Programmatic Advertising: What the Spend Data Won’t Tell You

Programmatic advertising is the automated buying and selling of digital ad inventory in real time, using data signals to match ads to audiences across display, video, audio, and connected TV. It accounts for the vast majority of digital display spend globally, and most large advertisers run it as a core channel. The problem is not how it works technically. The problem is how most marketers think about what it is actually doing for them.

The mechanics are well understood. The strategic thinking around it, in my experience, is not. Too many teams treat programmatic as a performance channel, optimise toward last-click metrics, and end up spending a lot of money reaching people who were already going to buy. That is not a technology problem. It is a planning problem.

What Programmatic Advertising Actually Is

At its core, programmatic is an infrastructure layer. Publishers make inventory available through supply-side platforms (SSPs). Advertisers bid for that inventory through demand-side platforms (DSPs). A data management layer sits in the middle, informing who gets shown what, at what price, in what context. The whole thing happens in milliseconds, at scale, across thousands of sites and apps simultaneously.

There are several buying models worth understanding. Real-time bidding (RTB) is the open auction model, where inventory is bought competitively in real time. Private marketplace deals (PMPs) give buyers access to premium inventory with more transparency and control. Programmatic direct locks in guaranteed inventory at a fixed price, removing the auction entirely. Each has different cost profiles, inventory quality implications, and use cases.

The major DSPs include The Trade Desk, DV360 (Google’s platform), Amazon DSP, and a handful of others depending on your region and vertical. Each has different strengths in terms of inventory access, data integrations, and measurement capabilities. The choice of platform matters less than most vendors would have you believe. The strategy you run through it matters considerably more.

Why Most Programmatic Strategies Are Too Narrow

Early in my career, I overvalued lower-funnel performance channels. I was drawn to the clean attribution story: someone clicked, someone converted, here is your cost per acquisition. It felt like proof. What I came to understand, after managing hundreds of millions in ad spend across more than 30 industries, is that a significant portion of what performance channels get credited for was going to happen anyway. You are often paying to intercept demand that already existed, not paying to create new demand.

Programmatic retargeting is the clearest example. You build a retargeting pool of everyone who visited your site, you serve them ads, they convert at a higher rate than cold audiences, and the platform reports a strong return on ad spend. But how many of those people would have come back and converted without seeing the ad? The honest answer is: a meaningful proportion of them. You are not always causing the conversion. You are often just present at the moment it was going to happen regardless.

This is not an argument against retargeting. It is an argument for not letting retargeting dominate your programmatic mix. If your audience pool is shrinking because you are not running enough upper-funnel activity to refill it, your retargeting performance will look strong right up until it does not. You have been harvesting, not planting.

Think about it like a clothes shop. Someone who walks in and tries something on is far more likely to buy than someone who has never been in the door. Retargeting is the equivalent of following up with the person who already tried the jacket on. That is worth doing. But if you stop bringing new people into the shop entirely, you will run out of warm prospects to follow up with. Growth requires reaching new audiences, not just recapturing existing intent. That is a principle that applies directly to how you structure your programmatic investment.

If you are working through your broader go-to-market approach, the thinking around audience architecture and channel mix fits into a wider set of decisions. The Go-To-Market and Growth Strategy hub covers those upstream planning questions in more depth.

Audience Architecture: The Part That Actually Drives Performance

The single biggest lever in programmatic is not your bidding strategy or your creative rotation cadence. It is the logic of who you are trying to reach, and why. Get that wrong and you are optimising the wrong thing at scale.

Audience architecture starts with a clear view of your customer segments: who they are, where they sit in the buying experience, and what signals indicate they are genuinely in-market versus passively browsing. From there, you build a targeting structure that reflects those segments, using a combination of first-party data, contextual signals, and third-party audience data where appropriate.

First-party data is the most valuable input you have. CRM lists, site visitor data, email engagement data, and purchase history all give you a foundation that no third-party provider can replicate. The deprecation of third-party cookies has made this more important, not less. Teams that have invested in clean, structured first-party data are in a materially better position than those who relied on cookie-based targeting and are now scrambling to replace it.

Contextual targeting has had a resurgence for good reason. Reaching someone reading about home renovation when you are selling power tools is a reasonable proxy for intent, without requiring any personal data signals. It is not as precise as behavioural targeting at its best, but it is more reliable, more brand-safe, and more durable as a strategy.

Lookalike modelling, where you use your best customers as a seed audience and the DSP finds people with similar characteristics, is a useful prospecting tool when the seed audience is large enough and clean enough to produce meaningful signals. The mistake is using too small or too broad a seed. A seed audience of 500 people produces a lookalike that is mostly noise. A seed of 50,000 qualified converters produces something worth testing.

The Inventory Quality Problem Nobody Talks About Enough

Open exchange programmatic has a quality problem that has been well documented but is still not taken seriously enough by many buying teams. Ad fraud, made-for-advertising (MFA) sites, brand safety incidents, and low-viewability placements are not edge cases. They are systemic features of open auction buying at scale.

I have seen audit reports on programmatic campaigns where a meaningful share of impressions were served on sites that no brand would knowingly associate with, or in environments where the ad had almost no chance of being seen. The platform reported delivery and impressions. The business got very little in return.

There are practical steps that reduce this exposure. Inclusion lists, where you specify the sites and apps you will buy from rather than excluding bad ones reactively, are more effective than blocklists. Pre-bid brand safety filters from third-party verification providers add a layer of protection before the impression is bought. Private marketplace deals give you access to known, premium inventory at a higher CPM but with materially better quality assurance.

The cost of doing this properly is a slightly higher CPM. The benefit is that your impressions are actually reaching real people in environments that reflect well on your brand. That trade-off is almost always worth making, particularly for brands where reputation is a meaningful commercial asset.

Frequency, Fatigue, and the Experience You Are Creating

Frequency management is one of the most consistently neglected disciplines in programmatic. It is easy to set a campaign live and let the platform optimise toward the cheapest impressions. The result is often that a small subset of your audience sees your ads dozens of times while the rest of your target market barely sees them at all.

This matters for two reasons. First, the person who has seen your ad 40 times in a week is not becoming more likely to buy with each additional impression. At some point, frequency becomes irritant. You are not building brand equity. You are eroding it. Second, the people you are not reaching at all represent missed opportunity to build awareness and consideration in your addressable market.

Effective frequency caps, cross-device deduplication, and campaign-level frequency management rather than just ad-level caps all contribute to a more even, more effective distribution of impressions. It requires more active management than most teams apply, but the payoff in both efficiency and experience quality is real.

Creative fatigue compounds frequency problems. Running the same creative for months while the platform continues to serve it to the same audience is a slow drain on performance that rarely shows up clearly in the dashboard. Build rotation into your creative strategy from the start, and monitor engagement signals at the creative level rather than just at the campaign level.

Measurement: What Programmatic Can and Cannot Tell You

This is where I want to be direct, because measurement is where the most expensive mistakes in programmatic happen.

View-through attribution, where a conversion is credited to an ad that was served but not clicked, is one of the most misleading metrics in digital advertising. It sounds reasonable in theory: someone saw your ad, then later converted, so the ad contributed. In practice, the default attribution windows are often long enough that almost any conversion can be attributed to almost any impression. The platform looks good. The actual causal relationship is unclear at best.

I judged the Effie Awards, which are specifically focused on marketing effectiveness. The entries that stood out were the ones that could demonstrate real business impact with honest measurement methodology, not the ones with the most impressive-looking dashboards. The discipline of asking “how do we know this worked, not just that it happened” is rarer than it should be.

Incrementality testing is the more honest approach. Running holdout groups, where a matched segment of your audience does not see the ads, and comparing conversion rates between the exposed and unexposed groups gives you a genuine read on whether the advertising is causing incremental outcomes. It is more complex to set up than standard attribution, but it produces a number you can actually trust.

Media mix modelling (MMM) has seen a resurgence as cookie-based attribution has become less reliable. It is not perfect, but it gives you a higher-level view of how channels are contributing to business outcomes over time, without relying on individual user tracking. For brands spending meaningfully across multiple channels, it is worth the investment.

The honest position on programmatic measurement is that it is directional, not definitive. You can get a reasonable read on relative performance, trends, and efficiency. You cannot get a precise causal accounting of every conversion. Accepting that honestly is more useful than pretending the attribution model gives you certainty it does not actually provide.

Connected TV and the Expanding Programmatic Landscape

Programmatic has expanded well beyond display. Connected TV (CTV) is now one of the fastest-growing areas of programmatic spend, and for good reason. It combines the reach and brand impact of television with the targeting precision and measurability of digital. For brands that have historically relied on linear TV, or brands that could never afford linear TV, it opens up genuinely new options.

The same principles apply. Audience architecture matters. Frequency management matters. Inventory quality matters, perhaps even more in CTV where the premium nature of the environment is part of the value proposition. The measurement challenges are, if anything, more pronounced because the connection between a CTV impression and a downstream conversion is harder to track than in display.

Programmatic audio, digital out-of-home (DOOH), and in-app environments are all part of the same ecosystem now. The ability to coordinate messaging across these environments through a single DSP, using consistent audience segments, is one of the genuine advantages of the modern programmatic stack. Whether most teams are exploiting that coordination effectively is a different question.

How to Structure a Programmatic Investment That Makes Commercial Sense

After years of running programmatic across categories from retail to financial services to FMCG, the structure that consistently performs best is one that mirrors the buying experience rather than concentrating at a single point in it.

Upper funnel activity, using broad but contextually relevant targeting to build awareness among people who do not yet know your brand, needs to be a consistent line item, not a campaign burst. The temptation to cut it when short-term pressure arrives is understandable, but the consequences show up in your retargeting pool size and your organic search volume six months later.

Mid-funnel activity, reaching people who have shown some engagement signal but have not yet expressed strong purchase intent, is where programmatic can do genuinely useful work. Sequential messaging, where you serve different creative to people based on what they have already seen or engaged with, is a tactic that most teams have access to but few use well.

Lower funnel retargeting and prospecting against high-intent signals is where most teams over-index. It should be part of the mix. It should not be the whole mix. If your programmatic strategy is primarily retargeting and lookalike prospecting optimised toward last-click conversions, you are harvesting demand more than you are building it. That is a viable short-term approach that creates a fragile long-term position.

For a broader perspective on how programmatic fits into a full growth architecture, including how it connects to demand generation, market penetration, and channel sequencing, the Go-To-Market and Growth Strategy hub is worth working through. Programmatic does not exist in isolation. Its value depends on how it connects to everything else you are doing commercially.

Understanding market penetration as a growth strategy is a useful complement to programmatic planning. If you are trying to grow share in an existing market, the role of awareness-building programmatic is different from a scenario where you are entering a new category. The channel strategy should follow the commercial objective, not the other way around.

The teams that get the most out of programmatic are the ones that treat it as a planning problem first and a technology problem second. The platforms are capable. The question is whether the strategy running through them is coherent. That requires someone asking the uncomfortable questions about what the spend is actually doing, not just what the dashboard says it is doing.

There is a reason why go-to-market execution feels harder than it used to. Channels have multiplied, measurement has fragmented, and the gap between activity and outcome has widened. Programmatic is not immune to that. If anything, its complexity makes it easier to mistake motion for progress. The discipline of connecting media investment to real business outcomes is what separates the teams that use programmatic well from the ones that just use it a lot.

Scaling a programmatic operation also requires the kind of organisational rigour that BCG’s work on scaling agile teams points to: clear ownership, fast feedback loops, and the willingness to kill what is not working rather than optimise it indefinitely. Most programmatic programmes have campaigns that have been running long past their useful life because nobody made the call to stop them.

Growth hacking as a concept has always had a complicated relationship with programmatic. The growth hacking framework is useful for rapid experimentation, but programmatic rewards consistency and structural thinking more than it rewards clever one-off tactics. The best programmatic programmes I have seen are boring in the best sense: disciplined audience architecture, clean measurement, consistent creative refresh cycles, and a clear connection to commercial objectives. They do not win awards for innovation. They deliver results quarter after quarter.

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