46% of B2B companies are planning hybrid agency-plus-in-house models in 2026, agencies report 4.8:1 ROI versus 3.2:1 for in-house teams in like-for-like comparisons, and AI tooling has compressed what one in-house marketer can do across paid, content, lifecycle, and reporting. Most CMOs are still allocating in-house vs. agency by tradition — whatever was outsourced two years ago stays outsourced, whatever was built in-house stays in-house, and next year’s team plan is a hiring exercise. The question that produces better outcomes isn’t “should this function be in-house or agency.” It’s “what should this function look like in 2026, given AI compression, given the work that compounds in value over years, and given the work that has become commoditized.”
TL;DR
- The agency-vs-in-house framing is a remnant of an earlier era where the choice was binary. The 2026 framing is three-part: which functions belong in-house because they require institutional knowledge that compounds, which should be outsourced because they’ve been commoditized by AI, and which should be eliminated entirely because the underlying work has stopped justifying itself.
- AI tools have meaningfully compressed what one operator can do. The implication isn’t “smaller teams.” It’s “different shape of team”: fewer specialist generalists, more GTM engineers, more strategic positioning roles, fewer execution layers.
- The GTM engineering role is the highest-leverage marketing hire available in 2026. One or two people who can build, wire, and operate AI-driven workflows across the marketing stack often replace three to five traditional specialists.
- The diagnostic that a team plan reflects 2026 economics: the resulting org chart looks meaningfully different from 2022, with new roles, fewer roles in some categories, and a clear thesis about what each function is for.
The Three-Part Decision Framework
Each marketing function should be evaluated against three questions, not one.
Question 1: Does this function require institutional knowledge that compounds over years?
Functions like positioning leadership, brand strategy, customer marketing, analyst relations, and pricing partnership all require depth in the company’s history, customers, market position, and internal dynamics. The first six months of a hire in these roles produces less value than the second twelve months. These functions belong in-house and benefit from tenure.
Question 2: Has the underlying work been commoditized by AI tools?
Functions like first-draft content production, paid-media campaign operation, lead scoring administration, and basic analytics have been substantially compressed by AI tooling. The work still needs to happen, but the headcount required to do it has fallen — and the work is increasingly indistinguishable across providers. These functions are candidates for either AI-driven automation or agency consolidation.
Question 3: Has the underlying need disappeared?
Some marketing work has become unnecessary, not just commoditized. Functions like extensive MQL processing, traditional cold-list email marketing, generic top-of-funnel awareness campaigns at certain ARR thresholds, and many forms of mid-funnel “nurturing” produce less return than the cost of operating them. Before deciding in-house vs. agency, ask whether the work should be done at all.
Where Each Function Should Live (Most of the Time)
The decision framework produces a pattern at most $50M+ B2B SaaS companies that looks roughly like this.
In-house, full-time:
- Positioning and messaging leadership (Director or VP level)
- Brand and creative direction (Director level, supported by AI tooling)
- Customer marketing — activation, expansion, advocacy
- Analyst relations and influencer relations
- Marketing operations and data architecture
- Demand-gen leadership (not execution)
- GTM engineering (one or two people)
- Field marketing operations and event execution
- Product marketing
Hybrid or agency-supported:
- Paid media execution (in-house strategy, agency or AI-augmented execution)
- Content production at scale (in-house editorial direction, agency or AI for volume)
- SEO and content strategy (hybrid; AI tools compressing what’s needed)
- Video and creative production (project-based agencies remain efficient)
- PR and communications (agency for media relations, in-house for narrative)
Likely to be eliminated or radically compressed:
- Junior content marketer (compressed by AI tooling)
- SDR manager (compressed by AI SDRs)
- Junior marketing analyst (compressed by AI analytics)
- Marketing automation administrator (compressed by warehouse-native architecture)
- Specialist generalists doing thin coverage of multiple channels
Why GTM Engineering Is the New High-Leverage Hire
The GTM engineering role didn’t exist as a defined position three years ago. By early 2026, LinkedIn lists 3,000+ open roles with that title. OpenAI pays around $250K for senior GTM engineers; Ramp pays around $184K. The salaries reflect the leverage.
A GTM engineer is someone who can:
- Build and operate workflows that span the data warehouse, CRM, marketing automation, AI tools, and ad platforms
- Design and wire AI agents for specific marketing functions (content production, lead scoring, campaign reporting, lifecycle messaging)
- Operate reverse ETL pipelines, identity resolution, and segmentation in the warehouse
- Write SQL, work with dbt or similar transformation tools, manage API integrations
- Translate marketing strategy into automation that runs at scale
One GTM engineer typically replaces three to five operators who previously each managed a slice of the same stack. The function shape is different — not “marketing ops manager + content manager + paid media manager + email manager” but “GTM engineer + editor + brand director” — and the resulting team is structurally faster.
The mistake most CMOs make is hiring the GTM engineer as a senior individual contributor with no clear charter. The role works best with explicit ownership of the marketing automation stack, AI workflow design, and the marketing data architecture. Without those ownership lines, the GTM engineer becomes a senior expensive operator instead of a force multiplier.
What Most Companies Get Wrong
The most common pattern that produces under-leveraged team structures:
Hiring the 2022 org chart with 2026 budgets. A demand-gen manager, a content marketing manager, a marketing-ops administrator, a paid-media specialist, a lifecycle marketer. Five hires, none of them GTM engineers, none of them positioning leadership at the right altitude. The function operates at 2022 productivity with 2026 cost.
Outsourcing the wrong layer. Companies often outsource the strategic work (brand positioning to an agency) and keep the executable work in-house (content production headcount). The pattern should be inverted: keep positioning and strategy in-house where institutional knowledge compounds, outsource or automate the execution where it doesn’t.
Treating agency vs. in-house as a binary choice. The hybrid model is usually right, but the boundaries matter. Agencies that are accountable for outcomes (pipeline, MQLs, brand lift) tend to produce 4.8:1 ROI. Agencies that are accountable for outputs (content produced, ads run) tend to produce 2–3:1 ROI. The contract structure matters as much as the in-house vs. agency split.
Not eliminating work that doesn’t need to be done. The third question in the framework — should this work exist at all — is the question most companies skip. The marketing function carries activities, programs, and headcount that survived multiple reorgs without anyone asking whether the underlying work still produces return.
The Diagnostic for the CEO or CFO
The cleanest test of whether a marketing team plan reflects 2026 economics: look at the proposed org chart against the 2022 org chart. If the headcount has grown but the role mix is the same, the team is being built for the wrong economics. If new roles appear (GTM engineering, AI workflow design, customer marketing leadership), old roles disappear (junior content manager, demand-gen specialist, marketing-ops administrator), and the strategic-vs-execution ratio has shifted toward strategy — the team is being built for 2026.
A secondary test: ask the CMO which functions they’d prioritize if forced to cut 20% of headcount. CMOs who can answer with specific roles and a clear thesis are operating with conviction about function design. CMOs who can’t are operating with role inertia, which produces the same team plan year after year regardless of how much the underlying economics have changed.
The Bottom Line
The agency-vs-in-house question is the wrong frame. The right frame is function design — which work compounds in value over years, which has been commoditized by AI, and which has stopped justifying itself entirely. Marketing functions built around this framework look meaningfully different from those built two or three years ago: more positioning leadership, more GTM engineering, less execution headcount, and a clearer thesis about what each role is for. The teams that complete the redesign produce structural leverage competitors with traditional org structures cannot easily match. The teams that don’t, keep adjusting last year’s plan by ±10% and wondering why the productivity gains they expected from AI never showed up at the bottom line.
Additional Resources
From the Zaitz Marketing Knowledge Library:
- AI Agent Workflow Redesign vs. Bolt-On — The workflow-level companion to the role-level redesign described here
- The Defensible Martech Stack: Warehouse-Native, Not Best-of-Breed — Why GTM engineering depends on warehouse-native data architecture
- How AI Built Our Content Engine — A worked example of what the redesigned content function looks like
Want a second read on your measurement setup?
Start with a Growth Architecture Review. We will map your channel mix, audit your attribution, and show you where the real leverage is.
Book a Conversation →