Why Your MarTech Stack Is Bigger Than It Needs to Be

Most companies know how to add a new marketing tool to the stack. Almost none of them have a plan for getting rid of one.

The three questions

Frans Riemersma has co-created the Martech Map with Scott Brinker for years. He's seen what companies buy, why they buy it, and what happens when it doesn't work out. His framework is refreshingly simple.

First: what are your use cases? Not a 200-item RFI. Twenty or thirty items, focused on what campaign managers actually need to be able to do. Frontend first. Don't over-engineer it.

Second: integrations. What connects to what, and how well?

Third — and this is the one most people skip: how do you stop using this system? License terms. Data migration. User removal. The whole exit.

"That's where you're really agile in buying your Martech, scaling it, using it, learning from it — but also maybe deprecating it."

The signs your stack has become a mess

The signs are often hidden. You don't know what you're no longer using. People who left the company still have access. License fees keep renewing quietly. And then one day the CFO slams a door.

Frans's solution: someone in Marketing Ops needs to own governance — not as a one-off project, but as an ongoing role. Walk over to the people using the tools. Are you still using it? Until when? Clean, lean, mean.

The misconception, he notes, is that this is about saving money. It's not. It's about clarity. You can't build something coherent on top of a stack you don't understand.

On AI: play first, deploy later

Frans cites a recent MIT report finding only 5% of AI projects made it to production. His response is not alarm — it's perspective.

"It shows more to me that we have too high expectations."

Don't try to operationalize AI tools before you've played with them. Don't assume colleagues know how they work either. Set aside 30 minutes a week — the same advice he gave about learning Martech years ago — and experiment. Get comfortable with the failure mode before you build anything critical on top of it.

One nuance worth noting: prompting takes time. The first round of outputs rarely saves hours. The value compounds later. Expect that.

Security: the risk nobody talks about

The security risks in AI tools are not fundamentally different from the risks in regular SaaS. Governance is the gap. Frans's practical fix: get legal in the room — not compliance-first legal, but someone who understands the business and can make constructive decisions.

Future-proofing: what can you leave out?

Strip it back. Ask what you can remove without breaking revenue — without losing the customers who actually pay the bills. Rationalise that first. Then reverse engineer from what's left.

"Bring it down to the core, to the essence. And then it becomes so much more simple to say: we need an AI agent here, we can add some GenAI there."

Keep it simple, stupid. Hasn't expired yet.

The bottom line

Frans Riemersma has spent years mapping the entire Martech landscape — and his conclusion isn't that you need more of it. It's that you need to be far more deliberate about what you keep, what you add, and how you plan to leave.

The question most vendors will never ask you: how will we make it easy for you to walk away?

Start asking it yourself.

Find Frans on LinkedIn: linkedin.com/in/fransriemersma