Advertising Exchanges: What Marketers Get Wrong About Programmatic Supply

An advertising exchange is a technology platform where publishers sell ad inventory and advertisers buy it, in real time, through automated auctions. The price is set by demand, the transaction happens in milliseconds, and the whole system operates largely without human intervention. That is the clean version. The reality is more complicated, and the gap between the clean version and the reality is where most marketing budgets quietly leak.

Understanding how exchanges actually work, where they sit in the broader programmatic stack, and what they mean for your go-to-market economics is not a technical exercise. It is a commercial one. The decisions you make about supply path matter as much as the decisions you make about creative or targeting.

What Is an Advertising Exchange and How Does It Fit in the Stack?

The programmatic ecosystem has a habit of generating acronyms faster than most people can absorb them. DSP, SSP, DMP, PMP, SPO. Strip it back and the structure is straightforward. Publishers connect to supply-side platforms (SSPs) to make their inventory available. Advertisers connect to demand-side platforms (DSPs) to bid on that inventory. The exchange is the marketplace in the middle where those bids are matched to impressions.

In practice, the lines between SSPs and exchanges have blurred significantly. Google’s Ad Exchange (now part of Google Ad Manager), OpenX, Magnite, PubMatic, and Xandr all function as both the sell-side infrastructure and the auction venue. When you buy programmatically through a DSP, your bid is typically submitted to multiple exchanges simultaneously. The exchange that wins the auction charges the advertiser and pays the publisher, taking a percentage margin in between.

That margin is one of the first things marketers should understand. It is not always disclosed clearly, and it varies by exchange, deal type, and the specific path your budget takes through the system. The BCG work on commercial transformation makes a point that applies here: commercial clarity is a prerequisite for commercial growth. If you do not know what you are paying for each layer of the stack, you cannot optimise it.

Open Exchanges vs. Private Marketplaces: The Trade-Off That Actually Matters

Open exchanges are exactly what they sound like. Any buyer with a DSP seat can bid on any inventory that publishers have made available. The scale is enormous. The quality is variable. And the fraud exposure is real.

I spent several years managing programmatic strategy across a portfolio of clients that collectively represented substantial display and video spend. The open exchange numbers always looked good in the dashboard. CPMs were low, click-through rates were reported, view-through conversions were attributed. The problem was that when we started pulling on threads, a meaningful portion of that activity was not doing what we thought it was doing. Impressions were being served to audiences that bore no resemblance to the intended target. Placements were appearing on domains we would never have consciously chosen. Some of it was outright fraud.

Private marketplaces (PMPs) emerged partly in response to these concerns. A PMP is an invitation-only auction where a publisher offers inventory to a selected group of buyers at a floor price. You get more control over where your ads appear, better data about what you are buying, and generally higher-quality inventory. The trade-off is higher CPMs and reduced scale. Preferred deals take this further: a fixed price agreed directly between buyer and publisher, with the buyer getting first access to inventory before it goes to auction.

Programmatic guaranteed sits at the far end of the spectrum: a direct deal, automated execution, fixed volume, fixed price. It combines the control of a direct buy with the operational efficiency of programmatic. For brand campaigns where placement quality matters, it is often the right call.

The choice between these models is not purely a quality question. It is a go-to-market question. If your strategy depends on reaching new audiences at scale, open exchange with strong exclusion lists and supply path optimisation may be the right tool. If you are running a brand campaign for a regulated category or a premium product, the reputational risk of open exchange placements may outweigh the CPM savings. That calculation belongs in your planning process, not your post-campaign report.

Go-to-market decisions like these, where channel economics intersect with brand risk and audience strategy, are exactly the kind of thinking covered in more depth across the Go-To-Market and Growth Strategy hub.

Supply Path Optimisation: Why the Route to the Impression Matters

For a long time, programmatic buying operated on a simple assumption: the DSP finds the best impression at the best price, and the advertiser benefits. What that assumption missed is that the same impression can be available through multiple paths, each with different costs, different intermediaries, and different levels of transparency.

Supply path optimisation (SPO) is the practice of identifying the most efficient and transparent route from your DSP to the publisher. It involves reducing the number of intermediaries, preferring exchanges where the publisher has a direct relationship, and eliminating resellers who add cost without adding value. ads.txt and sellers.json were introduced specifically to make this more tractable, by creating public records of authorised sellers for each domain.

The practical implication for marketers is that SPO is not just a trading desk concern. It affects your working media ratio, which is the percentage of your budget that actually reaches an audience versus the percentage consumed by the stack. In a poorly optimised programmatic setup, that ratio can be surprisingly low. In a well-optimised one, you get meaningfully more reach or more quality for the same budget.

When I was building out the programmatic capability at iProspect, one of the first exercises we ran was a supply path audit for a major retail client. The findings were instructive. A significant share of spend was flowing through resellers who had no direct relationship with the publishers the client thought they were buying. Cutting those paths and consolidating to direct exchange relationships improved the effective CPM and the quality of placements simultaneously. The budget did not change. The output did.

Ad Fraud, Brand Safety, and Viewability: The Problems That Do Not Go Away

These three issues have been discussed in the industry for over a decade. They are still live problems. The reason they persist is not that the industry lacks solutions. It is that the incentive structures within the ecosystem do not always align with the advertiser’s interest.

Ad fraud takes many forms: bot traffic that generates fake impressions, domain spoofing where low-quality inventory is misrepresented as premium, ad stacking where multiple ads are layered in a single placement with only the top one visible, and pixel stuffing where ads are served in a one-by-one pixel frame. Verification vendors like IAS and DoubleVerify provide measurement and blocking tools, but they are not a complete solution. Sophisticated fraud adapts.

Brand safety is a related but distinct issue. Your ad appearing next to content that conflicts with your brand values is not fraud, but it is a real commercial and reputational risk. The automated nature of exchange buying means that without strong exclusion lists, keyword blocklists, and content category filters, your creative will end up in places you would not choose. This is not a hypothetical. Major brands have faced genuine reputational damage from programmatic placements on extremist content or misinformation sites.

Viewability is the question of whether an ad was actually visible to a human. The MRC standard for display is that 50% of the ad’s pixels are in view for at least one second. That is a low bar. An ad that meets the viewability standard is not necessarily an ad that was seen, processed, or remembered. When I judged the Effie Awards, one of the consistent observations among the panel was that effectiveness entries which relied heavily on programmatic display metrics rarely demonstrated the kind of attention or memory encoding that drives actual business outcomes. Viewability is a floor, not a ceiling.

None of this means programmatic display is ineffective. It means the metrics it generates need to be read with appropriate scepticism. The Forrester work on intelligent growth models has long argued that measurement frameworks need to account for what is actually driving outcomes, not just what is easiest to count. That applies directly here.

The Attribution Problem Inside the Programmatic Stack

Earlier in my career, I placed too much weight on lower-funnel performance metrics. The numbers looked compelling. CPAs were low, return on ad spend was high, and the platform reporting made it easy to feel like the system was working. The problem I came to understand over time is that a meaningful portion of what performance channels claim credit for was going to happen anyway. Someone who has already decided to buy something and types your brand name into a search engine is not being converted by the ad they see. They are being intercepted by it.

The programmatic stack has a version of this problem that is specific to its architecture. When a DSP reports that an impression contributed to a conversion via view-through attribution, it is making a claim that is very difficult to verify independently. The exchange has an interest in that attribution being generous. The DSP has an interest in it being generous. The verification vendor is measuring what it can measure, not what it cannot. The result is a measurement environment that systematically flatters the stack.

This does not mean exchange-bought media is not working. It means the reported numbers are not a reliable guide to how much it is working. Incrementality testing, holdout groups, and media mix modelling give you a more honest picture. They are harder to run and less satisfying to report, but they are closer to the truth. Tools like behavioural analytics platforms can add a qualitative layer to what the programmatic stack tells you about audience engagement, helping you triangulate between what the numbers say and what users are actually doing.

The Vidyard Future Revenue Report makes a related point about pipeline attribution in B2B contexts: the channels that look most efficient in last-touch models are often not the ones generating the initial demand. The same logic applies to programmatic display. The exchange that claims the conversion may not be the medium that created the intent.

How Exchanges Fit Into a Growth-Oriented Media Strategy

The commercial case for advertising exchanges is real. They provide access to scale that direct buying cannot match, at price points that make prospecting economics work. For brands that need to reach new audiences efficiently, the open exchange, properly managed, is a legitimate tool.

The mistake is treating exchange-bought media as a performance channel and measuring it like one. Display impressions served through an exchange are, in most cases, interruption advertising. They work the way all interruption advertising works: through repeated exposure, through contextual relevance, through creative quality, and through the cumulative effect of being present in the right environments over time. Expecting them to generate direct, attributable conversions at scale is asking the wrong question.

A better frame is reach quality. How many of the people you reached were genuinely in your target audience? How many of those impressions were served in environments that reinforce your brand positioning rather than undermine it? How much of your budget actually reached a human being in a visible placement? These questions are harder to answer than “what was the CPA?” but they are more useful for making strategic decisions.

There is also a growth dimension that is easy to miss. Advertising exchanges, at their best, are a mechanism for reaching people who do not yet know they want what you offer. That is a fundamentally different job from capturing existing intent through search. The brands I have seen grow most consistently over time are the ones that invest in both, not the ones that optimise relentlessly for the bottom of the funnel and wonder why their growth plateaus. The clothes shop analogy holds here: someone who has already tried on the jacket is far more likely to buy it than someone who has never walked through the door. Exchange media, when it works, gets people through the door.

BCG’s analysis of evolving go-to-market strategy in financial services makes a point that generalises well: growth requires reaching audiences at earlier stages of their decision-making, not just converting the ones who have already decided. Exchange media, properly planned and honestly measured, is one of the tools that can do that job.

If you are thinking through how exchange-bought media fits into a broader growth architecture, the Go-To-Market and Growth Strategy hub covers the strategic frameworks that sit around these channel-level decisions, from audience planning to commercial measurement.

What Marketers Should Actually Do With Advertising Exchanges

A few principles that have held up across the accounts I have managed and the campaigns I have overseen.

Know your supply path. If you cannot tell me which exchanges your DSP is buying through, what the fee structure looks like at each layer, and which publishers are actually delivering your impressions, you do not have enough visibility to optimise. Request a supply path analysis from your trading desk or agency. It is a reasonable ask and a revealing exercise.

Use private marketplaces for brand-sensitive campaigns. The CPM premium is real but so is the risk mitigation. For campaigns where placement context matters to brand perception, the economics of a PMP are usually justified.

Set viewability and fraud thresholds that mean something. The default settings in most DSPs are not aggressive enough. Work with your verification vendor to understand what thresholds are achievable in your categories and set targets accordingly. Accept that higher thresholds will reduce scale. That is the point.

Test incrementality, not just attribution. Run holdout tests. Compare conversion rates between exposed and unexposed groups. The results will often be humbling, but they will be honest. Humbling and honest is more useful than flattering and misleading.

Treat exchange media as a reach tool, not a conversion tool. Set objectives accordingly. Measure against reach quality metrics: on-target percentage, viewability, brand-safe placements, frequency distribution. Use separate channels and separate measurement frameworks for conversion activity.

When I took over a loss-making agency and started rebuilding its programmatic offering, one of the first things I changed was how we reported exchange performance to clients. We stopped leading with CPAs and started leading with reach quality. Some clients pushed back. The ones who stayed with us long enough to see the results understood why it mattered. The ones who left for agencies promising lower CPAs mostly came back later, after the numbers stopped adding up.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions