Programmatic Advertising Is Broken. Here's What Replaces It.

Every year, the industry publishes another report about ad fraud, MFA sites, and supply chain waste. Every year, the platforms promise more transparency. And every year, the problem persists — because the opacity isn't a flaw. It's a feature.

Programmatic advertising was designed around intermediaries. DSPs, SSPs, exchanges, verification vendors, data brokers — each layer takes a cut, each layer adds complexity, and each layer has a financial incentive to keep the system exactly as opaque as it is. The advertisers paying for it can't see through the stack. The publishers creating content get a fraction of the spend. The intermediaries in the middle extract the rest.

I've seen this from the inside, running media technology at holding companies that move billions through these pipes. The uncomfortable truth is that the people who could fix the transparency problem are the ones who profit most from its absence.

Made-for-advertising sites — pages engineered to attract programmatic spend with no genuine audience — get a lot of attention. And rightly so: by some estimates, 15-20% of programmatic display spend ends up on MFA inventory. That's billions of dollars buying impressions that no human being with any purchase intent will ever see.

But MFA is just the most visible symptom of a deeper pathology. The real disease is a buying system that optimises for cost-per-impression without meaningful quality signals. When your buying algorithm's primary objective is cheap reach, it will inevitably find the cheapest inventory — and the cheapest inventory is almost always the worst.

Verification tools help, but they're reactive. They tell you after the fact that your ads appeared somewhere worthless. What you need is a system that never buys that inventory in the first place — one that evaluates quality before the bid, not after the impression.

This is why we built Lancelot. Not as another verification layer bolted onto a broken system, but as an intelligence layer that maps the supply chain, identifies MFA and fraud patterns, and prevents waste before it happens. The distinction matters: verification is an audit; what's needed is prevention.

The next generation of media buying won't look like programmatic as we know it. It will be AI-driven, but the AI will work for the advertiser — not for the intermediary.

That means buying systems that can evaluate publisher quality in real time, that understand the context of a page rather than just its category label, and that optimise for business outcomes rather than proxy metrics. It means supply chain transparency enforced by technology, not promised by vendors.

The shift is already underway. Brands that have brought media buying in-house, or work with specialist partners rather than holding company trading desks, are seeing better outcomes — not because in-housing is inherently superior, but because it removes the layers of intermediary incentive that corrupt the current system.

The question for every CMO is straightforward: do you know where your programmatic spend actually goes? Not the verification report — the actual, end-to-end flow of money from your budget to a human being's screen. If the answer is no, you're funding a system designed to prevent you from finding out.

The post-programmatic world isn't a world without automation. It's a world where automation serves the buyer instead of the intermediary. Where quality signals are built into the bidding logic, not layered on as an afterthought. Where the supply chain is auditable by design, not opaque by default.

This requires new infrastructure. Lancelot is part of that infrastructure — an AI system that maps the programmatic supply chain, scores inventory quality before purchase, and gives advertisers the visibility they've been promised for a decade but never actually received.

But infrastructure alone isn't enough. It also requires a willingness to challenge the incumbent model — to accept that the system you've been buying through has structural incentives that work against you, and to demand something better. The technology to fix programmatic exists. The question is whether the industry has the will to use it.