Shorten the Supply Chain: The Case for Programmatic Direct Deals

Open any modern media plan and trace a single impression from brief to publisher. It will pass through a demand-side platform, an agency trading desk layer, one or two curation marketplaces, a supply-side platform, and in many cases a header-bidding wrapper on the publisher side. Each hop adds a fee, a log file, and a latency budget. Each hop also adds an opportunity to lose signal, misattribute spend, or buy the wrong inventory at the right price.

This is the complexity tax. It is not a bug. It is the design of an auction ecosystem that pays intermediaries to exist between buyer and seller. The more intermediaries a dollar touches, the more the system earns and the less the advertiser sees.

The industry's answer has been supply-path optimization. Audit every intermediary. Rank SSPs by take rate. Prefer shorter paths to known publishers. It is good work, and it has real returns. But SPO still assumes the auction is the path. It tries to find the cleanest pipe inside a system that was engineered to have pipes.

There is another option. Stop running the auction for inventory you can buy directly.

It helps to remember why the stack exists. Real-time bidding solved a real problem in 2009: publishers had unsold inventory, advertisers needed audience scale, and a neutral auction layer let both sides match at clearing prices they could not negotiate manually. Header bidding extended the same idea to give publishers yield parity across demand sources. Curation layers emerged to re-aggregate that inventory by audience, context, or brand safety criteria.

Each layer solved a specific problem for a specific side of the market. Together they built a supply chain that neither side fully controls. The DSP and SSP were originally translation layers. The agency trading desk was a services wrapper on top. Curation was the fix for the mess those three created. Each new layer was rational on its own. The aggregate is a five-hop path to inventory the buyer could often name by URL before the auction started.

Programmatic direct is a category, not a product. Three mechanisms sit inside it, and the difference between them is the difference between a guarantee, a reservation, and a first-look invitation. Buyers who conflate them lose negotiating leverage before the deal is signed.

A programmatic guaranteed (PG) deal is a fixed-price, fixed-volume contract executed through the deal ID infrastructure. The publisher commits inventory. The buyer commits spend. There is no bidding. CPM is negotiated up front and does not fluctuate. This is the closest analog to a traditional insertion-order buy, wearing programmatic plumbing.

PG is the correct choice when the campaign depends on specific inventory delivering at a specific time: a tentpole sponsorship, a category-exclusive takeover, a launch window that cannot tolerate under-delivery. Brand campaigns with tight flight windows and high creative investment belong here.

A preferred deal (also called an unreserved fixed-price deal) sets a negotiated CPM and gives the buyer first-look access to a publisher's inventory, but does not guarantee volume. The buyer can pass on any given impression. The publisher can fill the impression elsewhere if the buyer declines. It is a reservation on the right to bid first, at a known price.

Preferred deals are the workhorse of direct programmatic. They work when a buyer wants predictable pricing on premium inventory but needs the flexibility to apply audience filters. Most always-on brand spend on premium publishers should sit here, not in open exchange.

A PMP is an invitation-only auction. The publisher curates the participating buyers. The buyers bid against each other on a defined inventory pool. Pricing is dynamic. The inventory is curated. The auction itself is private.

PMPs are the correct tool when scale matters more than guarantee, when the buyer wants exposure across a defined publisher set, and when price discovery is valuable because demand varies. Performance campaigns against premium inventory often belong here, because the audience tail is wider and the CPM efficiency of an auction helps.

The question is never "which deal type is best." It is "which deal type matches this campaign's constraints." The framework below is the one we use when structuring a buy.

The open exchange has a legitimate role. It is prospecting media. It is the right tool when the audience is undefined, the creative is tolerant of context variation, and price discovery genuinely matters. For everything else, the direct stack is shorter, more accountable, and usually cheaper on a fully-loaded basis once intermediary fees are backed out.

The case for direct is not just fewer fees. The case is accountability. Five specific gains separate a direct-first book of business from an auction-first one.

Transparent pricing. A negotiated CPM is a known CPM. There is no bid shading, no dynamic floor, no post-hoc log-file reconciliation to understand what was actually paid. Finance teams and marketing teams can agree on what a thousand impressions cost before the campaign runs.

Brand safety by construction. The ad exchange ecosystem approaches brand safety through blocklists, contextual filters, and post-impression verification. Direct deals approach it by choosing the publisher. The difference is subtractive versus additive. One removes risk after the fact. The other never invites it.

Supply-path control. In a direct deal, the buyer knows the publisher, knows the inventory, and in PG and preferred deals knows the price. There is no debate about whether an SSP took an extra two percent. The path is the deal ID, and the deal ID terminates at the publisher.

Signal integrity for measurement. Attribution models are only as good as the signal that feeds them. Direct deals deliver cleaner signal: known publishers, known placements, stable creative context, minimal auction variance. Cross-platform attribution (something we've written about before in the context of media economics) gets easier when the input data is not noise.

Relationship leverage. An advertiser spending seven figures through PG on a publisher becomes a strategic partner. An advertiser spending the same amount through an open exchange is a statistic. The commercial terms on year two, three, and four of a direct relationship compound in a way that auction buying structurally cannot.

A manifesto that does not name its trade-offs is a sales pitch. Direct deals have three.

The first is scale flexibility. A direct-first program commits to a known set of publishers. If a campaign needs to double its reach in 72 hours, the auction is the tool that can absorb the expansion. Direct deals cannot turn up volume past the publisher's inventory ceiling.

The second is price discovery. A negotiated CPM might be above the clearing price a free auction would have produced. This is the cost of predictability. Buyers who live in the open exchange often pay less per impression on paper, though they frequently pay more when supply-path fees are accounted for. The honest answer is that it depends on the category and the publisher.

The third is audience reach beyond the publisher set. Some audiences live in the long tail. Private-data cohorts, niche B2B segments, and geographically narrow targets sometimes require open-exchange reach to hit coverage goals. Direct deals can cover the premium core of a media plan, but the full tail usually requires a complementary auction layer.

The strategic point: these trade-offs are features, not bugs. A buyer who wants the auction's flexibility is buying the auction's fees. A buyer who wants the direct deal's accountability is giving up some of the auction's optionality. Pick the trade-off that matches the campaign, not the one that matches the industry's default settings.

The operational shift from auction-first to direct-first is not a tooling change. Most of it is an allocation discipline. The five-step version we use when restructuring a client's book of business:

This sequence is compatible with any agency relationship. It does not require new tools. It does require the discipline to redraw the media plan around the inventory you actually want, instead of the audiences you hope to find. The difference between a strong media buyer and a media buying agency often shows up at exactly this step.

The ad-tech industry has spent fifteen years building a supply chain and the last five years apologizing for it. Supply-path optimization is an improvement on the status quo. It is not a strategy. It is a tax audit.

The strategy is to buy less of what the supply chain sells and more of what publishers sell. Programmatic direct deals are the mechanism. They are not a nostalgic return to insertion orders. They are the application of programmatic plumbing to the relationships that actually produce accountable media.

If your current book of business runs 80 percent open exchange and 20 percent direct, flip it. The open exchange has a role, and it is prospecting. The direct stack should carry the rest. The math usually works the first quarter. The relationships compound after that. And the complexity tax stops being the cost of doing business, because you stopped doing that business.

Shorten the supply chain. Do not audit it.