Programmatic Media Is Buying Efficiency, Not Marketing Strategy

Programmatic media is an automated method of buying and selling digital advertising inventory in real time, using data and algorithms to match ads with audiences across display, video, audio, and connected TV. It has made media buying faster, cheaper, and more scalable. What it has not done is make media strategy easier, or smarter, or more effective by default.

The efficiency is real. The strategic thinking it replaced is the problem.

What Programmatic Media Actually Is

Before getting into what goes wrong, it is worth being precise about what programmatic media actually covers, because the term gets used loosely in ways that obscure important distinctions.

Programmatic buying refers to the use of automated technology to purchase digital advertising. The dominant mechanism is real-time bidding, where individual ad impressions are auctioned in milliseconds as a page loads. A demand-side platform (DSP) bids on behalf of the advertiser, drawing on data about the user, the context, and the advertiser’s targeting parameters. The winning bid serves the ad. The whole process happens before the page finishes loading.

That is the open auction, or open exchange. But programmatic also includes private marketplaces (PMPs), where publishers invite specific buyers to bid on premium inventory. And it includes programmatic direct, or programmatic guaranteed, where a deal is agreed in advance at a fixed price but executed through programmatic infrastructure rather than insertion orders.

The channels have expanded significantly. Programmatic now covers display, online video, connected TV (CTV), digital audio, digital out-of-home, and native. The infrastructure is the same. The strategic considerations vary considerably by channel.

What connects all of it is automation and data. The automation handles the transaction. The data informs the targeting. Neither of those things is a substitute for knowing what you are trying to achieve and why.

Why Programmatic Efficiency Became a Strategic Trap

Earlier in my career, I overvalued lower-funnel performance signals. I was not alone in this. Most of the industry was, and a significant portion still is. When programmatic retargeting became widely accessible, it looked like a performance marketing breakthrough. You could serve ads to people who had already visited your site, abandoned a basket, or shown clear intent signals. The conversion rates were strong. The cost-per-acquisition looked excellent.

What I eventually understood was that much of what that retargeting was “converting” was going to happen anyway. These were people already in the purchase funnel, already moving toward a decision. The ad was capturing credit for intent that existed before the impression was served. The efficiency was real. The incrementality was often not.

Programmatic amplifies this dynamic. Because it optimises toward whatever signal you give it, and because the easiest signals to measure are the ones closest to conversion, most programmatic campaigns end up concentrating spend on audiences who were already likely to convert. The algorithm does exactly what it is told. The problem is that what it is told is usually a proxy for business value rather than business value itself.

Think of it this way. Someone who walks into a clothes shop and tries something on is far more likely to buy than someone browsing the window. If you only ever invest in reaching people who are already inside the shop and already holding the item, your conversion rate looks impressive. Your growth does not. You are not building a pipeline of new customers. You are harvesting the ones already there.

Programmatic, run without strategic intent, defaults to this pattern. It finds the low-hanging fruit and calls it performance. Genuine growth requires reaching new audiences, not just recapturing existing intent. That is a strategic decision that no DSP algorithm will make for you.

If you are thinking about where programmatic sits within a broader commercial framework, the Go-To-Market and Growth Strategy hub covers the structural decisions that should sit upstream of any channel investment.

The Targeting Problem Nobody Talks About Honestly

Programmatic targeting is sold as precision. In practice, it is often approximate at best and actively misleading at worst.

Third-party audience data, the kind you buy through a data management platform or directly through a DSP, is built on probabilistic modelling. A user categorised as “in-market for a car” has not told anyone they are shopping for a car. They have been assigned that label based on behavioural signals that an algorithm has interpreted as indicating intent. The accuracy of that interpretation varies enormously, and most advertisers have no visibility into how the segment was constructed.

I have run campaigns where third-party audience segments delivered almost identical results to broad, untargeted buys on the same inventory. The targeting cost more. The performance was indistinguishable. The data vendors had no incentive to surface that finding, and the DSP had no incentive to challenge it either. The whole ecosystem profits from the perception of precision.

First-party data is a different matter. If you are targeting based on your own customer data, your CRM lists, your site visitors, your app users, you are working with something real. First-party data in programmatic is genuinely powerful. The problem is that most advertisers either do not have enough of it, do not have it in a form that connects cleanly to programmatic infrastructure, or have not invested in the clean room or identity resolution layer needed to use it properly.

Contextual targeting has made a quiet comeback since the deprecation of third-party cookies became a serious operational concern. Targeting based on the content of the page rather than the inferred characteristics of the user is less precise in theory but often more reliable in practice. It is also considerably easier to verify. You can see where your ads ran. You cannot see whether the audience segment you paid for was what it claimed to be.

Inventory Quality and Where Your Ads Actually Run

The open exchange is vast. It is also, in significant portions, not a place where advertising works particularly well.

Invalid traffic, which covers both bot traffic and various forms of ad fraud, remains a genuine structural problem in programmatic. Viewability rates on open exchange inventory are often poor. Brand safety controls, even with sophisticated inclusion and exclusion lists, are imperfect. The environments where ads run on the open exchange are frequently low-quality, low-attention contexts where even a perfectly targeted impression generates minimal commercial impact.

When I was running agency operations and we started doing serious supply path optimisation work for clients, the results were consistently the same: concentrating spend on fewer, higher-quality supply paths improved performance without increasing cost. We were buying fewer impressions, paying similar or slightly higher CPMs, and seeing better outcomes. The efficiency of the open exchange is partly an illusion created by cheap inventory that does not work.

Private marketplace deals and programmatic direct give you more control. You know which publishers you are buying from. You can negotiate on placement quality, not just audience parameters. You have a direct relationship with the supply side that creates accountability in a way that open exchange buying does not.

For brand-building objectives especially, the environment matters. An ad seen in a high-quality editorial context, on a publisher whose brand the reader trusts, carries different weight than the same ad served in a cluttered, low-attention environment. Programmatic makes it easy to ignore this distinction because it abstracts the inventory into audiences and CPMs. That abstraction is useful. It is also dangerous if you let it obscure what you are actually buying.

Reach, Frequency, and the Audience Saturation Problem

Frequency management in programmatic is one of the most consistently mismanaged areas I have seen across agency and client-side operations. The problem is structural. When you buy programmatically across multiple DSPs, multiple channels, and multiple data sources, there is no unified frequency cap. Each platform manages frequency independently. The result is that the same user can be served the same ad, or close variations of it, many more times than any frequency cap you have set would suggest.

Beyond a certain point, additional impressions to the same person do not increase the probability of conversion. They decrease brand sentiment. I have seen this in campaign post-analyses where survey data showed brand perception declining among heavy ad-exposed segments while lighter-exposed segments showed positive movement. The algorithm was optimising for reach within a known audience. It was simultaneously eroding the brand with that audience.

The other side of this is reach itself. Programmatic campaigns often look like they are reaching broad audiences because the impression counts are large. In practice, many of those impressions are concentrated on a relatively small number of users who happen to be highly reachable across the sites and apps in the supply path. True reach, the number of distinct people you are actually reaching at meaningful frequency, is often much lower than gross impressions suggest.

For growth-oriented campaigns, this matters enormously. If your programmatic activity is repeatedly hitting the same 15% of your target audience while 85% of them never see your brand, you are not building the broad mental availability that drives long-term growth. You are running an expensive remarketing operation dressed up as a brand campaign.

Solving this requires active management: unified frequency capping where possible, reach measurement rather than just impression measurement, and deliberate investment in channels and supply paths that extend reach rather than deepen penetration of an already-reached audience. Some useful frameworks for thinking about this kind of growth challenge are covered in resources like Forrester’s intelligent growth model, which addresses how to think about expanding versus deepening market presence.

Measurement in Programmatic: What the Numbers Are Actually Telling You

Programmatic generates an enormous amount of data. Impressions, clicks, viewability rates, completion rates, conversions, cost-per-acquisition, return on ad spend. The volume of reporting creates the impression of rigorous measurement. Most of it is measuring the transaction, not the outcome.

Click-through rate is almost useless as a performance indicator for most programmatic display activity. The population of people who click on display ads is not representative of your target audience. Optimising toward clicks changes who your campaign reaches in ways that are rarely aligned with commercial objectives.

Viewability is a threshold, not a measure of attention. An ad that is 50% in view for one second is technically viewable. Whether anyone processed it is a different question entirely. The industry has spent years arguing about viewability standards while largely ignoring the more important question of whether the ads are actually being seen and processed by humans who care about them.

Last-touch attribution, which remains the default in many programmatic setups, credits the final touchpoint before conversion. In a multi-channel environment, this systematically overstates the value of lower-funnel, intent-capture activity and understates the contribution of upper-funnel, awareness-building work. It creates a measurement architecture that, over time, causes budget to migrate toward channels that look good under last-touch attribution rather than channels that are actually driving growth.

I judged the Effie Awards over multiple cycles, and one of the consistent patterns I observed was that the campaigns with the most sophisticated measurement approaches were not the ones with the most granular programmatic reporting. They were the ones that had thought carefully about what commercial outcome they were trying to influence and had built measurement frameworks around that outcome, not around the data that was easiest to collect.

Incrementality testing, media mix modelling, and brand lift measurement are all imperfect. They are also considerably more honest about what programmatic is and is not contributing than the default reporting dashboards most advertisers rely on. The goal is honest approximation, not false precision.

Tools like those covered in Semrush’s overview of growth tools can complement programmatic measurement, particularly for understanding organic and paid interaction effects that single-channel reporting misses entirely.

Connected TV and the New Programmatic Frontier

Connected TV has become one of the more interesting developments in programmatic over the past several years, and also one of the more overhyped.

The promise is genuinely compelling: the reach and brand-building power of television, combined with the audience targeting and measurement capabilities of digital. For certain advertisers, particularly those who have been priced out of linear TV or who need more precise audience segmentation than linear allows, CTV programmatic offers real strategic value.

The reality is more complicated. CTV inventory is fragmented across dozens of streaming services and platforms, each with different measurement standards, different identity resolution approaches, and different levels of transparency about what you are actually buying. Frequency management across CTV is even more complex than in display because the identity graphs connecting household-level TV viewing to individual-level digital data are imprecise.

Measurement is the most significant challenge. The metrics that work for digital display, clicks, conversions, viewability, do not translate to a lean-back TV viewing environment. Brand lift and reach measurement are more appropriate, but they require investment in measurement infrastructure that many advertisers have not made. Without that infrastructure, CTV programmatic risks being evaluated on the wrong metrics and either over-invested in or abandoned for the wrong reasons.

For brands with genuine upper-funnel objectives and meaningful budgets, CTV programmatic is worth serious consideration. For brands looking for a cheaper way to run performance marketing, it is probably not the right channel. The environment does not support that kind of activity, and the measurement architecture is not built for it.

How to Actually Run Programmatic Well

Given everything above, what does good programmatic practice actually look like?

Start with the strategic objective, not the channel. Programmatic can serve awareness, consideration, and conversion objectives. The setup, targeting approach, creative format, measurement framework, and success metrics should all be different depending on which objective you are pursuing. The mistake most advertisers make is running the same programmatic campaign for all three and wondering why the results are ambiguous.

Invest in your first-party data infrastructure before you invest heavily in programmatic scale. First-party data is the most reliable targeting input you have. If your CRM data is not connected to your programmatic buying, if your site data is not informing your audience strategy, if you have not mapped your customer segments to programmatic audience parameters, you are leaving the most valuable part of programmatic targeting unused while paying for third-party data that is far less reliable.

Be deliberate about supply path. Do not default to open exchange for everything. Identify the publishers and environments that matter for your brand and your audience, and build direct or private marketplace relationships with them. The premium in CPM is usually more than offset by the improvement in inventory quality and the reduction in invalid traffic.

Build measurement that reflects commercial reality. Agree on what success looks like before the campaign launches, and make sure the measurement framework can actually capture it. If you are running an awareness campaign, measure brand lift and reach. If you are running a performance campaign, measure incrementality, not just attributed conversions. The measurement framework shapes what gets optimised. Get it wrong and the algorithm will optimise toward the wrong thing with impressive efficiency.

Manage frequency actively and monitor reach genuinely. Set frequency caps at the campaign level, not just the placement level. Measure unique reach, not just impressions. Actively look for evidence of audience saturation and adjust spend accordingly. The algorithm will not do this for you. It will keep serving impressions to reachable users because that is what it is built to do.

And audit your supply chain. Know which DSPs you are using, which SSPs they connect to, which publishers are actually serving your impressions, and what fees are being extracted at each layer. The programmatic supply chain is complex and not fully transparent by default. Transparency requires active effort. The advertisers who have done this work consistently find that a meaningful portion of their programmatic spend was not reaching the audiences or environments they thought it was.

Understanding how programmatic fits within a broader growth architecture is part of a larger set of decisions around go-to-market strategy. If you are working through those structural questions, the Growth Strategy hub at The Marketing Juice is a useful place to think through the upstream choices that shape what programmatic can and cannot deliver.

There is also a useful parallel with how growth-stage companies think about pipeline. Vidyard’s research on untapped pipeline potential highlights how much revenue opportunity sits in audiences that are not being reached at all, a finding that maps directly onto the reach saturation problem that programmatic campaigns frequently create.

The Honest Assessment

Programmatic media is genuinely useful infrastructure. It has made media buying more efficient, more scalable, and more data-informed than it was before. For advertisers who use it well, it is a powerful tool for both brand building and demand capture.

The problem is not programmatic itself. The problem is the strategic vacuum it often operates in. When media buying becomes automated and the transaction becomes invisible, it is easy to mistake the efficiency of the process for the effectiveness of the investment. Those are not the same thing.

I have seen this pattern repeatedly across agencies and client-side operations. Programmatic budgets grow because the reporting looks good. The reporting looks good because the metrics are easy to measure and the algorithm is good at optimising toward them. The business results are ambiguous because the metrics being optimised are proxies, not outcomes. Nobody challenges the setup because the dashboard is green.

The discipline required to run programmatic well is not technical. The platforms are sophisticated and the tooling is good. The discipline required is strategic: knowing what you are trying to achieve, being honest about whether the measurement framework can tell you if you achieved it, and being willing to challenge the default settings rather than accepting the algorithm’s choices as strategy.

Programmatic is an execution layer. Strategy still has to come from humans. That has not changed, and it will not.

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