Programmatic and Display Advertising
Programmatic advertising agency, in-house or AI: what pays off for you?
Programmatic has the highest entry threshold: access to a buying platform, its running costs and the required knowledge only pay off from a certain media volume. Below it, in-house is rarely sensible, not because the team could not do it, but because the fixed effort does not spread. Above it, the question flips: then it is about data ownership, transparency in buying, and who pays the middlemen. AI takes over monitoring and evaluation here, not the buying; that runs automated anyway.
What we do: We run buying, placement checks and contact distribution on a platform access in your name and disclose every cost layer; you get control without the fixed costs of your own operation.
By Svenja Müller, Strategic Team Lead Programmatic
Last updated on
The decision in 30 seconds
Small to medium media volume in the channel
Agency with AI. The fixed effort for platform access, tools and knowledge spreads across many accounts with us, across one with you.
Large, permanent volume and a desire for full transparency
At this size, control over buying and data weighs heavily. Agency with AI plugs into exactly that: platform access and audiences run in your name, we steer on top and disclose every cost layer.
Own platform access exists, but the evaluation lags behind
Monitoring, quality checks and evaluation run automated; the steering stays with you. Agency with AI plugs into exactly that: we run the monitoring on your access, decisions stay at your end.
Demanding audiences and measurement with limited internal capacity
Agency with AI. Platform competence and automated monitoring together. That is our own model.
The four models compared
Every cell names a model's strength and its limit. The rating in front says which of the two prevails. There is no points total and no overall winner: both would feign a precision that does not exist here.
StrengthNeutralWeaknessStrength in our model
| Criterion | In-house | Agency | In-house with AI | Our modelAgency with AI |
|---|---|---|---|---|
| Product and business understandingHow well someone knows which environment your brand should stand in, and which it must not. | Strength You know which topics are sensitive for your brand, before anyone outside notices. That list is written down completely nowhere; it lives in the house. What is missing is the overview of which environments are even available for buying. | Neutral From many campaigns it is known which environments hold what they promise. But what harms your brand is a question only your house can answer, and a generic exclusion list rarely hits it exactly. | Strength The same closeness, and exclusions can be tracked automatically instead of being maintained by hand every month. Only what was once named gets tracked automatically: the sensitive list in people's heads stays manual work. | Neutral We take your exclusions as the baseline and track them automatically, and we bring in which environments are actually available for buying. What is sensitive is your call. What is buyable is our knowledge. |
| Platform competenceCommand of the buying platform, its settings and its quirks. | Weakness Buildable, but expensive: access, training and continuous practice only pay off at permanently high volume. Whoever has built it is independent. | Strength The usual reason this channel gets outsourced. Access, contracts and practice are in place; the effort spreads across many accounts. | Weakness Takes over routine, but replaces neither platform access nor onboarding. AI lowers the entry threshold less here than in other channels. | Strength Specialists plus automated monitoring. In practice, misconfigurations and anomalies surface faster that way. |
| Data and audiencesWhich data drives the targeting, and who owns it. | Strength The strongest reason for this model. Your own customer data stays in-house, audiences are built first-hand, and the insights stay with you through any switch. | Neutral Brings experience and access to data partners. With your own customer data it takes clear agreements, otherwise a dependency emerges that only shows when you switch. | Neutral Value models and lookalike groups can be computed from your own data instead of using bought segments. That takes maintained data paths. | Neutral Models on your data, run with external experience. Who owns the audiences and results that get built is put in writing before the start, and in your favor. |
| Media buying and transparencyHow much of the budget actually arrives as advertising delivered. | Strength The chain is short and traceable, every intermediate layer visible. In return, framework contracts are missing, and with them the terms that only volume produces. | Neutral Better terms through volume, and it knows which intermediate layers are common. The flip side is the transparency question, which only contractual disclosure settles. | Neutral Supply-chain evaluations make visible where money gets stuck. That requires access to the log data. | Neutral Automated evaluation plus someone who can name the intermediate layers. We disclose the complete cost breakdown without you having to ask for it. |
| Brand safetyIn which environments ads appear, and what must not appear there. | Neutral Nobody knows the house's own limits better, and edge cases are decided by someone with brand responsibility. The upkeep of exclusion lists falls asleep in the daily routine. | Neutral Maintains lists and tools on schedule and sees critical environments earlier, because it watches many accounts. The brand's finer points it has to learn. | Neutral Environments can be assessed by content instead of only by word lists, which reduces unnecessary exclusions. Drawing the line remains a decision, not a calculation. | Strength Content-level assessment automated, edge cases with people. In news environments this difference is most visible. |
| Inventory qualityWhether ads appear where they can actually be seen. | Neutral Full control over buying lists and the option to build lasting relationships with good environments. That costs time, which is scarce in the daily routine. | Neutral Knows environments from many accounts and spots low-grade reach faster. Without a clear agreement, cheap buying becomes a quiet side calculation. | Neutral Anomalies in viewability, invalid traffic and recurring patterns surface automatically. Someone still has to react. | Strength Permanent checks plus experience. In this channel one of the points where the most money can be saved. |
| Frequency and pacingHow often the same person sees the same ad, across campaigns and formats. | Neutral Well controllable across your own campaigns, because everything sits in one hand. As soon as other channels join, the shared view is missing. | Neutral Thinks across channels when it also steers the others. If it does not, the same gap emerges as internally. | Strength Patterns in the contact distribution surface early, say when a small part of the audience gets most of the contacts. By hand that is barely visible. | Strength Automated detection plus intervention by people. Excessive frequency is the most common silent loss in this channel. |
| Measurement and attributionWhether it can be proven that the ads made a difference. | Neutral The connection to internal business numbers is there, and real impact tests can be set up without a conflict of interest. The methodological experience for it is often missing. | Weakness Brings methodological knowledge, but sits in a conflict of interest when judging its own performance. An independent measurement basis matters more here than in any other channel. | Neutral Evaluations and control groups can be created and repeated automatically. Designing the test remains the actual work. | Neutral Automated evaluation, test design together, and we ourselves insist on being judged against a number that does not come from our own report. |
| Interplay with other channelsWhether programmatic feeds the same goals as search, social and affiliate. | Neutral When all channels sit in-house, that is the big advantage: one view, one budget, one decision. If only this channel sits internally, the opposite emerges. | Neutral If it steers several channels, it shifts budget by impact instead of by responsibility. If it steers only this one, it optimizes its own slice. | Weakness Cross-channel evaluations can be produced automatically, provided the data comes together in one place. That is exactly where it usually fails. | Strength Shared data foundation plus automated evaluation. Which is why we steer programmatic together with search and social instead of next to them. |
What can AI take over today?
Check environments and supply paths permanently
Suspicious vendors, invalid traffic and weak viewability surface automatically from log data, instead of drowning in a spot check.
Evaluate the contact distribution
Recognize how contacts spread across the audience and where the same people are reached too often. By hand that cannot be done.
Build audiences from your own data
Compute value models and lookalike groups from existing customer data, instead of using bought segments.
Classify environments by content
Rate pages by content instead of by word lists. That prevents whole news environments from being excluded over a single term.
Translate creatives into format families
Generate banner formats in all required sizes and language versions from one approved creative.
Consolidate evaluations
Bring numbers from buying platform, measurement vendor and web analytics together and name deviations before anyone goes looking.
What should AI not decide on its own?
Choose the buying platform
Which platform, which contracts and which operating model fit is a strategic decision with a long commitment and costs that accrue regardless of usage.
Negotiate
Terms, disclosure of costs and access to raw data are negotiated between people. These points decide how much of the budget arrives as advertising delivered.
Draw the brand's lines
In which environment a brand does not want to appear is a question of stance. A content-level classification prepares the case; someone with brand responsibility has to decide.
Judge its own impact
Attribution is difficult here, and any model can be built to look good. Designing the measurement and choosing the benchmark belong in hands that do not need a particular result.
Decide the channel's role
Whether programmatic should build reach, harvest demand or open a market follows from the business. Without that direction, every automation optimizes for the most convenient metric.
Typical situations and what we recommend
Small team without platform experience
Programmatic is supposed to be added, but there is neither platform access nor anyone who has worked with it.
RecommendationAgency with AI. At small volume the fixed effort does not pay off internally; with us it spreads across many accounts. Customer data and the audiences that get built run in your name from the start.
Large company with its own performance team
Several channels run internally, the volume is permanently high, and the question of transparency in buying is on the table.
RecommendationAgency with AI as the existing team's partner, step by step. Start with the part where your own data makes the biggest difference; we keep steering the rest until it is cleanly handed over. A switch in one step is risky in this channel.
High budget with few internal resources
A lot is being invested, but in-house nobody reads along daily, and the reports come from the party that also spends the money.
RecommendationAgency with AI, with a measurement basis that does not come from the buying side alone. Automated monitoring of inventory quality and contact distribution is the most effective safeguard here, and we disclose every cost layer.
Brand campaign with high demands on the environment
Reach matters, but the environment matters more, and a misstep would be public.
RecommendationAgency with AI. Your brand side draws the lines; implementation and continuous checking are ours. Automated content-level classification protects reach instead of cutting it across the board.
Our solution
Programmatic is the channel where the model question hangs most on size. We ourselves work with specialists and automated monitoring side by side, and offer three routes.
We take over completely
Platform, buying, audiences, brand safety and measurement sit with us; goals and budget with you. That fits below the volume threshold at which your own setup pays off.
We work with your team
You run the platform; we add specialist knowledge, quality checks and measurement design. That fits while building your own operation and in the transition period after.
We automate your workflows
Inventory quality checks, contact distribution evaluation and consolidation of the numbers run automated at your end. That fits when the steering is solid and the checking eats time.
What we build to automate this control work is described at hurra.ai: permanent checks of environments and supply paths, evaluation of the contact distribution, consolidation of numbers from several systems. It replaces no steering; it makes visible what otherwise goes unnoticed.
Go to hurra.aiEverything on this channel at hurra.com: Display advertising at hurra.com.
Frequently asked questions
From which budget does programmatic in-house pay off?
The threshold follows from the fixed effort, meaning platform costs, tools, staff and onboarding, divided by your media volume. Any general number would be a guess; platform costs and staffing differ too much for that. With us, that fixed effort spreads across many accounts, and platform access still runs in your name: you get the control without the fixed costs.
Can AI steer programmatic campaigns on its own?
The buying runs automated anyway; that is the nature of the channel. AI additionally contributes to monitoring, quality checks and evaluation. Platform, budget frame, brand environment and measurement remain decisions made by people.
How do I ensure transparency in media buying?
Through three things: contractual disclosure of all cost layers, access to the buying platform's log data, and a measurement basis that does not come from the buying side alone. That applies to every agency, ours included.
Who owns the audience data when an agency buys?
That should be settled contractually before the first campaign starts. The audiences and insights that get built are the most valuable part of this channel; if they sit with the vendor, a switch is not a switch but a fresh start.
What is the difference between display and programmatic?
Display describes the ad form, meaning banners and other ad spaces. Programmatic describes the buying: automated and per contact. Programmatic today also covers video, audio and digital out-of-home, so it is the broader description.
How do I measure whether programmatic works?
Through control groups instead of last-touch attribution. Regional tests or deliberately holding out audiences show the difference the channel actually makes. The setup costs preparation, but it is the only route to a reliable statement.
And if the question is the agency after all
If the decision for an agency is already made and it is only about the choice, the agency comparison overview puts hurra.com next to the agencies we regularly compete against, with a source on every statement.
Content and images on this page may have been created with AI support.