Programmatic Media Is Powerful. Most Brands Use It Wrong.

Programmatic media is the automated buying and selling of digital advertising inventory, using data and algorithms to target audiences in real time across display, video, audio, connected TV, and beyond. Done well, it gives marketers reach, precision, and efficiency that manual buying simply cannot match. Done poorly, and it becomes an expensive way to serve ads to bots, brand-unsafe placements, and audiences who were never going to buy from you anyway.

The gap between those two outcomes is wider than most media plans acknowledge.

Programmatic sits at the intersection of media strategy, data strategy, and commercial accountability. If you want to understand how it fits into a broader approach to growth, the Go-To-Market and Growth Strategy hub covers the wider context, including how media investment connects to positioning, audience development, and business outcomes.

What Programmatic Media Actually Is

Strip away the acronyms and the programmatic ecosystem is straightforward in concept. Publishers make ad inventory available through supply-side platforms. Advertisers bid for that inventory through demand-side platforms. The transaction happens in milliseconds, informed by audience data, context, and whatever targeting parameters the buyer has set. A user loads a page. A bid is placed. An ad is served. The whole process completes before the page finishes rendering.

That automation is genuinely powerful. It replaced a world of insertion orders, direct publisher relationships, and manual trafficking that was slow, expensive, and opaque in its own way. Programmatic made it possible to reach audiences at scale, across thousands of publishers simultaneously, with far greater control over who you were targeting and what you were paying.

The problem is that the complexity of the ecosystem, the layers of technology between advertiser and publisher, and the sheer volume of data involved have made it easy to confuse activity with effectiveness. Dashboards full of impressions, click-through rates, and view-through conversions can look compelling while the underlying business case is eroding quietly in the background.

I have sat in enough media reviews to know that the question “is this working?” is asked far less often than “how do we optimise this?” Those are not the same question. One interrogates the strategy. The other assumes it.

The Funnel Bias Problem in Programmatic

Earlier in my career I placed enormous weight on lower-funnel performance metrics. Conversion rates, cost per acquisition, return on ad spend. The numbers were clean, the logic felt airtight, and the reporting looked good in client decks. It took years of running larger programmes, across more industries, to understand what I was missing.

Much of what lower-funnel programmatic is credited for was going to happen anyway. Retargeting someone who has already visited your product page, added an item to their basket, or searched for your brand by name is not the same as creating new demand. You are capturing intent that already existed. The ad may have been the final nudge, but the customer was already on the path. You are taking credit for the last metre of a experience you did not start.

Think about a clothes shop. Someone who walks in and tries something on is dramatically more likely to buy than someone who walks past the window. The conversion rate looks great. But the interesting question is what got them through the door in the first place. Programmatic retargeting is very good at serving the person standing in the changing room. It is considerably less good at building the kind of brand awareness and consideration that fills the shop with new customers.

This matters commercially because a business that only optimises for existing intent will eventually run out of new customers. The pool of people already in market for your product is finite. Growth requires reaching people who are not yet thinking about you, which means upper and mid-funnel investment, which means accepting that some of the metrics will be softer and the attribution will be messier. That is not a flaw. It is the nature of building demand rather than harvesting it.

Programmatic has the tools to operate across the full funnel. Display and video for awareness. Native and content formats for consideration. Retargeting and intent-based targeting for conversion. The issue is that most programmes are heavily weighted toward the bottom because the measurement is easier, the feedback loops are faster, and the numbers look better in the short term. Sustainable growth requires a more balanced approach.

Where Programmatic Spend Actually Goes

The programmatic supply chain is long. Between the advertiser’s budget and the publisher’s ad slot, there are demand-side platforms, supply-side platforms, data management platforms, verification vendors, ad servers, and various intermediaries who take a margin at each step. The actual percentage of a media budget that reaches working media, meaning the actual cost of the impression served to a real human, has been a subject of serious scrutiny for years.

This is not a conspiracy. It is a structural feature of a complex automated ecosystem. But it is a feature that every advertiser should understand and interrogate. When I was managing large programmatic budgets across multiple markets, one of the most commercially significant exercises we ran was a supply chain audit. Not because we suspected fraud, but because understanding where the money was going was basic financial accountability. The results were consistently instructive. Costs that had been buried in technology fees, margins taken by intermediaries that had not been clearly disclosed, and inventory quality that varied significantly across the supply path.

Programmatic Guaranteed and Private Marketplace deals exist partly to address this. By negotiating directly with publishers for reserved inventory at agreed prices, buyers get more control over placement quality and supply path transparency. The trade-off is less scale and less flexibility than open auction buying. For brand-sensitive categories or premium inventory, that trade-off is usually worth making.

Invalid traffic and brand safety are related concerns. Ad fraud, where impressions are served to non-human traffic, remains a genuine problem at scale. Brand safety, where ads appear adjacent to content that damages brand equity, is a risk that has caught major advertisers out publicly. Verification tools from providers like DoubleVerify and Integral Ad Science help, but they are not perfect, and they add cost. Treating them as optional is a false economy.

Audience Strategy Is the Real Competitive Advantage

The technology in programmatic is largely commoditised. Most large advertisers are running through similar DSPs, using similar bidding strategies, accessing similar inventory. The actual point of differentiation is audience strategy, which is a function of how well you understand who you are trying to reach and what data you have to find them.

First-party data is the foundation. CRM lists, website visitor data, purchase history, email engagement. These are audiences you have a real relationship with, and they are the most reliable signal you have. Lookalike modelling extends that, using the characteristics of your best customers to find new audiences with similar profiles. Third-party audience segments have their uses but come with significant caveats around accuracy, recency, and data provenance.

The deprecation of third-party cookies, which has been slower to arrive than originally signalled but is still directionally real, has increased the premium on first-party data significantly. Advertisers who have invested in building owned data assets, consent frameworks, and clean room infrastructure are better positioned than those who have relied on third-party segments bought through the platform. This is not a new insight, but the urgency of acting on it has increased.

Contextual targeting deserves more credit than it currently gets. Reaching people based on the content they are actively consuming, rather than behavioural profiles assembled from cross-site tracking, is both more privacy-compliant and more intuitive as a targeting logic. Someone reading a long-form piece about home renovation is probably more receptive to a hardware brand than someone who visited a hardware site three weeks ago and has been followed around the internet since. Context is a signal. It just fell out of fashion when behavioural targeting became possible, and it is now being rediscovered.

Understanding how audience strategy connects to go-to-market execution is something I explore more broadly in the growth strategy section of this site. The principles that govern who you target in programmatic are the same principles that should govern your entire market approach.

Measurement and Attribution: The Honest Version

Programmatic platforms are very good at telling you what happened inside the platform. Impressions served, clicks generated, view-through conversions recorded. What they are not good at, and what most reporting does not acknowledge clearly enough, is telling you what would have happened without the advertising. That counterfactual is the actual measure of effectiveness, and it is genuinely hard to establish.

Last-click attribution is the most common approach and the most misleading. It assigns full credit for a conversion to the last touchpoint before purchase, which in a programmatic context is usually a retargeting ad served to someone who was already going to convert. It makes retargeting look extraordinarily efficient. It makes upper-funnel investment look like it contributes nothing. Neither conclusion is accurate.

Multi-touch attribution models are better but still imperfect. They distribute credit across touchpoints according to rules, whether position-based, time-decay, or data-driven, but the rules are still models, not measurements. View-through attribution, where a conversion is credited to an ad the user saw but did not click, is particularly prone to inflating performance figures. If your view-through window is set to 30 days and your brand has reasonable market presence, you will attribute a lot of organic conversions to programmatic impressions that had nothing to do with them.

Incrementality testing is the closest thing to a reliable answer. By holding out a portion of your target audience from seeing ads and comparing their conversion rate to the exposed group, you get a genuine read on how much the advertising is actually moving the needle. It requires discipline, a large enough audience to generate statistical significance, and a willingness to accept that the answer might be uncomfortable. I have run incrementality tests that showed retargeting programmes delivering far less incremental value than the attributed metrics suggested. The response was to rebalance the mix, not to suppress the finding.

Marketing mix modelling provides a complementary view at a higher level of aggregation, looking at the relationship between media investment and business outcomes over time. It is imprecise and retrospective, but it captures effects that digital attribution misses entirely, including the contribution of brand-building activity to long-run sales. The combination of MMM for strategic direction and incrementality testing for tactical optimisation is more honest than either approach alone. Feedback loops built on honest measurement compound over time in ways that flattering attribution never will.

Connected TV and the Expanding Programmatic Landscape

Programmatic has expanded well beyond display and search retargeting. Connected TV is now one of the fastest-growing programmatic channels, bringing the targeting and measurement capabilities of digital to television-scale audiences. Audio programmatic, covering streaming music and podcast inventory, has matured significantly. Digital out-of-home, where screens in retail environments, transport hubs, and public spaces can be bought programmatically, is growing.

Each of these channels has different characteristics, different measurement challenges, and different roles in the funnel. CTV in particular is interesting because it sits in the upper funnel by nature, large-screen, lean-back, high-attention, but with targeting capabilities that traditional broadcast television never had. The ability to reach a defined audience segment with a video ad in a premium, brand-safe environment, at a fraction of the cost of a broadcast buy, is a genuine capability shift.

The measurement challenge in CTV is real. Attribution from a television screen to a purchase is inherently indirect. Brands that insist on direct response metrics from CTV will either be disappointed or will find ways to manufacture the numbers. The more honest approach is to treat CTV as a brand and consideration channel, measure it accordingly with brand lift studies and search uplift analysis, and resist the temptation to squeeze it into a last-click framework. BCG’s work on brand and commercial strategy has consistently shown that long-term brand investment and short-term activation work best when treated as complementary rather than competing.

The broader point is that programmatic is no longer a synonym for display advertising. It is an infrastructure for buying media across formats and environments. The strategic questions, who are you trying to reach, at what stage of the purchase experience, with what message, measured against what outcome, remain constant regardless of which channel you are buying through.

In-House Versus Agency: The Real Trade-offs

The question of whether to run programmatic in-house or through an agency is one I have been on both sides of. Running a media agency, I have had clients who moved programmatic in-house and clients who moved it back out. The right answer is genuinely context-dependent, and anyone who tells you otherwise has a vested interest in one outcome.

The case for in-housing is real. Owning your data, your technology stack, and your audience segments means you are not dependent on an agency relationship for access to your own commercial assets. Transparency is higher. The team develops deep institutional knowledge of your specific business. For large advertisers with sufficient volume to justify the infrastructure, it can be commercially compelling.

The case against is equally real. Building a genuine programmatic capability in-house requires experienced people who are hard to hire and hard to retain. Technology costs are significant. The breadth of exposure that an agency team gets across multiple clients and categories is genuinely valuable, and it is difficult to replicate internally. Hybrid models, where strategy and data ownership sit in-house but execution and technology management sit with a specialist partner, often make more practical sense than a binary choice.

What I would caution against is making the in-house decision primarily as a cost-saving measure. The savings are often real in the short term and erode as the true cost of building and maintaining the capability becomes apparent. The decision should be driven by strategic control, data ownership, and talent strategy, not by the assumption that removing an agency margin automatically improves performance. It does not, unless the capability you build is genuinely better than the one you replace. The tools available to both in-house teams and agencies have converged significantly, which means the differentiator is increasingly the people using them, not the platforms themselves.

What Good Programmatic Looks Like in Practice

Good programmatic starts with a clear brief. Who are you trying to reach? What do you want them to think, feel, or do? Where are they in their relationship with your brand? What does success look like in business terms, not just media terms? These are not complicated questions, but they are frequently skipped in the rush to get campaigns live.

Audience architecture matters. Separating prospecting audiences from retargeting audiences, applying appropriate frequency caps to each, and ensuring the creative and messaging are calibrated to where each audience sits in the funnel. Serving a brand awareness message to someone who abandoned a basket yesterday is a waste. Serving a hard conversion message to someone who has never heard of you is worse.

Creative is consistently underinvested in programmatic. The targeting gets all the attention, but the ad itself is what the audience actually experiences. Dynamic creative optimisation can help, but it is not a substitute for strong creative strategy. The best-targeted ad in the world will not move anyone if the message is flat. I have seen programmes where the creative had not been refreshed in six months and the team was still optimising bids as if that was the limiting factor. It was not.

Governance matters too. Brand safety settings, inventory exclusion lists, viewability thresholds, and frequency caps should be reviewed regularly, not set once at campaign launch and forgotten. The programmatic ecosystem is not static. New inventory sources emerge, publisher quality changes, and the default settings in most DSPs are not optimised for your specific brand safety requirements. Agile operational frameworks apply as much to media governance as they do to product development. Regular review cycles and clear accountability for quality settings are not optional extras.

Finally, connect programmatic performance to business outcomes, not just media metrics. If your programmatic programme is delivering improving CTRs and CPAs but revenue is flat, something is wrong. Either the attribution is misleading you, the audiences are wrong, or the channel is not the constraint. Media metrics are proxies. Business outcomes are the point. Keep that hierarchy clear and you will make better decisions. Go-to-market strategy at its best treats media investment as one lever among many, calibrated to business objectives rather than platform optimisation scores.

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