Of every dollar that enters a programmatic DSP, only 43.9 cents reaches a real person as a viewable impression on legitimate inventory. The rest disappears into intermediary fees, made-for-advertising sites, invalid traffic, and layers of ad tech that add cost while removing accountability. That's not a fringe number from a critic. That's from the ANA's own 2025 Programmatic Transparency Benchmark. The industry spends $786 billion a year on digital advertising, and a significant chunk of it is burning for reasons most advertisers never see reported anywhere near their dashboards.
We're going to cover the structural problems that make digital advertising so prone to waste and misaligned incentives. Not as a theoretical exercise, but because we've seen these patterns in accounts we've inherited, and because understanding them is the only way to run campaigns where the spending is actually working.
The incentive problem that never gets fixed
Most digital advertising agencies charge fees as a percentage of what you spend. Typical rates sit between 10% and 20% of managed media. That model sounds reasonable until you think about what it actually incentivizes.
An agency earning 15% on a $400,000 annual budget takes home $60,000. If they talk you into increasing that budget to $600,000, their take goes to $90,000. They earned 50% more without improving a single campaign. Meanwhile, if they found a way to cut wasteful spend and deliver the same outcomes on $300,000, they'd make $45,000 less than they started with.
The math tells the story. Under percentage-of-spend pricing, every conversation about efficiency is a conversation the agency is financially disincentivized to have. We've written about this in detail in our post on the percentage-of-spend agency model and what advertisers should ask instead.
This isn't a character flaw. It's a structural problem. Good people operating under a broken incentive model will still drift toward the behavior the model rewards. The fix isn't finding a more ethical agency. It's changing the fee structure.
What "we're scaling your campaigns" often means. When an agency tells you they want to scale your campaigns, ask what specifically they're proposing to scale. Increasing budget without improving conversion rate or reducing cost per acquisition isn't scaling. It's spending more. The two things sound similar but have very different effects on your business.
What ad ops actually is and why it's a black box
Ad operations (ad ops) is the machinery that sits behind every digital advertising campaign. It covers how creatives get trafficked into platforms, how targeting parameters are set and updated, how bids get adjusted, how pacing works, how reporting pulls together, and how discrepancies between platforms get investigated and resolved.
At most agencies, ad ops is treated as a back-room function. You talk to an account manager. The account manager talks to an ad ops team. The ad ops team makes changes in the platforms. You get a report that summarizes what happened. What you almost never see is the actual raw data. Which placements actually ran. What the viewability was at each placement. What the actual CPMs paid were before platform fees. Which audience segments got the most spend and why.
That invisibility is where a lot of problems get introduced. Not necessarily through intentional misdirection, though that happens. More often through complexity that compounds over time. Someone sets up a campaign with 40 placement exclusions and then someone else adds a budget reallocation and six months later nobody on the team can fully explain why one placement is consuming 60% of the impression volume. The complexity becomes its own form of accountability avoidance.
Red flags in your current setup- Agency can't provide a domain-level placement report showing where programmatic impressions ran
- Budget increases are proposed without specifics on which campaigns or placements they'll go to
- Monthly reports show only impressions, clicks, and CTR without conversion data at the campaign or ad group level
- Agency handles all platform access directly and doesn't add client as an admin on their own accounts
- Viewability metrics are absent or lumped in without per-placement breakdowns
- Campaign recommendations come without the data that supports them
- Changes to bid strategy, audiences, or creative rotation are made without notification
We've covered the pattern of agencies using information asymmetry as a retention mechanism in our post on why most agencies hide their metrics. If an agency won't give you admin access to your own accounts, that's not a policy. It's a dependency they're building.
The programmatic supply chain problem
Programmatic advertising runs through multiple layers of technology between an advertiser and a publisher. A demand-side platform (DSP) connects to supply-side platforms (SSPs) through real-time auctions. Data management platforms add audience targeting signals. Verification platforms check for fraud and viewability. Ad servers track and report delivery. Each layer charges fees. Most of those fees are never disclosed as a line item.
The ANA's K2 Intelligence Report documented what happens when agencies take a principal position on media buys. Instead of acting as an agent purchasing on your behalf, the agency buys inventory wholesale and resells it to you at a markup. K2 found that markups on principal transactions range from 30% to 90%, and some ad server markups were running at 200% to 250% above the actual cost, with none of it disclosed in client contracts or reporting.
That's not ancient history. The incentive structure that made those arrangements attractive to agencies hasn't changed. What's changed is that more advertisers know to ask about it.
Made-for-advertising sites
Made-for-advertising (MFA) sites are web properties built to generate programmatic ad revenue rather than to serve an actual audience. They publish thin content, structure pages to maximize ad slot density, and pump their inventory into the open exchange at low CPMs. Without supply path optimization and active exclusion lists, your campaigns end up running there.
The problem is that MFA inventory often passes basic brand safety checks. It's not flagged as explicit content. It doesn't trigger category exclusions. It just quietly absorbs impression volume while delivering no meaningful audience value. We've seen accounts where 40% or more of programmatic impressions were running on MFA domains, and the agency's report showed healthy CTRs because they were only reporting on performance, not on quality.
This is exactly why transparency at the placement level matters and why we emphasize it in how we run programmatic campaigns.
How metrics get used to obscure what's actually happening
Digital advertising produces a lot of numbers. Click-through rates. Impressions served. Cost per click. Reach and frequency. View-through conversions. Engagement rates. The availability of metrics makes it easy to build a report that looks like evidence of performance without the report actually showing whether the business is benefiting.
| Metric | What it shows | What it doesn't show |
|---|---|---|
| CTR | Ratio of clicks to impressions | Whether clicks converted or had any business value |
| Impressions served | Volume of ad delivery | Viewability, human traffic rate, audience quality |
| Cost per click | Average cost of a click | Quality of the traffic or intent of the user |
| View-through conversions | Conversions attributed to ad exposure | Whether the ad actually influenced the conversion |
| ROAS | Revenue divided by ad spend | Margin, incrementality, or whether revenue would have happened anyway |
| Brand lift survey | Self-reported awareness change in a sample | Whether awareness change translates to purchase behavior |
The pattern we see most often is reports that lead with volume metrics and bury conversion data. Impressions went up, clicks went up, cost per click went down. Somewhere on page four, cost per acquisition went up significantly. The leading metrics are selected because they're moving in the right direction. The lagging metric that actually matters is de-emphasized.
We've seen accounts where agencies were reporting "strong performance" on CPM and CTR while the actual cost per acquired customer had tripled over 18 months because the audience targeting had gradually drifted toward cheap but unqualified traffic. Nobody flagged it because nobody was required to flag it.
The attribution shell game
When a brand runs Google Ads, Meta ads, and programmatic simultaneously, attribution becomes genuinely hard. Someone might see a programmatic display ad on Monday, click a paid social ad on Thursday, search for the brand name on Friday, and convert through the branded search campaign. Google claims full credit because it was last click. Meta claims it because the click happened. The DSP claims a view-through because the display impression ran before the conversion.
All three numbers show up in all three dashboards. If you add them up, total attributed revenue often exceeds actual revenue by 3x or more. This isn't fraud. It's how attribution math works when every channel applies its own model to its own data. But it creates a situation where every channel looks profitable in its own report and nobody has to answer for whether the combined spend is actually justified.
The fix isn't finding the right attribution model. It's building measurement that connects ad spend to actual business outcomes at the level that matters. For e-commerce that's usually ROAS at a margin threshold. For lead gen it's pipeline quality and close rates, not just lead volume. That kind of measurement is harder to build and harder to report on, which is why most agencies don't lead with it. Read our overview of how programmatic advertising actually works to understand where the measurement challenges start in the supply chain.
What advertiser-side accountability actually looks like
It starts with access. You should have admin-level access to every platform running your campaigns. Not view-only access. Full access, regardless of whether you plan to make changes yourself. Any agency that won't provide this is creating a dependency, not a service.
It continues with the right questions. Not "how did performance look this month" but "what is the cost per acquired customer at margin, and how has that trended over the last six months." Not "what did CTR do" but "which campaigns are profitable and which aren't."
What a transparent relationship actually looks like
A flat-fee agency has no financial incentive to recommend spending more. An agency that discloses every fee layer in the programmatic stack gives you the information you need to make real decisions. An agency that shows you placement-level data lets you see what you're actually buying.
None of these things are unusual asks. They're the baseline of a relationship that's actually aligned with your outcomes rather than your budget size.
It ends with structure. The three things above aren't premium add-ons. They're what a working relationship looks like when the agency's incentives aren't pointed against you.
What to demand from any agency- Admin-level access to every platform your budget runs through, before you sign anything
- Placement-level reporting on every programmatic campaign, not just aggregate metrics
- A full fee breakdown separating working media, DSP fees, SSP fees, data, and verification costs
- Confirmation in writing that the agency does not take principal positions on media buys
- Cost per acquired customer trending over time, not just campaign-level CTR and CPM
- Written notice before any bid strategy, audience, or creative rotation change is made