Marketing is in its third major push to bring work in-house. This time the reason won't surprise anyone. It's AI.
The earlier waves are worth remembering. Brands brought work in-house to save money, or because they had lost trust in their data and their media partners. The share of Association of National Advertisers members with in-house agencies rose from 42% in 2008 to 78% by 2018.
Then the hidden costs arrived. Some talent drifted back out. In-house teams often never got a seat at the leadership table. And the true cost of the software and the staff turned out higher than the agency's hourly rate had suggested.
The question from leadership is different now. It's no longer whether you can afford an agency. It's whether all this AI efficiency is producing better marketing.
The evidence so far is thin. In Duke University's 2026 CMO Survey, no marketing technology activity scored above 5 on a seven-point scale, and that included generating a return from the technology (as reported by MarTech).
Efficiency isn't what the board was promised. Results are. So I wouldn't decide everything at once. I'd run a pilot.
Here's one model:
Choose one campaign. Before you begin, write down what will count as proof:
Then compare those numbers with how you work today.
This approach depends on your visual identity, the look and feel of the campaign, and how clear your brand guidelines are. I've used a good many AI tools for this kind of work. Some are very good. Some aren't great at anything. That's exactly why you pilot.
Moving work in-house isn't a mistake. Moving it without a way to prove it's working is how the old mistakes get repeated.
This wave is driven by AI, which promises faster and cheaper production of marketing content. Earlier waves were driven by cost savings and loss of trust in data and media partners.
Hidden costs. Some talent left, in-house teams often lacked a seat at the leadership table, and the true cost of software and staff was higher than expected compared with agency rates.
Pick one campaign. Let an outside agency build the creative toolkit, have the in-house team own and approve it, and use AI for versions and adaptations. Define success measures in advance and compare the results with your current process.
Track the time and cost of each version, how often work is sent back for brand reasons, and whether campaign results hold up compared with your current way of working.