Ad Tech Ecosystem: What Marketers Are Paying For

The ad tech ecosystem is the network of platforms, tools, and intermediaries that sit between an advertiser’s budget and a publisher’s inventory. It includes demand-side platforms, supply-side platforms, data management platforms, ad exchanges, verification vendors, and a growing layer of identity and measurement infrastructure. Understanding how these pieces connect, and where value is actually created versus extracted, is one of the most commercially important things a senior marketer can do.

Most marketers use parts of the stack without understanding the whole. That gap costs money.

Why Most Marketers Misread the Stack

When I was running a performance-led agency, we had a client who was genuinely proud of their programmatic setup. They had a tier-one DSP, a well-configured DMP, and a verification layer that looked solid on paper. Their reporting showed strong CTRs and a CPM that seemed competitive. What the reporting did not show was that roughly 40 cents of every pound they spent was disappearing into fees, arbitrage margins, and inventory that no human being ever saw.

This is not unusual. It is, in fact, the default state for most programmatic buyers who have not audited their supply chain in the last 12 months.

The ad tech ecosystem is genuinely complex, and that complexity is not accidental. Opacity benefits intermediaries. The more difficult it is to trace a dollar from brief to impression, the easier it is for margin to disappear at each handoff. Understanding the architecture is the first commercial defence a marketer has.

If you want to understand how this fits into a broader commercial marketing approach, the Go-To-Market and Growth Strategy hub covers the strategic context that makes decisions like these land properly rather than sitting in isolation.

How the Programmatic Supply Chain Actually Works

At its simplest, programmatic advertising works like this: an advertiser uses a demand-side platform to bid on ad impressions, which are made available by publishers through supply-side platforms, with an ad exchange sitting in between to facilitate the auction. The whole process happens in milliseconds, which is impressive engineering and also a useful distraction from the question of where the money goes.

Here is what sits inside that chain in practice:

• DSP (Demand-Side Platform): The buying interface. Advertisers or their agencies use DSPs to set targeting parameters, bid strategies, and creative rules. The DSP takes a fee, either as a percentage of media spend or a CPM markup, depending on the commercial arrangement.
• Ad Exchange: The auction mechanism. Connects buyers and sellers in real time. Major exchanges include Google’s Ad Exchange, OpenX, and Magnite. Each exchange has its own auction dynamics, which affect clearing prices and win rates.
• SSP (Supply-Side Platform): The publisher’s tool for making inventory available and maximising yield. SSPs also take a fee, typically from the publisher’s side, though the economics often flow back through clearing prices.
• DMP (Data Management Platform): Stores and segments audience data for targeting. Becoming less central as third-party cookies decline and first-party data becomes the primary signal.
• Verification and Measurement Vendors: Tools like brand safety platforms, viewability measurement, and ad fraud detection. These sit across the chain and add cost, but they also add genuine value if configured properly.

Each of these layers has a margin. The cumulative effect is that a meaningful portion of a programmatic budget never reaches a real human audience. Industry bodies have tracked this for years, and while estimates vary, the structural reality is consistent: the chain is expensive.

Where Value Is Created and Where It Is Extracted

Not every layer of the stack is extractive. Some of it creates genuine value. The problem is that most marketers cannot tell the difference without doing the work.

A DSP that gives you access to quality inventory, accurate audience signals, and transparent reporting is worth its fee. A DSP that re-sells inventory it has bought cheaply at a markup, while calling it “premium programmatic,” is not. Both exist. The difference is not always visible in the interface.

The same logic applies to data. A DMP that helps you activate genuinely useful first-party segments is a legitimate part of the stack. A third-party data provider selling audience segments built on modelled inferences from questionable sources is selling you noise at a premium. I have seen both situations up close, and the reporting often looks identical until you test it properly.

One of the most useful things I learned from managing large programmatic budgets across multiple verticals is that the question to ask is not “is this working?” but “what would happen if we turned it off?” That test is uncomfortable, but it is the only honest one. GTM complexity has increased significantly, and the ad tech stack is one of the places where that complexity creates the most commercial risk.

The Identity Crisis Reshaping the Ecosystem

The deprecation of third-party cookies has been discussed so extensively that it has become background noise for many marketing teams. That is a mistake. The identity question is the most structurally significant shift in digital advertising since the introduction of programmatic buying itself.

Third-party cookies enabled cross-site audience tracking, frequency capping across publishers, and attribution models that could follow a user from first exposure to conversion. All of that is degrading. What replaces it is not yet settled, and the ecosystem is genuinely fragmented in how it is responding.

The main solutions being proposed fall into a few categories:

• First-party data activation: Using your own customer data, email lists, CRM records, and on-site behaviour to build audiences and inform buying. This is the most defensible approach, but it requires investment in data infrastructure that many brands have not made.
• Contextual targeting: Returning to placement-based logic rather than audience-based logic. Matching ads to content rather than to users. Less precise in theory, but increasingly competitive in practice as the quality of contextual signals improves.
• Identity graphs and clean rooms: Collaborative data environments where advertisers and publishers can match audiences without sharing raw personal data. Technically sophisticated and commercially promising, but not yet standardised.
• Platform-native audiences: Relying on the walled gardens, Google, Meta, Amazon, which have their own logged-in user data and are largely insulated from the cookie deprecation problem. Effective, but it increases dependency on platforms that have their own commercial interests.

Brands that have spent the last few years building first-party data assets are in a genuinely stronger position. Those that have not are now paying to solve a problem they could have avoided.

The Walled Gardens and What They Cost You

Google, Meta, and Amazon collectively account for a dominant share of digital ad spend. They are effective, measurable within their own ecosystems, and easy to buy. They are also opaque about how their auctions work, how their algorithms optimise, and what their data actually represents.

Earlier in my career, I was a strong advocate for lower-funnel performance channels. The reporting looked compelling. CPAs were trackable, ROI calculations were clean, and it was easy to build a case for continued investment. Over time, I came to believe that a significant portion of what performance channels claimed credit for was going to happen anyway. Capturing existing intent is not the same as creating demand. The walled gardens are very good at the former and largely indifferent to the latter.

This matters for how you think about the ad tech ecosystem. If your entire strategy runs through Google and Meta, you are not really engaging with the open programmatic ecosystem at all. You are buying into closed systems with their own measurement logic, their own attribution models, and their own commercial interests in showing you results that justify continued spend.

That does not make them wrong. It makes them partial. A mature media strategy uses the walled gardens where they are genuinely efficient and builds outside them where the commercial case supports it. BCG’s work on evolving go-to-market models reflects a similar principle: channel mix decisions should follow commercial logic, not inertia.

Measurement in Ad Tech: Honest Approximation vs. False Precision

One of the things I observed repeatedly when judging the Effie Awards is how differently agencies and brands think about measurement. The entries that impressed me most were not the ones with the most sophisticated attribution models. They were the ones that were honest about what they could and could not prove, and that made their case through business outcomes rather than platform metrics.

Ad tech measurement has a precision problem. The numbers look exact. Impressions, clicks, viewability rates, conversion paths: all of it is reported to multiple decimal places, which creates a false sense of accuracy. The reality is that every number in your ad tech stack is a model output, not a ground truth. The model reflects the assumptions of whoever built it, and those assumptions often serve the platform’s commercial interests.

Last-click attribution, for example, is still widely used despite being demonstrably wrong as a model of how advertising works. It systematically over-credits the final touchpoint and under-credits everything that created the conditions for conversion. It persists because it is easy to implement and because it tends to favour the channels that are easiest to measure, which are often the channels that are also easiest to buy more of.

Multi-touch attribution is better but not neutral. Media mix modelling is more strong but requires data quality and analytical investment that many organisations do not have. Incrementality testing is the most honest approach, but it requires turning things off, which is politically difficult and commercially uncomfortable.

The right approach is honest approximation. Accept that you cannot measure everything precisely. Build a measurement framework that triangulates across multiple signals rather than relying on any single model. And be genuinely suspicious of any number that tells you exactly what you want to hear.

Supply Path Optimisation: The Commercial Case for Caring

Supply path optimisation, or SPO, is the practice of reducing the number of intermediaries between an advertiser and a publisher. It sounds technical, but the commercial logic is simple: fewer hops means more of your budget reaches the actual impression, and you get better visibility into where your ads are appearing.

SPO became a serious conversation in the industry as programmatic buying matured and the cost of the supply chain became harder to ignore. The move toward direct deals, curated marketplaces, and preferred publisher relationships is partly a response to the opacity of the open exchange.

From a planning perspective, SPO is not just a procurement exercise. It is a quality signal. Advertisers who know which publishers their ads appear on, who have direct relationships with those publishers, and who have audited the path between their DSP and the impression are in a fundamentally stronger position than those who are buying blind through open exchange.

This connects to a broader point about growth strategy: the decisions that compound over time are rarely the flashy ones. Cleaning up your supply chain, building direct publisher relationships, and investing in first-party data infrastructure are unglamorous. They are also among the highest-return activities available to a programmatic buyer.

What a Commercially Sensible Ad Tech Stack Looks Like

There is no universal answer here, because the right stack depends on your objectives, your budget scale, your internal capability, and your category. But there are principles that hold across most situations.

Start with what you can actually use. A sophisticated DMP is worthless if you do not have the data quality or the analytical resource to activate it properly. A premium DSP with full transparency settings is only valuable if someone is actually reviewing the reports. Tech decisions should follow capability decisions, not precede them.

Audit your supply chain annually. Where is your money going? What percentage is reaching working media? What does your inventory actually look like? These questions should have answers, and those answers should be reviewed regularly. Forrester’s work on agile scaling makes the point that operational visibility is a prerequisite for meaningful optimisation. The same applies here.

Invest in first-party data before you need it. The brands that are best positioned in the post-cookie environment are the ones that started building first-party infrastructure years ago. If you have not started, start now. The compounding value of a clean, consented, well-structured first-party data asset is significant.

Treat verification as a baseline, not a differentiator. Brand safety, viewability, and fraud detection should be table stakes, not premium add-ons. If your current setup does not include these by default, that is a problem to fix before anything else.

Measure what matters to the business, not what the platform makes easy to measure. Platform metrics are proxies. Revenue, market share, customer acquisition cost, and customer lifetime value are outcomes. Build your measurement framework around the latter and use the former as directional signals, not proof points.

The Go-To-Market and Growth Strategy hub has more on how to connect media and channel decisions to commercial outcomes rather than treating them as separate disciplines.

The Structural Tension Between Innovation and Accountability

The ad tech industry produces a constant stream of new formats, new targeting capabilities, and new measurement solutions. Some of it is genuinely useful. A lot of it is solutions looking for problems, dressed up in language designed to make marketers feel behind if they are not adopting it immediately.

I have sat in enough agency new business pitches, on both sides of the table, to know that novelty is often used as a proxy for value. A new ad format is exciting. A rigorous audit of your existing supply chain is not. But the latter is almost always more commercially productive than the former.

The question to ask of any new ad tech capability is not “is this interesting?” but “does this solve a real problem we have, and can we measure whether it works?” If the answer to either part of that question is unclear, the right response is to wait. The ad tech ecosystem moves fast, but the fundamentals of commercial marketing do not. Reach the right people, with the right message, at a cost that makes business sense, and measure the outcome honestly. Everything else is detail.

BCG’s analysis of go-to-market strategy in B2B markets makes a related point about complexity: organisations that add capability without adding clarity tend to create cost without creating value. The ad tech stack is one of the most common places that pattern plays out in marketing.

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