Programmatic Advertising: What It Costs You to Get It Wrong

Programmatic online advertising is the automated buying and selling of digital ad inventory using real-time bidding, data targeting, and algorithmic decision-making. It replaced the manual insertion order process that once defined digital media buying, and today it accounts for the vast majority of digital display spend globally. If you are running display, video, or connected TV at any meaningful scale, you are almost certainly running programmatic.

The mechanics are well documented. The commercial reality of how most brands actually use it is considerably messier. Programmatic is one of those channels where the gap between what the technology can do and what most teams are actually doing with it is wide enough to drive a bus through.

What Programmatic Advertising Actually Is (and What It Is Not)

Programmatic advertising is a system, not a channel. It is the infrastructure through which display, video, native, audio, digital out-of-home, and connected TV inventory is bought and sold programmatically, meaning through software rather than through direct negotiation between a buyer and a publisher.

The core components are the demand-side platform (DSP), where advertisers bid for impressions; the supply-side platform (SSP), where publishers make inventory available; and the ad exchange, which connects the two. A real-time auction happens in milliseconds every time a page loads or a video pre-rolls. The winning bid gets the impression. The whole process is invisible to the user.

What programmatic is not: a guarantee of quality, a substitute for strategy, or a channel that optimises itself without human oversight. I have seen more than a few marketing teams treat programmatic as a set-and-forget channel, handing it to a DSP, setting a target CPA, and walking away. The results are usually a masterclass in efficient spending on the wrong things.

The technology is genuinely impressive. The way it is commonly deployed is often not. That gap is where most of the commercial risk lives.

How the Programmatic Supply Chain Works

Understanding the supply chain matters because it directly affects where your money goes. The simplified version: advertiser money flows through a DSP, into an ad exchange, through an SSP, and eventually to a publisher. Each layer takes a fee. The question is how much reaches the actual media, and how much disappears into the plumbing.

The programmatic supply chain has historically been opaque in ways that benefit intermediaries more than advertisers. Supply path optimisation (SPO) emerged as a response to this, with advertisers and agencies trying to reduce the number of hops between their DSP and a publisher’s inventory, cutting out resellers and reducing fees. It is a legitimate and commercially important practice that most mid-market advertisers have not yet implemented properly.

Programmatic buying modes also matter more than most briefs acknowledge. Open marketplace (OMP) buying gives you scale and low CPMs, but also maximum exposure to fraud and low-quality inventory. Private marketplace (PMP) deals give you direct access to specific publisher inventory at negotiated terms. Programmatic guaranteed (PG) is essentially the old direct buy, automated. Each has a different risk and cost profile, and the right mix depends on your objectives, not just your budget.

Early in my career I spent a lot of time optimising toward the cheapest CPM. It took a few years of looking at actual campaign outcomes to understand that cheap inventory is cheap for a reason. The economics of programmatic reward volume, not quality, unless you build quality into your buying strategy deliberately.

The Real Cost of Ad Fraud and Brand Safety

Ad fraud is not a niche concern. It is a structural feature of the open programmatic ecosystem. Invalid traffic, domain spoofing, click fraud, and made-for-advertising (MFA) sites all exist at meaningful scale. If you are running open marketplace programmatic without active fraud prevention, a material portion of your budget is not reaching real people.

The tools to address this exist. Verification vendors can filter invalid traffic and enforce brand safety at the impression level. Inclusion lists, rather than exclusion lists, are a more effective approach to brand safety for most advertisers, though they require more upfront work to build and maintain. The brands that take this seriously treat verification as a standard cost of doing programmatic, not an optional add-on.

Brand safety is a related but distinct issue. Appearing next to content that contradicts your brand’s values, or that you simply would not want to be associated with, is a real risk in open programmatic. It is also a risk that scales with your spend. The larger the budget, the more impressions, the more exposure. Contextual targeting and curated supply paths both reduce this risk, but neither eliminates it entirely.

When I was running agency operations, brand safety incidents were always uncomfortable conversations with clients. Not because they were catastrophic, but because they were preventable. The technology to avoid most of them existed. The discipline to implement it consistently was what was often missing.

Why Most Programmatic Strategies Are Over-Indexed on Retargeting

This is the one I come back to most often, because it connects to a broader misunderstanding about what performance marketing actually does.

Retargeting is the dominant use case for programmatic in most mid-market and enterprise setups. You pixel your site visitors, build audiences, and serve ads to people who have already shown intent. The conversion rates look good. The CPA looks efficient. The attribution model confirms the story. Everyone is happy.

Except a significant proportion of those conversions were going to happen anyway. The person who put something in their basket and then saw your retargeting ad was already in the buying mindset. You may have accelerated the decision marginally, or you may have just taken credit for a sale that was already coming. The measurement infrastructure in most programmatic setups cannot reliably distinguish between the two.

I spent years watching this dynamic play out across client accounts. The retargeting numbers always looked strong. But when we ran holdout tests, the incremental contribution was consistently lower than the attributed numbers suggested. Sometimes significantly lower. The channel was efficient at capturing existing intent. It was doing almost nothing to create new demand.

Think of it like a clothes shop. Someone who has already tried something on and walked out is far more likely to come back and buy than someone who has never been in the store. Retargeting is very good at following that person down the street. Prospecting programmatic is what gets new people through the door in the first place. Most programmatic budgets are heavily weighted toward the former and underinvested in the latter.

Growth requires reaching new audiences, not just recapturing existing intent. Programmatic is well suited to both objectives, but the targeting, creative, and measurement approach for each is fundamentally different. Treating them as the same thing is where a lot of wasted spend originates. If you are building a go-to-market strategy that relies on programmatic as a growth driver, the broader principles of growth strategy are worth revisiting before you brief the DSP.

Data Strategy: First-Party, Third-Party, and the Deprecation Problem

Programmatic advertising has always been data-dependent. The value proposition is that you can reach the right person, in the right context, at the right moment. That requires data. The question of which data, from where, and how reliably it maps to real people is where most programmatic strategies get complicated.

Third-party cookie deprecation has been discussed so extensively that it has almost become background noise, but the practical implications are still working through the industry. Third-party audience segments, the kind you buy from a data provider and layer onto your DSP targeting, are of variable quality at the best of times. The signal loss from cookie deprecation makes them less reliable still. Contextual targeting, which does not depend on user-level data, has seen a corresponding resurgence in interest.

First-party data is the asset that most advertisers should be building more aggressively. CRM audiences, site visitor segments, and customer match lists all give you targeting that is grounded in real commercial relationships rather than probabilistic inference. The CPMs are often higher when you target against first-party data, but the relevance and conversion quality tend to justify it.

Clean rooms, where advertisers and publishers can match data without sharing raw user information, are the direction the industry is moving for more sophisticated first-party data activation. They require meaningful first-party data assets to be worth the effort, which is another reason to treat CRM and data strategy as a prerequisite for programmatic sophistication, not an afterthought.

Measurement: Where Programmatic Attribution Goes Wrong

Programmatic measurement is one of the areas where the industry has consistently oversold its capabilities. View-through attribution, which credits a conversion to an ad that was served but never clicked, is the most obvious example. A user sees a display ad, does nothing, converts later through a different channel or directly, and the programmatic campaign takes the credit. The attribution window is usually configurable, which means the advertiser can choose how generous they want to be with themselves.

Last-click attribution has the opposite problem: it gives all the credit to the final touchpoint and none to the programmatic impressions that may have contributed to the decision earlier in the process. Neither model is accurate. Both are used constantly.

Incrementality testing is the most honest way to understand what programmatic is actually contributing. Hold out a portion of your target audience, do not serve them ads, and compare conversion rates between the exposed and unexposed groups. The difference is the incremental lift. It requires methodological discipline and usually produces numbers that are less flattering than the attributed figures, which is probably why it is less commonly used than it should be.

I judged the Effie Awards over multiple cycles, and the measurement quality in programmatic-heavy entries was consistently one of the weaker areas. Brands would present impressive attributed numbers with almost no discussion of incrementality, holdout testing, or how they had accounted for organic demand. The work was often good. The measurement framework often was not honest enough to prove it.

Programmatic analytics tools give you a perspective on performance. They are not a neutral record of what happened. The difference matters when you are making budget allocation decisions. Forrester’s research on intelligent growth models makes a related point: the brands that grow consistently tend to be the ones that are most honest about what their measurement is and is not telling them.

Choosing the Right DSP for Your Objectives

The DSP choice is consequential and often treated as less important than it is. The major platforms (DV360, The Trade Desk, Amazon DSP, and others) have meaningfully different strengths, data access, inventory relationships, and fee structures. The right choice depends on your objectives, your data assets, and the channels you are prioritising.

DV360 integrates tightly with the Google ecosystem, which is an advantage if you are running YouTube, Display Network, or relying heavily on Google audience data. The Trade Desk has strong CTV and audio capabilities and is often preferred for open web programmatic at scale. Amazon DSP offers unique purchase intent data that is genuinely differentiated for e-commerce and CPG advertisers. These are not interchangeable tools.

The fee structures also vary, and the total cost of running programmatic (DSP fees, data costs, verification costs, creative serving) can add up to a significant portion of your gross media spend. Understanding the all-in cost, not just the media CPM, is a basic commercial discipline that is often skipped in the excitement of setting up campaigns.

When I was growing the agency from around 20 people to over 100, one of the most commercially significant decisions we made was around which DSP partnerships to invest in and which to walk away from. The wrong DSP for a client’s objectives is not just a performance problem. It is a trust problem, because the numbers will look worse than they should and the client will not always know why.

Where Programmatic Fits in a Full-Funnel Strategy

Programmatic is a delivery mechanism, not a strategy. It can serve awareness objectives at the top of the funnel, consideration objectives in the middle, and conversion objectives at the bottom. The mistake is assuming that a single programmatic setup, optimised for CPA, will serve all three effectively. It will not.

Upper-funnel programmatic requires different creative formats, different success metrics (reach, frequency, brand recall), and different audience targeting logic. You are trying to introduce your brand to people who do not know it yet, or deepen familiarity with people who have had limited exposure. The optimisation signal is not a conversion event. It is attention, context quality, and reach against a defined audience.

Mid-funnel programmatic is where contextual relevance and sequential messaging become important. You are trying to move someone from awareness to active consideration. The creative needs to do more work. The targeting needs to reflect where someone is in their decision process, not just who they are demographically.

Lower-funnel programmatic, including retargeting, is where most budgets are concentrated and where the attribution inflation problem is most acute. It has a role. It is just not the only role, and for most brands that want to grow rather than just harvest existing demand, it should not be the dominant one.

BCG’s work on brand and go-to-market strategy alignment makes a point that applies directly here: the brands that sustain growth are the ones that invest across the funnel, not just at the bottom where the measurement is easiest. Programmatic makes it technically possible to do this at scale. Whether most brands actually do it is a different question.

If you want a broader framework for how programmatic fits into a growth plan, the articles and thinking collected in The Marketing Juice’s go-to-market and growth strategy hub cover the underlying principles in more depth.

What Good Programmatic Governance Looks Like

Programmatic without governance is a slow leak. The campaigns run, the dashboards show numbers, and the budget depletes, but nobody is asking the hard questions about where the money is actually going or what it is actually doing.

Good governance starts with clear ownership. Someone needs to be accountable for the supply path, the brand safety settings, the fraud verification setup, and the audience strategy. In agency relationships, this accountability often gets diffused across multiple people and nobody owns it clearly. In-house teams sometimes have the opposite problem: one person owns everything but does not have the bandwidth to manage it properly.

Regular log-level data audits, where you look at the actual impression-level data rather than the aggregated dashboard, are one of the most effective ways to identify where spend is going that should not be. Most advertisers never do this. The ones that do consistently find something worth fixing.

Cadence matters too. Programmatic campaigns that are reviewed monthly are not being managed properly. The auction dynamics, frequency caps, audience saturation, and creative fatigue all move faster than a monthly review cycle can catch. Weekly reviews of key signals, with monthly deeper dives into supply path and audience quality, is a more appropriate rhythm for active programmatic investment.

Tools like those covered in Semrush’s growth tool roundup and the behavioural analytics approaches discussed at Crazy Egg can complement programmatic governance by giving you a clearer picture of what is actually happening on-site after the click, not just what the DSP reports. The two perspectives together are more honest than either alone.

Vidyard’s analysis of why go-to-market feels harder than it used to touches on a theme that resonates here: the proliferation of tools and channels has made it easier to be busy and harder to be effective. Programmatic is a good example of a channel where activity is easy to generate and genuine commercial impact is harder to prove.

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