01 Overview
Programmatic buying promised precision, but it also introduced a quiet arithmetic problem. A dollar entered the auction, and by the time an impression rendered on a screen, only part of that dollar had actually bought media. The rest went to data, verification, fees, and the layered plumbing that connects buyers to sellers. The title of this article borrows a familiar image: the shrinking dollar. In programmatic campaigns, the working portion of each dollar can feel closer to a few dozen cents once every intermediary has taken its share.
This is not a scandal so much as a structural feature. Every hop between the advertiser and the publisher can carry a cost, and every added layer of targeting or measurement adds another line item. A campaign strategy for the programmatic age has to start by acknowledging that the number on the invoice is not the number that reached the audience. The metric that matters is the retained share: how much of gross spend actually purchased viewable, brand-safe, on-target impressions.
Understanding that gap changes how you plan. Instead of optimizing only toward a low headline cost per thousand impressions, you optimize toward a healthy retained share and a clear picture of what each layer contributes. The goal is not to eliminate every intermediary, because some add real value. The goal is to know which ones earn their keep and which ones simply thin the dollar.
02 What this metric really measures
The retained share measures the proportion of gross media spend that survives the journey from budget to rendered impression. It is not a single platform metric, and it rarely appears as one clean line in a report. Instead, it is reconstructed from several inputs: the gross spend, the fees charged by each intermediary, the cost of audience data, the cost of verification and brand-safety tools, and the effective cost per thousand impressions actually paid for inventory that rendered.
When people talk about the shrinking dollar, they are usually describing the difference between what an advertiser budgets and what a publisher receives. That difference is not automatically waste. Some of it pays for fraud detection, viewability measurement, and audience segmentation that genuinely improves outcomes. But the metric is honest about one thing: every fee reduces the buying power of the remaining budget. If ten percent goes to data and another slice goes to verification, the media bid is smaller than the headline suggests.
The metric also captures the difference between served and viewable impressions. An impression can be served, counted, and billed while never being seen. If a meaningful share of impressions are not viewable, the effective cost of a seen impression rises well above the nominal cost per thousand. The retained share is therefore a reality check on efficiency, not just a measure of fees.
03 How it moves with your levers
Several levers move the retained share, and most of them trade off against one another. Bid density and auction competition are the first lever. In a crowded auction, clearing prices rise, and the same budget buys fewer impressions. In thinner auctions, prices fall, but the inventory may be lower quality or less relevant. Floor prices set by sellers act as a second lever: a high floor protects publisher yield but can push buyers toward cheaper, less desirable placements.
Viewability and brand-safety filtering are powerful but costly levers. Tightening these filters reduces the pool of eligible inventory and often raises the price of what remains. That can lower the retained share in raw terms while raising the quality of each impression. Audience data is another lever: first-party data tends to be cheaper and more durable than third-party segments, which carry their own fees and can decay in accuracy. Creative weight and load speed matter too, because heavy creatives that fail to render waste spend without producing a viewable impression.
Supply-path optimization is the most direct lever on the number of intermediaries. Reducing redundant hops, consolidating demand-side paths, and negotiating transparent fees can raise the retained share without changing the audience or the creative. The key is to treat each lever as a deliberate trade, not a default setting. Every filter, every data layer, and every intermediary should justify its cost against the quality of the impressions it produces.
04 Reading the number without fooling yourself
The most common misread is treating the headline cost per thousand impressions as the true cost of reaching people. It is not. That figure usually excludes data fees, verification costs, and the share of impressions that were never viewable. When you add those back in, the real cost of a seen impression can be substantially higher. Teams that compare headline costs across campaigns without normalizing for these factors are often comparing apples to oranges.
A second misread is assuming that a higher retained share is always better. It is not, if the cheaper path delivers poor inventory, weak viewability, or audiences that do not convert. A campaign that retains eighty cents of every dollar but reaches the wrong people is worse than one that retains fifty cents and reaches the right ones. The retained share is a diagnostic, not a score. It tells you how much of your budget is working as media, but it does not tell you whether that media is doing its job.
A third misread is ignoring the difference between served and viewable impressions. A report can show a healthy cost per thousand while a large share of those impressions were never seen. Always pair the retained share with viewability and on-target rates. Read together, these numbers reveal whether the dollar is being thinned by fees or by waste. The honest reading is a range, not a single figure, and it should be tracked over time rather than judged in isolation.
05 Benchmarks to sanity-check against
Rough benchmarks are useful as guardrails, not as targets. Across programmatic display and video, the retained share of gross media spend commonly falls somewhere in a broad range, and the specific figure depends heavily on how many intermediaries sit between buyer and seller, how much data and verification are layered in, and how competitive the auctions are. Treat any single number as directional. A campaign with many intermediaries and heavy data costs will naturally retain less than one with a consolidated path and first-party data.
Viewability is a related guardrail. A meaningful share of impressions in open auctions are not viewable, and the effective cost of a seen impression rises accordingly. If your viewability rate is low, the retained share looks better than it is, because you are paying for impressions that never had a chance to be seen. On-target rates work the same way: if a large share of impressions reach the wrong audience, the working dollar is smaller than the fee math suggests.
Use these benchmarks to ask better questions rather than to declare victory or failure. Is the retained share stable over time? Does it improve when you consolidate supply paths? Does it fall when you add a new data layer? Does higher viewability come at a cost that is justified by better outcomes? Benchmarks give you a frame for those questions. They do not replace the discipline of measuring what actually happened in your own campaigns.
06 Scaling what works
Scaling a programmatic campaign changes the arithmetic. As spend grows, you often move from premium, tightly controlled inventory into broader pools. Raw prices may fall, but the retained share can fall too, because broader inventory tends to require more filtering, more verification, and sometimes more intermediaries. The cheap impression at scale is rarely the same impression that performed well in a small test. This is why scaling should be gradual and measured, not a sudden multiple of the original budget.
A useful approach is to scale the path, not just the spend. If a consolidated supply path produced a healthy retained share at a modest budget, expand within that path before adding new ones. If first-party data drove efficiency, invest in expanding that data rather than renting more third-party segments. If viewability held up, keep the filters that protected it. Each of these choices preserves the working share of the dollar as volume grows.
Watch for the point where marginal inventory stops earning its keep. At some scale, additional impressions come from placements that are cheaper but less effective, and the retained share declines without a corresponding gain in outcomes. The discipline is to know your threshold and to stop before the dollar thins past the point of usefulness. Scaling is not about spending more; it is about spending more on the same quality of working media.
07 Frequently asked questions
Is the shrinking dollar a sign of fraud or waste? Not necessarily. Some of the reduction pays for legitimate services such as verification, fraud detection, and audience data. The problem is not that fees exist; it is that they are often invisible and unexamined. When you can see each layer and its cost, you can decide whether it earns its place. The goal is transparency, not the elimination of every intermediary.
Should I always choose the path with the highest retained share? No. A higher retained share is only meaningful if the impressions it buys are viewable, brand-safe, and on target. A cheaper path that reaches the wrong audience is not efficient, even if more of each dollar goes to media. Pair the retained share with quality metrics before making a decision. The best path is the one that balances working media with working outcomes.
How often should I review these numbers? Programmatic auctions change constantly, so a periodic review is more useful than a one-time audit. Look at the retained share, viewability, and on-target rates together, and track them over time. Sudden changes can reveal new intermediaries, shifting auction dynamics, or creative problems. A steady review cadence turns the shrinking dollar from a mystery into a manageable variable, and that is the foundation of a durable campaign strategy.