The Flight to Premium: Why the Open Exchange Keeps Losing Share

For most of the last decade, the open exchange was where programmatic buying started. It was cheap, it was everywhere, and it gave buyers access to more inventory than any single publisher deal ever could. That default is quietly ending. Budgets are moving toward private marketplaces and curated supply, and the shift is large enough that it now changes how campaigns should be planned.
This article looks at what the 2026 data shows, why buyers are making the move, and how to think about the price difference.
The numbers behind the shift
The clearest picture comes from transaction data. An analysis by Searchlab in March 2026, built on Index Exchange and PubMatic data, found that open exchange spending dropped from 48% of programmatic in 2023 to roughly 36% in 2026, while private marketplace deals climbed to about 42%. Other 2026 reporting puts the combined PMP and curated share closer to 41%, so the exact figure depends on who is counting, but the direction is consistent across sources.
Curation is the fastest-moving piece. Searchlab's reporting on Audigent and Magnite data has curation layers handling 22% of all programmatic spend in 2026, growing around 45% year over year. Experian puts it differently but tells the same story: more than 66% of open-exchange spend, worth over $100 billion a year, now runs through curated private marketplaces.
Why buyers are moving
Three pressures explain most of the shift.
Inventory quality. HUMAN Security's Bot Baseline reporting puts fraud at 14–18% of open exchange impressions, compared with 3–5% on private marketplace deals. That gap is hard to argue with when a quarter of a budget is at stake.
Waste. Basis estimated global programmatic waste and inefficiency at about $26.8 billion in 2025. The same research found 54% of advertisers believe generative AI has contributed to a decline in overall media quality, which makes low-cost, high-volume inventory harder to trust than it was two years ago.
Availability. This one is simple and often missed. Premium publishers, most CTV apps and the bulk of retail media inventory are not sold on the open exchange at any price. If a plan needs those environments, a deal is the only way in.
The price question
Curated inventory costs more per impression, and the gap is real. PubMatic's 2026 analysis puts deal CPMs around 38% above open exchange on average. Benchmark data for display shows open exchange clearing near $5.85 against $8.20 for private marketplace, a spread that has grown more than 60% since 2024.
The more useful comparison is cost per impression that is actually viewable and actually human. Once fraud and viewability are factored in, several 2026 analyses find the effective cost roughly evens out. Paying twice the CPM for inventory that delivers is not a premium, it is a different way of buying the same outcome.
Supply path optimisation adds to this. The ANA's work on programmatic supply chain transparency has repeatedly found that cutting redundant intermediaries raises working media spend by 15–20%. Fewer hops usually means more of the budget reaching a publisher.
What this means in practice
None of this makes the open exchange obsolete. It makes it a specific tool rather than a starting point.
Use the open exchange for scale, with strong pre-bid filtering in place. For broad reach at low cost it still does a job nothing else does as cheaply.
Use curated deals for control. Curated marketplaces sit between the open auction and a negotiated PMP: vetted supply pools with transparent pricing and no per-publisher negotiation. Minimums are lower than direct PMPs, which makes them realistic for mid-market budgets.
Do not assume a PMP is automatically better. A deal is only as good as its curation. A poorly assembled invite list can perform no better than the open auction it was meant to replace, so ask who is actually in the deal before committing budget.
Audit the path, not just the price. Knowing how many intermediaries sit between a bid and a publisher is often more valuable than shaving a dollar off a CPM.
Where this leaves planning
The honest summary is that the open exchange has been demoted, not killed. It has gone from the default buying environment to one option among several, and the burden of proof has flipped. Buyers used to justify why they were paying more for a deal. In 2026 they increasingly have to justify why they are running open auction inventory without a filter on it.
For anyone planning Q4 and beyond, the practical move is to stop treating the split as fixed. Look at what share of spend currently runs open versus curated, compare the two on fraud-adjusted and viewability-adjusted cost rather than raw CPM, and let that number decide the allocation. That comparison is usually more revealing than any industry benchmark.


