How does AI affect the different types of marketing agency? Unevenly. It goes first for the work in each discipline that is repetitive and data-heavy, which is a different set of tasks in a media shop than in a PR firm. And because small and large agencies now use the same tools, it rewards the specialists who know things the tools do not.
Nita Song, president of IW Group, made that second point sharply in Marketing Dive. AI, she argues, is the first big technology shift that does not reward agency size: a 10-person boutique and a 1,000-person holding company shop can use the same platforms. In her words, "The tools don't know the difference."
Does AI favour big agencies or small ones?
For most of advertising's history, scale won. Bigger production budgets, bigger research departments, more room to absorb risk. Song's case is that AI flattens much of that. Research that once took months can now take days, for everyone.
Sociologists who study the digital divide spotted a similar pattern years ago. The first divide was about access: who had the technology at all. Once access spread, a second divide opened up around skills and use: who could actually get value from it. Agencies are crossing that line now. Access to AI is close to universal. The gap is in how well each agency uses it, and what it brings that the model has never seen.
Song's example is a good one. A large language model can summarise what Asian-American consumers bought last quarter. It cannot explain why a message lands differently with a second-generation Chinese-American household in Los Angeles than with a first-generation Vietnamese-American household in Houston. That knowledge lives in communities, not datasets.
How does AI change each type of marketing agency?
Here is our read, discipline by discipline. The pattern is consistent: AI takes the chores first, and the value moves to judgement.
| Agency type | What AI absorbs first | What becomes more valuable | The risk of standing still |
|---|---|---|---|
| Creative | Versioning, resizing, first drafts, mood boards | The idea, taste, and knowing which concept deserves media money | Being priced like a production line |
| Media and performance | Bid management, pacing, routine optimisation, reporting | Measurement, budget allocation, commercial negotiation | Clients running optimisation themselves |
| CRM | Segmentation, journey building, campaign deployment | Customer insight, offer strategy, data governance | Becoming a platform operator rather than an adviser |
| PR and comms | Media monitoring, coverage reports, first-draft releases | Relationships, narrative judgement, crisis calls | Monitoring fees quietly disappearing |
| Influencer | Creator search, performance reporting, contract admin | Creator relationships, cultural fit, creative direction | Margin squeezed by self-serve platforms |
| Social | Scheduling, community triage, listening summaries | Platform-native ideas and real-time cultural reading | Competing with in-house teams on volume |
Two of our own clients fit the pattern. SAMY, a global influencer agency, adopted Defy Agents to streamline performance reporting, cutting three days of work to one. Enchant, a UK CRM agency, adopted Defy Monitor to automate insight and campaign deployment for all its clients. In both cases the automation targets the most repetitive part of the discipline, not the judgement.
Sports marketing follows the same split. Versioning matchday content for every platform and sponsor is a chore. Reading a fanbase's mood after a derby defeat is not.
Why are specialist agencies cutting services instead of adding AI ones?
The instinct when a new technology arrives is to bolt on a new department. The evidence points the other way. Campaign US cited Promethean Research's 2026 report on digital agencies: those that cut services grew 13%, against 9.8% for those that expanded and 5.7% for those that held steady. And 86% of the agencies surveyed describe themselves as specialists.
The examples are instructive. Dept closed a native mobile development division after outside firms offered similar services at "half the cost". Spark shut its in-house web development and studio teams because it could not compete with niche specialists. Conscious Minds dropped its research arm and community management to concentrate on social and creative, adding only specialities that deepen that core, such as production, AI and sports.
Spark's founder, Michael Peters, expects agencies to treat AI work the same way and outsource it. That is a very different instinct from trying to build everything in-house.
How to choose AI tools for your type of agency
- List your ten most repeated tasks. Timesheets, monthly reports, status meetings, client amends, manual resizing, competitor audits. Be honest about the hours each one takes.
- Sort them into chores and judgement. Chores are rules-based and data-heavy. Judgement is anything a client would notice if it got worse.
- Decide what is core. If a task sits outside your speciality, consider partnering or dropping it before you automate it.
- Pick tool categories by discipline. Media teams should look at measurement and optimisation; creative teams at versioning and pre-scoring; CRM at segmentation and deployment; PR and social at listening; influencer at reporting.
- Pilot on one client with a baseline. Write down the hours and the quality today, run the tool for a month, then compare.
- Reinvest the time in what tools cannot know. Cultural knowledge, relationships and taste are where your specialism lives.
Why does innovation come from connection, not competition?
If AI flattens access to capability, the old network advantage of owning every discipline under one roof weakens. The more interesting model is connection: specialist independents that partner with each other, share tools and pass work to whoever does it best. A CRM agency and a creative shop that trust each other can assemble an offer that looks a lot like a network's, without the overhead.
Innovation thrives in connection, not competition. Song makes a related point: multicultural consumers often start trends that later reach mainstream marketing. That knowledge only travels if the specialists who hold it are connected to the brands and agencies that need it.
What does this mean for independent agencies?
The playing field is flatter than it has ever been, but only if you have the capability to play on it. The networks have spent heavily on analytics, AI and intelligence stacks. The Defy platform gives independents the same analytical and AI capability without the build cost or the maintenance burden, so a specialist can spend its energy on the specialism.
The tools don't know the difference between a boutique and a network. Make sure your clients do.
