Should You Build or Buy Your AI Marketing Stack?

Your team doesn't need to build every AI workflow. Decide what to automate in-house, buy from a vendor, or outsource. The most valuable 30 minutes in marketing right now are the 30 minutes you spend thinking before you start building any AI automation.

What Happened

In a September 16, 2026 piece for Search Engine Land, Kevin Indig and Amanda Johnson called out a pattern across their client rosters: marketing teams are defaulting to in-house AI builds instead of sorting which work belongs to a partner, a consultant, or their own people. They cite MIT research showing 95% of enterprise genAI projects return zero, with strategic partnerships outperforming internal builds on success rate.

The Numbers in Context

MIT's review pegs enterprise genAI investment at $30 to $40 billion, with 95% of organizations seeing zero return. The prior benchmark most CMOs anchor to is the 2023 wave of pilot programs, where build-versus-buy was rarely a formal decision. The failure mode has not changed: tools that do not learn or integrate with how people already work get abandoned, regardless of how much internal engineering time went into them.

Why This Matters for HR Tech and FinTech Marketing Leaders

You are running lean teams under pressure to show AI productivity gains. The temptation to have a growth engineer wire up a custom citation tracker or a bespoke content scorer is real, and it feels cheaper than another SaaS line item. It is not. When you count prompting, QA, maintenance, and the meetings to rebuild it after the model changes, in-house builds routinely cost more than the platform fee you rejected. In regulated categories like FinTech and HR Tech, you also inherit the compliance surface of anything you build. That is a liability your legal team did not sign up for.

The Starr Conspiracy's Take

The 30-minute triage Indig describes is the single highest leverage exercise a B2B marketing leader can run this quarter. Buy when the problem is common. Hire expertise when the workflow is yours but the pattern is not. Build only when the workflow and the pattern are both proprietary, which is rarer than your team thinks. We walk clients through this same sort in our work on how AI is reshaping the B2B marketing operating model, because the teams winning right now are the ones who stopped confusing engineering activity with marketing outcomes.

What to Watch Next

Expect Q1 2027 budget cycles to surface the first honest accounting of in-house AI build costs. Watch for CFOs asking marketing to justify engineering headcount tied to AI workflows that a $2,000-per-month tool already solves. That conversation is likely, and you want to be ready with the triage before it lands.

Related Questions

When does building an AI workflow in-house actually pay off?

Build in-house only when the workflow, the data, and the decision logic are all proprietary to your business. If any of the three is common across your category, a partner has already amortized the cost across hundreds of clients. Your build will not catch up.

How should marketing leaders budget for AI tooling versus AI labor?

Treat AI tooling and AI labor as one line item, not two. Every tool you buy reduces hours; every hour you insource increases tool debt. Our view on demand states and the modern B2B buying cycle explains why the math has to be run against pipeline outcomes, not activity counts.

What is the biggest risk of over-building AI internally?

Maintenance debt. Models change, APIs deprecate, and the engineer who built it leaves. You end up with a fragile stack nobody owns and no partner support line to call when it breaks during a launch week.