In-housing has been the slow-moving story of the last decade, and it's accelerated to the point where most large advertisers now run at least some marketing function internally. The reasons are partly rational economics, partly trust erosion, and partly that the technology finally made it possible. Worth unpacking each.
The first driver is cost arithmetic that stopped making sense. When agencies were charging blended rates of $200-400 an hour for work that involved a lot of intermediate handoffs, clients eventually noticed they could hire two senior people in-house for what they were paying the agency for one FTE-equivalent of effort. This was always somewhat true, but it became impossible to ignore once procurement teams started benchmarking. The agency justification used to be "you're not just paying for labour, you're paying for craft, perspective, and the ability to scale up and down", and that's still real, but it's a harder sell when the day-to-day work is reformatting assets and updating campaign trackers.
The second driver is the rise of always-on marketing. The old agency model was built around campaigns: a discrete brief, a discrete deliverable, a launch, and a debrief. Modern digital marketing isn't shaped like that. Performance ads need daily optimisation, social channels need daily content, e-commerce listings need constant updating, and customer lifecycle programs run continuously. Paying an external agency on a project basis to do work that has no end and no edges is a structural mismatch, and the friction (briefing meetings, approval cycles, retainers that don't flex) makes it slow and expensive. Bringing it in-house removes the seams.
The third driver, and probably the most underrated, is data and martech ownership. Five years ago, the agency held the keys: they ran the DSP seats, owned the analytics tags, configured the ad accounts, built the dashboards. Today, the client increasingly insists on owning all of that directly, because the data is the actual strategic asset. When your CDP, your ad accounts, your measurement infrastructure, and your customer data live inside agency systems, you're in a hostage situation, switching agencies means rebuilding everything. Once a CMO has been through one painful agency transition, they tend to insist that next time the infrastructure stays put.
The fourth driver is trust erosion around media buying specifically. The 2016 ANA report on media transparency was the watershed moment, it documented that agencies were taking undisclosed rebates from media owners on top of client fees, and the trust never fully recovered. Add the ongoing concerns about ad fraud, opaque programmatic supply chains, and the gap between what was promised and what was delivered, and a lot of clients concluded that the only way to actually know where their media money went was to put their own people on the seat. Some of the largest advertisers now run their entire programmatic operation in-house specifically because they no longer trust that an agency's incentives are aligned with theirs.
The fifth driver is speed and proximity to the business. An in-house team sits in stand-ups with the product team, hears about the new feature launch on day one, knows which SKU is overstocked, and can ship a creative response in 48 hours. An agency hears about it in next Tuesday's status meeting, books a brief workshop for the following week, and ships in three. For brands competing on cultural relevance, where being early to a meme or news moment is the entire point, that latency is fatal. The agency model was built for a world where marketing cycles were measured in months, not hours.
The sixth driver is talent preferences shifted. A decade ago, the prestige career path for a young creative or strategist was holding-company agency → bigger holding-company agency → eventually some partnership. Today, a lot of that talent would rather work in-house at a brand they actually use, with better hours, equity that might be worth something, and the ability to see their work all the way through to business outcomes rather than handing it off at launch. This is a vicious cycle for agencies: when the best young talent goes to brands instead, the agency's capability gap widens, which gives clients more reason to in-house, which drains more talent.
The seventh driver, quietly the most strategic, is that AI tools made in-housing actually feasible for things that used to require an agency at minimum scale. A small in-house team of five with a good AI stack can now do work that used to require an agency team of twenty. The argument that "you need an agency because you can't justify hiring full-time specialists for every craft" weakens when one generalist plus the right tools can produce competent first drafts of copy, design, video, and analysis. The economic threshold for in-housing dropped sharply in the last 24 months.
It's worth being honest about what gets in-housed and what doesn't, because the picture is more nuanced than "clients are killing agencies." The work that moves in-house is usually: performance media, social content, lifecycle marketing, CRM, basic creative production, web and e-commerce, and analytics. The work that tends to stay with agencies is: brand strategy and positioning, big-platform creative ideas, integrated campaigns that need a fresh outside perspective, specialised capabilities the client can't justify hiring full-time (sonic branding, packaging design, deep cultural research), and crisis or reputation work where outside counsel is the entire point. The pattern is roughly: high-frequency operational work goes in, lower-frequency strategic and craft work stays out.
There's also an unspoken counter-trend worth naming: in-housing has its own failure modes, and some of the brands that aggressively in-housed five years ago are quietly bringing work back out. Internal teams get stale (you stop seeing work from outside your category, your reference set narrows). Hiring senior creative talent into a brand is hard because the best people want variety. Internal politics distort the work in ways an external partner is insulated from. And when the in-house team becomes a cost centre under a CFO's microscope, it gets cut first in downturns, at which point the brand has neither the internal capability nor the agency relationship to fall back on. The smart brands run a hybrid: in-house for the always-on operational layer, agencies for the things that benefit from outside provocation and specialised craft.
The strategic implication for agencies is uncomfortable but clear. Competing for the operational work is a losing game, clients can always do it cheaper themselves once they decide to. The agencies that remain valuable are the ones that do work clients can't easily replicate internally: genuinely original strategic thinking, distinctive creative craft, access to specialised talent and culture, and the kind of outside perspective that's only valuable because you don't sit in the building every day. The agencies that lost the most to in-housing are the ones that had quietly become extensions of the client's marketing department, present at every meeting, executing every request, indistinguishable from staff except more expensive. That role has no future.
