DOOH Demand Ecosystem

Programmatic digital-out-of-home (DOOH) is the application of real-time bidding, audience-based targeting, and standardized supply-chain protocols to physical screens in public venues. The mechanics borrow from web and CTV programmatic — an ad opportunity is described as a structured bid request, a waterfall of demand sources responds with bids or no-bids inside a few hundred milliseconds, the winner's creative is fetched and rendered, and a measurement event is fired back to confirm the play actually happened. The difference is that the impression is a public-space exposure on a fixed-location screen rather than a single-user browser session, which changes everything about how identity, attribution, and viewability work.

The Trillboards network operates on this programmatic substrate end-to-end. Every ad opportunity in 8,715 active screens across 22 countries is exposed to a multi-SSP supply chain over the IAB Tech Lab's OpenRTB 2.6 protocol, with the DOOH extension fields populated (venue category, screen dimensions, dwell-time estimates, sensing-derived audience signals where the screen supports them). Buyers reach this supply through their own DSP integrations to those SSPs; we do not publish or expose the identities of individual demand partners on this page or elsewhere on the site, but the aggregate shape of the supply chain — number of SSPs, number of DSPs reached, integration depth, latency distribution — is what this cornerstone documents.

Each Trillboards screen requests an ad whenever its scheduler decides the next slot in the playlist is a paid spot. That request becomes an OpenRTB bid request inside our ad server, enriched with the screen's venue category (IAB DOOH venue taxonomy), geo (state, city, lat/lon when consented), dwell-time bucket, and — where the screen is sensing-enabled — anonymized audience signals derived from on-device computer vision. The bid request is then fanned out in parallel to our connected SSPs.

Each SSP routes the opportunity through to its own DSP buyers, which respond inside their bid-window timeout (we set 300 ms for first-look auctions, 450 ms for second-look). The first bid above floor wins; the second-priced clearing rule (when a partner supports it) sets the actual paid price. We log every outcome in our vast_requests table: bid_received, no_bid, error, timeout, render_started, play_completed. The waterfall is structurally redundant — if SSP A returns no_bid, SSP B and C are already in flight; the aggregate fill is the union, not the intersection.

For a deeper view of how request volume converts to fills and impressions through the waterfall, see the Fill Rates chapter of the State of DOOH 2026 report.

Trillboards is integrated with multiple supply-side platforms covering the major programmatic DOOH demand pools — both global Tier 1 SSPs and DOOH-specialist exchanges. Through those SSPs we expose inventory to a much larger set of demand-side platforms, since a single SSP integration unlocks every DSP plugged into that SSP. The aggregate effect is that an ad opportunity on any Trillboards screen is reachable by a broad surface of programmatic buyers without requiring bilateral integration with each DSP.

The supply chain itself is verifiable end-to-end. Every SSP we integrate with publishes a sellers.json that identifies Trillboards as a publisher; we in turn publish ads.txt and app-ads.txt authorizing those SSPs on our domain, plus adagents.json for partners that aggregate inventory on our behalf. This four-file cross-walk is the modern IAB Tech Lab standard for supply-path transparency and is the floor requirement for inclusion in any reputable DSP's curated marketplace.

Full breakdown of integration depth, compliance status, and aggregate counts: SSP and DSP integration depth.

Latency is the single most under-discussed lever in programmatic DOOH. A 300 ms bid window sounds generous compared with web display's 100 ms norm, but DOOH ad calls cascade through several systems — the screen's ad client, our ad-decision service, each SSP, each SSP's DSPs, the creative-serving CDN — before bytes start flowing to the player. Every 50 ms of latency at the bid stage shifts spend toward slower decisioners and away from the buyers with the best identity match.

Our aggregate bid response latency sits below the IAB Tech Lab's 300 ms recommendation in the median case, with p95 within the protocol's declared timeout. We measure latency from the moment our ad server dispatches the OpenRTB request to the moment we have a complete response back (bid or no_bid), exclusive of network egress to the screen. Aggregate distributions — never per-partner, never per-screen — are documented in the latency benchmarks sub-page.

Full distribution, histograms, and IAB Tech Lab / AdExchanger industry comparisons: Programmatic DOOH bid response latency benchmarks.

Trillboards emits OpenRTB 2.6 bid requests with the DOOH extension fields populated. The DOOH extension — introduced in OpenRTB 2.6 alongside CTV — adds fields the web and mobile schemas didn't need: structured venue taxonomy IDs, screen physical dimensions, dwell-time bucket estimates, public-space disclosures, and the absence of any per-user identifier. Thedooh object replaces user as the impression-context root because DOOH has no user.

We populate venue taxonomy using the IAB Tech Lab's DOOH Venue Taxonomy 1.1, which gives buyers a stable cross-vendor vocabulary for venue type, foot-traffic profile, and dwell context. Audience segments ride the same OpenRTB segments array used in web and CTV, with segtax=4 identifying IAB Audience Taxonomy 1.1 nodes and segtax=600 identifying the Trillboards namespace for computer-vision-derived signals that have no direct IAB equivalent.

On the response side, we support VAST 4.3 wrapper and inline creatives, the OpenMeasurement (OMID) SDK for viewability, and the standard IFA-less measurement chain (proof-of-play confirmations, view-through impressions, quartile beacons). Every bid response is validated against the protocol schema before we accept it into the auction.

We expose 1,558 IAB Audience Taxonomy 1.1 nodes (segtax=4) across the network, covering interests, demographics, life-stage, in-market, and B2B segments. The full IAB AT 1.1 catalog is the broadest cross-vendor audience vocabulary the programmatic industry has standardized; every modern DSP knows how to target against it without a custom integration.

Alongside the IAB segments, we publish a Trillboards namespace (segtax=600) for computer-vision-derived signals that have no IAB equivalent. The declared classes include audience_group_composition (solo / pair / small_group / mixed_group / family_unit / coworkers), audience_intent_stage (awareness / consideration / decision / post_purchase / unknown), audience_attire_archetype (business / casual / athleisure / formal / uniform / streetwear / outdoor), audience_activity_macro (commuting / dining / shopping / leisure / waiting / working / transit), and a short structured audience_engagement_narrative rationale. These signals are aggregate and anonymous — never tied to a person, never persisted as a face attribute.

Full segment catalog and explanation of the two-namespace taxonomy: Audience signals available to buyers.

We have classified 61,146 creatives against the IAB content category taxonomy, age-rating tiers, and venue-appropriate-context flags. Every creative passing through Trillboards inventory is reviewed before it can serve — by automated classifiers (multimodal LLM + image classifiers + audio transcript matching) followed by human review on flagged or low-confidence verdicts.

The classification covers nine independent dimensions (alcohol, tobacco, gambling, political, religious, adult content, violence, profanity, plus the composite age-rating). Each dimension has both a confidence score and a binary decision. The decisions feed the ad-decision service's pre-bid filters so creatives that violate a venue's rules — say, alcohol on a screen in a family-restaurant — are dropped from the eligible set before the auction starts.

Full workflow, classifier composition, and review-outcome breakdown: DOOH brand safety: how Trillboards reviews creatives.

Buyers reach Trillboards inventory the same way they reach any other programmatic-DOOH publisher: through their existing DSP. If the DSP is connected to any of the SSPs we're integrated with, our inventory appears in the DSP's normal inventory browser, tagged with our publisher name and our DOOH venue taxonomy. The bidder doesn't need any special configuration to bid; the IAB AT 1.1 segments on the bid request are first-class fields, and the segtax=600 Trillboards namespace is treated as an extension that DOOH-aware DSPs already know how to read.

Buyers who want to invest more — direct insertion orders, private marketplace deals, custom audience activations — reach us through the demand-side onboarding channel on our developer portal. That channel routes the buyer to the appropriate SSP-side deal-creation flow or, for direct-sold relationships, into our account-management workflow. The bilateral relationship is supported on top of the same OpenRTB transport; we do not maintain a parallel direct-sold ad server.

The aggregate posture is: programmatic is the default, direct is layered on top when the relationship economics justify the administrative overhead, and the supply chain is fully transparent to any buyer who reads ads.txt / sellers.json / app-ads.txt. We don't publish partner identities on this site, but the cross-walk between our four-file disclosure and each SSP's sellers.json is the canonical record — and it's exactly the disclosure direction the IAB Tech Lab designed.