What to fire first when you bring marketing in-house

Founders who bring marketing in-house in 2026 are making decisions that would have looked reckless three years ago. They fire the agency before they hire the marketer. They cancel the SEO consultant before a content engine is running. They let the LinkedIn ghostwriter go without a replacement.

It works because the math underneath has changed. AI tools let one operator cover what used to need three or four vendors. The line items that made sense in 2022 are line items a small company can absorb today.

Here are the four vendors most founder-marketers cut on day one, and why.

1. The blog ghostwriter or content agency

This is the easiest cut and the most expensive line item for most B2B SaaS companies running between five and thirty employees. A typical retainer for monthly blog content runs $3K to $8K. The output is correct, on-brand, and forgettable. It compounds slowly because it’s built around a calendar, not a strategy.

The replacement is a process more than a tool. Set up the terminal, give Claude Code a clear voice doc and a backlog of opinions, and a founder can ship better content than most agencies, faster. Lenny Rachitsky has written about what separates content that compounds from content that fills slots; the answer is almost always “the operator behind it has a real opinion.”

Cut the retainer, redirect about half of that budget into a marketing engineer hire or part-time consultant, and ship better work at half the cost.

2. The SEO consultant doing keyword research

SEO consultants charging $4K to $10K a month to “do keyword research and content briefs” have been on borrowed time since Search Console data became free and good. The job they bill for is now a Claude Code prompt. Pull GSC. Pull SERP. Identify gap queries. Write a brief. None of that needs a human in the loop anymore.

The exception is structured technical SEO on a large site (over 10K URLs). Most 5-30 person B2B SaaS sites have under 200 pages, and technical SEO there is a one-time cleanup, not a monthly bill.

What you keep is someone with strong instincts on positioning and topic selection. That’s not the SEO consultant. That’s the founder or the marketing engineer.

3. The performance-ads agency for early-stage spend

If you’re spending under $50K a month on paid, you are not getting your agency’s best people, and you almost certainly aren’t getting their best work. Tomasz Tunguz has written about this pattern for years: paid agencies need scale to justify margin, and small accounts get the junior planners.

Founder-marketers running ads in-house with a competent operator hit better ROAS than mid-tier agencies. The tooling has caught up. The tuning is mostly automatic. What’s left is creative judgment, and that’s the founder.

Cut the agency. Keep $10K of the savings to hire a paid specialist on a project basis when you actually scale spend.

4. The founder’s LinkedIn ghostwriter

If you’re paying $1K to $3K a month for someone to ghost-write your founder posts, the product is louder than the founder is. Founders pay for this because they think they don’t have time to write. What most founders actually lack is a system for capturing their own thinking.

A Granola transcript of every customer call, a 30-minute weekly pull session with an AI agent that knows their voice, and a publishing routine produces founder-led content that beats any ghostwriter’s output. Andrew Chen has written about why founder-led content compounds in ways agency content doesn’t; the short version is that real opinions move people and polished prose doesn’t.

The savings here are smaller. The brand benefit is real.

What survives the cut

Two things are still worth paying for.

One strong human partner per function. Not a vendor. An embedded operator. A fractional CMO with operator chops, a designer who knows your visual system, a video producer for high-stakes assets. Project-based, not retainer.

The tooling stack. Anthropic, Granola or Otter, the analytics tools, the design tools, the project board. Total under $500 a month for most teams under thirty people.

This is the same shift I covered in the rise of the marketing engineer. The line between “person” and “tool” is collapsing, which means the line between “in-house” and “agency” is also collapsing. The middle layer, which is what most vendors are, is the part that disappears.

How to actually do this in week one

Three concrete moves:

1. Pull every marketing line item from your last three months of bookkeeping. Categorize each as platform (Salesforce, HubSpot, ads platforms), person (employees, contractors), or vendor (agencies, consultants). Most of the savings live in the vendor column.
2. For each vendor, ask: would a competent operator with AI tools produce comparable output? For most founder-marketer hires in 2026, the answer for the four vendors above is yes.
3. Don’t cancel everything at once. Cancel the worst contract this week, redirect that budget into a hire or into the next quarter’s tooling, and run for thirty days. Then cancel the next one.

If you want help mapping which vendors to cut and how to replace them with a managed content engine, book a call. I’ve been spending most of my time on this with founder-marketers running 5-30 person teams.

Jared Castronova is the founder of JAC Growth Marketing, where he builds AI-powered GTM systems for B2B companies.