Agencies keep clients from moving work in-house by fixing the account-ownership gap that drives the decision: smaller client rosters, one named lead who fully owns each account, and services an internal team cannot easily replicate. The Association of National Advertisers found 82% of member companies now run an in-house agency function, up from 78% in 2018 and 58% in 2013, according to ANA's 2023 report on the trend. Agencies that ignore the pattern lose their most profitable accounts to internal teams that cost less and know the brand better.
How many clients have already moved work in-house?
Most of them, at least partially. The ANA's 2023 study, "The Continued Rise of the In-House Agency," found that 82 percent of client-side marketers now operate some in-house agency function, and 65 percent had moved established business away from an external agency within the prior three years. Creative for digital media (social, search, email) and media strategy were the services most frequently pulled in-house.
The same study found a KPI shift that matters more than the headline number: cost savings dropped as the top reason to run an in-house shop, from 69 percent of respondents in 2018 to 62 percent in 2023, while "business performance" as a justification rose from 45 percent to 59 percent over the same period. Clients are no longer bringing work in-house only to cut costs. They are bringing it in-house because they believe the in-house team performs better.
Why do agencies actually lose accounts to in-housing?
The ANA's own respondents still work with external agencies at a 92 percent rate, so in-housing is rarely a full replacement. What agencies lose is the work that a distracted account team stopped doing well. Account managers on marketing and PR retainers are commonly assigned several clients at once, splitting time across social calendars, blog production, and asset creation for an entire roster. That workload makes it difficult for any one person to understand a single client's product, audience, and voice well enough to produce work the client could not have produced faster in-house.
The result is a specific client complaint: the deliverable reads like it came from someone who does not know the brand. That complaint, repeated enough times, is what turns "should we have an agency" into "why do we have an agency."
What can agencies do to keep clients from moving in-house?
Cut the roster before the client cuts the retainer. An account lead who owns four clients instead of twelve can learn each client's product, competitive set, and internal politics well enough to produce work an in-house generalist cannot match on quality, even if the in-house team is faster on turnaround. That requires charging more per account and hiring more staff relative to revenue, which is a real cost. It is smaller than the cost of losing the account outright.
Move the account team's output toward the categories the ANA data shows clients are not moving in-house: specialist creative, brand strategy work, and multi-market campaigns that require senior talent a company will not hire full-time for occasional use. Notion, Figma, Linear, Vercel, and Anthropic have each built substantial in-house communications functions and still route the specialist and surge work externally. The agencies that survived that shift did it by becoming the specialist the in-house team calls, not by competing with the in-house team on the volume work it now owns.
Does the 92 percent external-agency number mean agencies are safe?
No. It means agencies are safe doing different work than they did five years ago. The hybrid model, in-house core plus external specialist support, is now the default operating structure at large advertisers, and it rewards agencies that can prove a specific, senior capability the client cannot replicate at the price of one internal hire. It penalizes agencies still pitching the volume execution work that ANA's data shows is the first thing clients bring in-house.
The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.
