Programmatic is not failing because the technology is flawed. It’s failing because of what we chose to optimize for.
We built a system that rewards what is easy to measure instead of what actually drives growth. We confused cheap with efficient. We treated attribution as truth. And we let incentives quietly shape strategy.
If we want to fix programmatic, we have to confront three things:
- The incentives that built the swamp
- The metrics we mistake for impact
- The structural changes we’re still avoiding
Let’s start at the beginning.
The Swamp Wasn’t an Accident
Programmatic is capitalism at its most literal. We set machines loose and told them what success looks like. They did exactly what we asked.
The problem is we defined success in proxy metrics. Low CPMs. High CTR. Last-touch attribution. The system optimized perfectly for those signals, and the ecosystem filled in around it. More intermediaries. More supply. More inventory engineered to win auctions, not attention.
When you cannot clearly see what is happening between the DSP and the publisher, you create a perfect breeding ground for inefficiency. Not necessarily ill intent. Just incentives doing what incentives do.
Everyone played a role. Buyers, vendors, agencies, platforms. We rewarded outcomes that looked clean in a dashboard and acted surprised when the internet became cluttered and extractive.
The machine didn’t break. It followed instructions.
Cheap Is Not Efficient. Attribution Is Not Truth.
The two most damaging beliefs in programmatic are deceptively simple: that cheaper media is more efficient, and that attribution tells us what truly worked.
Cheap inventory often looks efficient because the metrics say it is. Retargeting campaigns frequently appear to outperform everything else because they capture people who were already close to converting. High video completion rates look impressive on a dashboard, even if the ad ran in an environment where no one was actively paying attention. The numbers validate the spend, so budgets follow the numbers.
Over time, we optimized toward what was easiest to measure rather than what was most meaningful to business growth. Clicks became a proxy for performance. Completion rates became a proxy for attention. Platform-reported return became a proxy for incrementality. The logic feels airtight because it is numerically supported.
But attribution, especially last-touch and platform-based models, only reflects the interactions we can see. It does not account for prior brand exposure, word of mouth, organic discovery, cultural relevance, or the cumulative effect of upper-funnel investment. When we treat that partial view as definitive truth, we overfund the bottom of the funnel and underinvest in the activities that create demand in the first place.
The problem is not measurement itself. Measurement is necessary. The problem is mistaking what is measurable for what is causal.
That is how cheap becomes “efficient,” retargeting becomes “heroic,” and strategy quietly collapses into reporting.
The One Rule That Changes Everything
If I could enforce one structural rule across programmatic buying, it would be this:
Start at the end first.
Define the business outcome you are trying to change. Then build the media plan backward from that. Not from platform defaults. Not from last quarter’s benchmarks. Not from what looks good in attribution.
If we did that consistently, parts of the ecosystem would start screaming. The long tail supply built on cheap impressions. The vanity metrics we use as emotional support. The structures that reward volume over value.
Prices would rise. Attention costs more than we have trained ourselves to accept.
But at least we would be buying something real.
