

Improvado's platform is built for exactly this layer: every ad account and analytics property mapped onto one governed data model, with campaign identity, taxonomy enforcement, and spend monitoring applied consistently across channels, so the leaks above show up as findings rather than staying invisible in a report that renders correctly.

Talk to an Improvado expert about putting every ad account on one governed data model.

Nobody Argues With Their Own Numbers

The go-to-market lesson and the operating lesson are the same lesson, which is why I keep coming back to it.

An external benchmark tells you the industry wastes money and invites you to assume you are the exception. Your own account, showing your own campaign, spending your own budget on nothing since March, does not invite anything. It just sits there. That is why the trial closes better than the pitch, and it is also why an internal audit that surfaces four specific campaigns beats a deck about efficiency at getting anything actually fixed.

So the question I would ask your own data, before your next budget conversation: what is it telling you that you have not wanted to hear?

Frequently Asked Questions

How much marketing budget is actually wasted?

For programmatic specifically, the best measured evidence is the ANA's Programmatic Media Supply Chain Transparency Study, which analyzed log-level data from 21 major advertisers and found that only 36 cents of every dollar entering a demand-side platform reaches the consumer, an efficiency gap the ANA sizes at $22 billion. Note the scope: that measures the programmatic supply chain, not your whole budget, and part of the gap is ad tech margin rather than pure waste. Be skeptical of round self-reported figures like "marketers waste a quarter of their budget." Those come from opinion surveys, several of the widely quoted ones are close to a decade old, and they measure awareness of waste rather than waste.

Why does wasted spend not show up in our dashboards?

Because a report groups by a field, and the most common leaks live in that field. A dashboard grouping by campaign name cannot reveal that campaign names are fragmented, since the defect is in the dimension it groups on, and the report still renders cleanly. Separately, most dashboards rank by spend or performance, so a campaign that spends money and delivers nothing ranks nowhere and gets filtered out of both ends. The waste is invisible because the instrument shares a component with the defect, not because nobody is looking.

What is a zero-delivery campaign and how does it keep spending?

It is a campaign in an active state that produces effectively no impressions or clicks, typically from a disapproved ad, an audience that shrank below deliverable size, a broken landing page, or a bid that stopped being competitive. It usually draws little or no budget once delivery collapses, which is exactly why it survives: the loss is the opportunity cost of budget allocated but not working, plus the flywheel of a campaign everyone believes is running. Nothing errors, because from the platform's perspective the campaign is simply losing auctions.

What is the fastest way to find wasted spend this week?

Run four checks against one channel and one month. Reconcile platform spend against your warehouse and your invoices, and treat the gap as your unattributable bucket. List every active campaign with spend above zero and delivery at or near zero. Count distinct campaign names against the number of campaigns you believe you run, to detect taxonomy drift. Check audience overlap between concurrent campaigns owned by different teams. Keep the scope narrow so you get a defensible number fast, because that number is what earns the mandate for the structural fix.

Is fixing naming conventions really worth the effort?

It is the highest-leverage unglamorous fix available. Taxonomy drift does not just cost you the fragmented campaign, it corrupts every roll-up built on top of the name field, which means every performance conclusion drawn from those roll-ups is drawn from fragments. It also compounds: each month of drift makes historical comparison harder. Campaign identity should be generated from a mapping table rather than hand-typed, so the convention is enforced at write time instead of audited afterward.

Why does this matter more when budgets are flat?

Because recovered waste becomes the only source of new money. Gartner's 2025 CMO Spend Survey found marketing budgets flat at 7.7% of company revenue for a second year, with 59% of CMOs saying they lack sufficient budget to execute their strategy, down five points from 2024. In a growth year, waste is an annoyance you fix eventually. In a flat year, the budget you recover from unattributable and zero-delivery spend is functionally your increase, and it does not require anyone's approval to claim.