Interactive advertising agencies, the three different shops hiding behind one phrase

"Interactive advertising agency" is a label with three different businesses inside it. One builds things: sites, apps, configurators, installations. One makes the advertising unit itself, the expandable, playable or shoppable creative. One buys digital media and reports on it. They have different cost structures, different fee shapes and different staff, and a buyer who does not pick one before the first meeting will be shown a credentials deck that blurs all three.

The three shops, side by side

If what you need is the first column, our city level guide to what an interactive agency builds and how it differs from an ad agency goes through the build scope, the fee shapes and the clauses line by line for that market. The rest of this page is about telling the three apart and briefing the second and third.

Why the format question decides the creative brief

Interactive creative is built to a published specification, and the specification has been moving away from fixed pixel sizes. The IAB's own page on its New Ad Portfolio, read on 22 October 2026, argues that "while many companies continue to produce traditional fixed-size ad formats, it is essential for brands, advertisers, publishers, and creative agencies to embrace the benefits of the IAB New Ad Portfolio", and that "flexible ad sizes improve delivery across diverse screen sizes and devices while future-proofing campaigns for emerging technologies such as augmented reality (AR) and virtual reality (VR)".

The practical consequence for a buyer is simple. Ask the studio whether it is delivering a fixed size set or a flexible set, and ask the media buyer which the inventory you are buying actually accepts. Those two answers are often given by two different suppliers who have never spoken, and the gap between them is where production money disappears.

How to brief each of the three

For a build, the brief is a product brief: who uses it, what they are trying to finish, which systems it has to talk to, and who owns the code and the design files at the end. Settle repository and design file ownership in writing before kickoff; it is the single most common unpleasant surprise.

For rich media creative, the brief is a specification plus a decision log. Name the platforms, name the formats, name the maximum file weights, and state who signs off the variants. Then write down who pays for a rebuild if the placement changes, because it will.

For media buying, the brief is a measurement brief before it is a budget. Name the conversion you care about, name the attribution model you will report on, and agree the fee shape. On the model, two platform facts are worth carrying into the conversation. Google Ads documentation, read on 22 October 2026, says "the first click, linear, time decay, and position-based attribution models are no longer supported by Google". Google Analytics documentation, read the same day, says there are three models available in its attribution reports: data driven attribution, paid and organic last click, and Google paid channels last click. An agency promising a bespoke multi touch split is promising its own model, which is allowed, and has to be labelled as theirs.

What nobody publishes about this market

There is no authoritative published count of interactive advertising agencies in the United States, and no independent registry of them. We looked for one to cite and there is none. Any list presented as the definitive set is a directory with an entry fee or an editor's judgement, and both are legitimate as long as they are described that way rather than as a census.

What is measurable is what an assistant tells a buyer who asks for one. Our own US reading on 22 October 2026 covered 635 scored readings on ChatGPT and Gemini, returning 19,793 citation URLs from 3,307 distinct hosts, with half of all citations coming from 51 hosts, and 1,053 namings across a roster of 24 vendor names counted once per reading. The host table and the question set are in the citation study; the naming side with its denominator is in the naming count. Citation and naming are two findings and are never added.

A disclosure, because we sell a measurement of part of what this page covers. AI Knows Us reads five AI surfaces and reports two separate findings, whether a business is named and whether its own pages are cited, and the two are never added. Our own reading on 22 October 2026 was named 0 and cited 0 across 635 readings, published as our own null result after 507 new pages. Where a rival is stronger, plainly: Profound publishes SSO and SAML, SOC 2 and up to nine engines, and our rendered pricing page carries none of those words and names five surfaces, not nine. Our cheapest plan that reads all five surfaces is Growth at 299 dollars a month in the US and the UK, and the 149 dollar SEO Start plan reads 0 questions, 0 credits and 0 engines, so it is not an AI visibility plan.

Questions buyers ask about interactive advertising agencies

Is an interactive agency the same as a digital agency

In everyday use the words have collapsed together, which is why the shortlist has to be built from what you want delivered rather than from the label. Ask for the last three pieces of work and look at what was actually shipped.

Should one firm do the build and the media

Splitting them gives you an unconflicted read on whether the media is working, because the firm reporting on performance is not the firm whose build is being judged. Keeping them together is faster and needs one internal owner of the numbers.

Who should own the creative files

You should, named in the contract, including source files and not only exports. The same applies to the repository on a build.

What fee shape is fairest for media buying

There is no universally fair shape, and each one biases behaviour. A percentage of spend rewards spending more. A flat retainer rewards doing less. A hybrid with a floor and a performance element is the usual compromise. Pick knowingly and write the bias into the review conversation.

When can we judge whether the engagement is working

Judge it on deliverables and measurements rather than elapsed time: assets delivered to spec and accepted, the share of placements that ran without a rebuild, cost per conversion on a named attribution model, and whether every rate in the report carries the count it was taken over.

The placement that cannot be bought

Every surface on this page can be bought. An assistant's answer to "who should I hire" cannot, because there is no auction to enter. It is assembled from sources, which makes it measurable on a recorded question set and not biddable at any budget.