What changed
In-housing stopped being a budget decision and became an org-chart decision. The ANA published its 2026 State of In-Housing report, The Resilient Rise of the In-House Agency, in June 2026. It surveyed 404 jurors of the 2026 ANA In-House Excellence Awards — brand-side marketers, agency professionals and consultants who work with in-house teams day to day. The jurors were five times more likely to say marketers are in-housing more than ever than to say marketers are pulling back: 35% agreed with the first statement, 7% with the second.
The reason people give has flipped. In 2023, 30% of respondents named cost efficiency as the primary benefit of an in-house agency. In the 2026 read, only 9% did. Instead, 53% said the in-house agency's primary role is strategic partner — upstream strategy, brand-building, the big creative idea. Adweek framed the same shift in its coverage: what started as a cost-cutting move is now pitched internally as a strategic play.
Scope is moving in one direction. 34% of respondents saw work migrating from external partners into the in-house team; 7% saw it moving back out. And the function most often handled in-house is the one this report cares about: social, at 58%, ahead of influencer and creator marketing at 37% and SEO at 35%. Two-thirds — 67% — of responding in-house agencies now handle media in some form.
The money is visible in the budget line
Gartner's 2026 CMO Spend Survey, reported by MarTech, shows in-house labor taking a larger share of the marketing budget: 24.5% in 2026, up from 21.9% in 2025. That is not a rounding error at any meaningful spend level. The same survey found awareness and conversion together absorbing 62.6% of media spend, with digital channels taking more than two-thirds of the total, while loyalty and retention spend fell 29% since 2024 to under 15% of media.
The platform did half the work
The operator's job on paid social changed underneath all of this. Apple's App Tracking Transparency removed roughly 70–80% of iOS users from deterministic tracking, with opt-in rates settling around 20–30%, according to Superscale's 2026 review of the category. The platforms answered with machine-learning ad serving — Meta shipped Andromeda, a transformer-based retrieval system, in late 2024 — that reads creative-level engagement instead of clean conversion signals. The practical consequence: the levers that used to justify an expensive external buyer (audience splits, bid strategy, ad-set architecture) got consolidated into the algorithm, and the lever that matters now is creative volume and diversity in 9:16 vertical video.
That is the quiet mechanism behind the in-housing numbers. Buying got easier to do in-house. Creative supply got harder.
Why it matters to a brand at this spend level
At $50K/month in paid social, the arithmetic is uncomfortably close. Published 2026 agency fee bands put paid media management at 10–20% of ad spend or a flat retainer of roughly $2,500–$15,000/month; The Remarkable Agency and Ad Snipper both report those same ranges, and The Remarkable cites WebFX's 2026 PPC pricing data for the 10–20% figure, noting lower rates at larger budgets. On a $50K monthly budget, 10–20% is $5,000–$10,000/month, or $60,000–$120,000 a year. (That last step is our arithmetic on their published bands, not a figure any source states.)
Against that, Ad Snipper puts a US in-house media buyer at $60,000–$95,000 base, $78,000–$124,000 fully loaded, and cites a Glassdoor average near $97,000; freelancers land at $50–$150/hour. Both Ad Snipper and Digimau sell services adjacent to these numbers, so treat them as market signal rather than audit.
So one senior in-house buyer costs about what the agency costs. The headcount swap is roughly neutral. Which is exactly why the ANA data shows cost efficiency collapsing from 30% to 9% as the stated reason — brands ran this math and found the savings were not the point.
What the swap actually buys and costs
| Line | Agency at $50K/mo spend | One in-house buyer |
|---|---|---|
| Direct cost | $5,000–$10,000/mo at 10–20%, or $2,500–$15,000 flat | $78,000–$124,000/yr loaded |
| Scales with spend | Yes, on percentage models | No |
| Creative production | Often in scope, often the first thing cut | Not included; separate cost |
| Coverage risk | Bench absorbs absence | Single point of failure |
| Tooling and measurement | Usually the agency's stack | You buy and run it |
The two costs that do not appear on either side of that table are the ones that decide the outcome. First, creative volume: if the algorithm now needs variety to find audience-ad matches, an in-house buyer without a production pipeline is a buyer with nothing to feed the machine. Second, measurement: once you no longer have an agency dashboard, someone on your side has to own server-side event quality, MER, and incrementality reads. Superscale's summary of 2026 practice is blunt about the direction — platform-reported numbers are directional inputs, not decisions.
The ANA respondents named their own top three threats in order: creative stagnation, talent burnout, and cost competitiveness as external agencies get leaner. The first two are what happens when you in-house buying without in-housing supply.
AI is filling the gap, but not closing it
On AI, the ANA read is modest: 49% said AI is letting the in-house team do more work while external spend stays the same, and only 19% said AI is letting them produce work that was previously outsourced. On creative development specifically, 18% reported a significant positive impact, 49% a moderate one, and 21% said it was too early to tell.
Gartner's survey points at why. Per MarTech's reporting, 70% of CMOs said their processes are not mature enough to scale AI, and 38% named a lack of internal AI expertise as the biggest barrier. The same survey found 49% of US consumers believe generative AI has made content quality worse, rising to 57% among Gen Z and millennials. Volume without quality control is a measurable risk, not a free win.
What to do this month
Basis Technologies' 2026 guide makes the framing point worth keeping: in-housing is not binary. It is a series of decisions about which specific capabilities sit inside, and hybrid arrangements are the norm, not the fallback. Three actions that follow from that.
1. Split your agency invoice into four lines and price each one
Most paid-social retainers bundle strategy, buying execution, creative production, and reporting into a single number. Ask for the split. Buying execution is the line the ANA data says brands are actually taking in-house — social is the most in-housed function at 58% — and it is the line the platforms automated hardest. Creative production and measurement are the lines where outside capacity still earns its fee. If the agency will not itemise, that is itself a data point. Note that The Remarkable Agency, one of our fee sources, discloses that it sells flat-fee paid media and prices a standalone account audit at $1,500–$5,000, which is a cheap way to get a second read before you restructure anything.
2. Count your monthly creative throughput before you move buying in-house
Write down how many distinct concepts — not variants, not crops — entered testing in each of the last three months, and how many were 9:16 vertical video. If that number is in the single digits, in-housing the buying seat will not improve performance, because the constraint is upstream. The ANA respondents ranked creative stagnation as their top threat and only 19% said AI had let them bring previously outsourced work in-house. Decide where concepts will come from before you decide who pushes the buttons.
3. Make the measurement handoff explicit and owned
With 70–80% of iOS users outside deterministic tracking, your server-side event quality is the thing keeping the algorithm calibrated. Before any scope change, name one person who owns the Conversions API feed, event deduplication, and the blended MER read — and confirm whether the pixel, the CAPI integration, the ad accounts, and the reporting warehouse are in your business manager or the agency's. This is a two-hour audit that prevents a two-month outage.
What we would watch next
- Whether the labor share keeps climbing. Gartner's 21.9% to 24.5% move is one year of data. A second consecutive rise in the 2027 survey would confirm a structural shift rather than a correction.
- Whether cost efficiency stays at 9%. If external agencies get leaner — the third threat ANA respondents named — the cost argument could swing back and pull work out again.
- Whether AI moves past the 19% line. The number to watch is the share of in-house teams saying AI let them produce work they previously outsourced. Today it is 19%. If that reaches 40%, creative production economics change and the hybrid split moves.
- Whether platform automation keeps absorbing the buying seat. More of the campaign setup surface moving into automated flows makes the in-house buyer role narrower and the creative role wider.
- Consumer tolerance for AI-made creative. 49% of US consumers already say generative AI made content quality worse. If that rises, volume-first creative strategies get expensive.
What we could not verify
- The full ANA report is behind a login. We read the ANA abstract page, IHALC's detailed summary, and Adweek's coverage. Every ANA percentage above comes from those write-ups, not from the source PDF.
- Adweek's article is partially paywalled; we read the opening section only.
- Ad Snipper and Digimau are vendor-published pricing guides selling adjacent services. Their salary and retainer bands are self-reported market observation, not audited data.
- The Remarkable Agency states in its own post that it sells flat-fee paid media, which is a direct interest in the flat-versus-percentage comparison.
- The Glassdoor salary average and WebFX PPC pricing data are cited second-hand through those posts; we did not open the originals.
- We attempted the IAB 2026 Outlook Study twice and got an empty PDF and a 404. No figure in this report comes from it.
- The ATT opt-in and 70–80% deterministic-loss figures come from Superscale, a practitioner source, not from Apple or Meta documentation. We did not open primary Meta material on Andromeda.
- No source we found publishes an in-house-versus-agency cost benchmark specific to paid social at the $50K/month tier. The $5,000–$10,000/month comparison is our arithmetic applied to published percentage bands.
- The ANA threat list order (creative stagnation, talent burnout, cost competitiveness) is reported as an ordered ranking in the secondary coverage; we could not see the underlying percentages.