Programmatic DOOH: How It Actually Works

Programmatic DOOH is the buying of digital out of home screen time through automated auctions rather than through a negotiated contract for a specific board over a specific month. That is the whole idea. Everything else is plumbing.

The plumbing matters, though, because the way it works creates very specific blind spots. Brands who buy it without understanding those blind spots end up with a healthy-looking report and a city that never actually felt like the campaign happened.

I should say plainly where I stand. AGM is a street-level operator. We do not sell programmatic screen inventory. I am writing this because clients ask us constantly how their street campaign should sit alongside a programmatic buy, and almost everything published on the subject is written by somebody selling the screens.

So here is the mechanical explanation, the honest limitations, and the part of the physical city that programmatic cannot reach at all.

What is actually happening when a programmatic DOOH ad plays

A digital screen somewhere has a slot to fill. That slot is a few seconds in a rotating loop, shared with several other advertisers.

The screen belongs to a media owner. Their inventory is exposed to the market through a supply-side platform, which packages the available slots with information about the screen: its location, its size and orientation, its venue type, the loop length, the number of slots in the loop, and an audience estimate for that location at that time of day.

On the other side, a buyer works in a demand-side platform. They set targeting rules, a flight period and a maximum bid.

When a slot becomes available, a request goes out. Buyers whose rules match the slot respond. The auction resolves, the winning creative is delivered to the screen, and it plays.

The whole exchange is automated and fast. The ad can be in the loop within moments of the decision.

That is the core. Now the parts that actually determine whether the buy works.

The three things that differ from digital advertising

Buyers coming from online media make the same three mistakes, because the vocabulary is identical and the medium is not.

One impression is not one person. On a website, an impression is roughly one browser. On a screen in a transit station, one play is seen by however many people happen to be there, which could be nobody or could be dozens. The impression figure you are buying is a modelled estimate derived from measured or sampled audience data for that location and hour. It is a reasonable estimate. It is not a count. Treat it as a planning currency rather than a fact.

There is no click. Nobody taps a screen in a lift lobby. Attribution is done by proximity, by device movement data, by matched panel surveys or by modelled uplift. All of these have real methodological limits. A programmatic DOOH report showing precise conversion attribution deserves a hard look at the methodology behind it.

Frequency is not controllable per person. You can cap how often your creative plays on a screen. You cannot cap how often a specific human sees it, because the system does not know who walked past. Someone whose commute runs through three screens you bought will see you constantly. Most of the audience will see you once.

What the targeting actually does

The targeting capability is the main selling point, and it is genuinely useful if you understand its resolution.

Location and venue type. The strongest and most reliable lever. Transit, retail, convenience, gym, office lobby, medical waiting, bar and restaurant, roadside. Venue type is a real proxy for mindset and it is the targeting I would trust most.

Daypart. Reliable and underused. Audience composition in a location at seven in the morning is genuinely different from the same location at seven in the evening, and buying only the hours that matter is the easiest efficiency available.

Triggers. Conditional delivery based on an external data feed. Weather, temperature, pollen, traffic, flight status, sports scores, stock levels, time until an event. Triggers are the most interesting capability in the format and the most frequently wasted. They work when the trigger genuinely changes the message’s relevance. They are decoration when the connection is contrived, and audiences notice contrived.

Audience segments. Built from aggregated mobile location data, showing which screens over-index for a given behavioural group. This is directional and should be treated as directional. The resolution is a location and a time window, not an individual.

Creative versioning. Different executions by location, hour or trigger state. Simple to do and consistently underused, because most campaigns deliver one file everywhere.

Where the format is genuinely strong

I am not here to talk anyone out of programmatic DOOH. It does several things well that nothing else does.

Speed. A campaign can be live in a day. If something happens in the news, in a sports result, in your own supply chain, the medium can respond. Traditional out of home cannot.

Flexibility. You can start, pause, shift markets and change creative mid-flight. That is a genuine operational advantage over a fixed contract.

Geographic precision at scale. Running a defined radius around hundreds of retail locations across a country is straightforward programmatically and would be a logistical nightmare any other way.

Testing. Multiple creative versions across comparable locations gives you real comparative data, if you set it up deliberately.

Low commitment entry. You can run a small defined test without a long contract, which matters for brands who have never used out of home.

The blind spots nobody selling screens will lead with

Here is the part that affects how you should plan the rest of your campaign.

You can only buy what is on an exchange. Programmatic reaches screens whose owners have made inventory available through a platform. Plenty of screens are not there. Plenty of physical surfaces are not screens at all.

Share of voice is divided by design. Your creative occupies a few seconds of a rotating loop shared with other advertisers. In most loops you are one of several. A person standing in that space does not experience a brand presence. They experience a sequence of brands, of which you are one.

Attention is incidental. Screens in transit, lobbies and retail catch people who are moving, queuing, or looking at their phones. The format is good at exposure and weak at anything requiring sustained attention. Complex messages do not survive a few seconds in peripheral vision.

Neighbourhood texture is invisible to it. A platform knows a screen’s coordinates and venue type. It does not know that the block has a particular culture, that a specific corridor is where your audience actually spends its Saturday, or that a wall on the route to the venue is seen by everybody who goes to that venue. Those are physical facts about a city that no feed contains.

Screens cluster where commercial real estate is. Transit hubs, malls, chains, office buildings, roadside. That is a real map of a city’s commercial infrastructure and a poor map of its cultural life. If your audience is young and their city is a set of neighbourhoods rather than a set of malls, the screen inventory is pointed at the wrong places.

Measurement runs on estimates. Worth repeating because it is the thing most often forgotten by the time the report circulates.

Creative specifications vary by screen and the lowest common denominator wins. Every screen has its own resolution, aspect ratio and permitted duration. Running across a broad pool means producing several versions or accepting that your file gets letterboxed, cropped or rejected on part of the inventory. Campaigns that supply one landscape file routinely lose the portrait screens without anyone noticing until the delivery report comes back light in a market.

Pauses are not instant everywhere. Flexibility is real but it is not uniform. Some inventory updates within moments and some sits behind a slower refresh, which matters if you ever need to pull a message quickly for a reason outside the campaign.

What the physical street does that programmatic cannot, and how it runs

This is where my side of the business sits, so take the bias into account and judge the argument on its merits.

A pasted poster campaign on a chosen corridor is not shared with anyone. It occupies that surface continuously for as long as it holds. A person who walks that block daily sees it every day. That is real frequency against a specific population in a specific place, and it is achieved by physically being there rather than by bidding.

It also reaches places with no screens. The corridors that actually define a neighbourhood’s character rarely have digital inventory, and those are frequently the exact streets a culturally-driven brand needs to be on.

And it signals something different. A brand present on the street in a neighbourhood reads as part of that place. A brand on a loop in a transit station reads as a brand that bought media. Both are legitimate. They are not interchangeable, and audiences distinguish between them whether or not they could articulate why.

The practical plan I recommend most often is simple. Use programmatic for breadth, speed and responsive messaging across a market. Use physical street formats for depth and frequency in the three or four neighbourhoods that actually matter to the brand. The combination outperforms either at the same total effort, because they fail in different places.

Wheatpasting & Poster Campaigns deliver that continuous corridor-level presence over weeks. Projection Advertising gives you large-format digital presence in the evening on surfaces that carry no screens and allow no permanent application, which is the closest physical analogue to a digital buy and is entirely yours while it runs. LED Trucks & Mobile Billboards put a full digital screen on the routes you choose with complete share of voice, which is the thing a shared loop structurally cannot offer. Sidewalk Decals & Stencils work at ground level on the final approach to a door, where no screen exists.

Crew, hours and how the physical layer actually runs.

Since the comparison only means something if you know what the other side involves, here is the operational reality.

Poster and snipe installs are overnight work, generally between roughly eleven at night and five in the morning. Crews work in pairs. One works the surface, one watches the street and manages material. Overnight because the surfaces are cool, the paste behaves properly, and the sidewalk is not full of pedestrians.

Surface preparation is the step that decides whether the campaign looks good in week two. Dust and grit prevent adhesion, and a wall that looks clean will not hold until it is swept down. Five minutes of prep is the difference between weeks of hold and curling corners by the second morning.

Paste holds on raw masonry, bare brick, concrete block and untreated plywood. It struggles on glossy sealed paint, metal, and anything carrying a silicone or anti-graffiti coating, because those coatings exist specifically to defeat adhesion.

Weather is a planning input. Humid markets give long working time and slow drying. Dry markets dry fast and force crews to work smaller sections with a wetter mix. Below freezing, water-based paste becomes unreliable. Heavy rain in the first hours before paste sets will cost you sites, although light rain during an overnight install is survivable.

Projection runs in the evening after full dark, with a crew of two to three, and it lives or dies on the surface. Flat, light-coloured, unobstructed and with controllable ambient light. A brightly lit wall defeats a projector regardless of its output.

LED trucks run routes on a schedule, and the variable that matters is dwell. A truck moving at speed down a highway delivers very little. A truck working congested streets, event approaches and queue zones at the right hours delivers a great deal. Routing is the whole craft.

Permission and the honest grey area

Programmatic has a clean answer here. You are buying authorised inventory from a media owner. That is one of its genuine advantages and I will not pretend otherwise.

Physical street work is a grey area. What is permissible varies by municipality, enforcement is inconsistent between cities and sometimes between blocks within one city, and the factors that determine exposure are who owns the surface, whether you have that owner’s permission, and what the local ordinance says about that space.

We reduce the uncertainty practically rather than legally. We work private surfaces wherever possible with permission from whoever actually controls the property. We document that permission. We keep installs non-destructive so there is never a damage conversation. We commit to a removal date and we remove on schedule, which is the main reason property owners let crews back.

We are not attorneys and none of this is legal advice. We tell clients plainly where the uncertainty sits for their markets and surfaces so they can take their own counsel. Brands with low risk appetite should know that projection and LED trucks sit differently from applied formats, and that we will route them accordingly.

What we talk clients out of

Buying programmatic DOOH as the entire out of home plan. It covers commercial infrastructure well and neighbourhood life poorly. If the brand needs to feel present in a place, screens alone will not do it.

Treating modelled impressions as a result. Use them to plan and compare. Do not report them as achievement without saying what they are.

Contrived triggers. Weather-reactive creative that has nothing to do with the product is a gimmick that costs setup time and buys nothing. Use triggers where relevance genuinely changes.

One creative file everywhere. Versioning by venue type and daypart is close to free and routinely skipped.

Long complex messages on shared screens. A few seconds of peripheral attention supports a mark, a short line, and nothing else.

Thin coverage across ten markets. This applies to both sides. Ten cities lightly covered produce nothing anywhere. Three markets at real density produce recall.

Assuming screen inventory sits where your audience lives. Check the actual screen map against the actual neighbourhoods before assuming the targeting solves it.

How to brief the two layers together

Decide what each layer is for before anything is bought.

Programmatic should carry the broad message, the responsive message and the geographic breadth. Set it by venue type and daypart rather than by raw reach, version the creative, and use triggers only where they genuinely change relevance.

The physical layer should carry depth in a defined area. Pick the corridors deliberately. Run for weeks rather than days, because frequency on a surface is the point. Keep the creative simple enough to work at walking distance and at a glance.

Then measure them differently, because they do different jobs. Programmatic gets modelled delivery and whatever attribution methodology you trust. The street layer gets counted evidence: geotagged install and removal photographs, site lists, and codes unique to each corridor where a response mechanic exists.

Campaigns that measure both with the same yardstick always conclude the wrong thing.

Frequently asked questions

Request a campaign quote

If you are planning a programmatic buy and want an honest read on what it will not cover, tell us the markets, the dates and the audience. We will tell you which neighbourhoods the screen map misses and what the physical layer should be doing alongside it.

Request a campaign quote and we will come back with corridors, crew numbers, install windows and a removal schedule.