Your personalization engine is a Ferrari with no driver. Start with a workshop, not a rollout.
Every month we sit with a founder who just bought a personalization tool. They spent five figures on software that promises to serve the perfect message to every user at the perfect moment.
Every month we sit with a founder who just bought a personalization tool. They spent five figures on software that promises to serve the perfect message to every user at the perfect moment. Then they look at us with hopeful eyes and ask, “Now what?” The honest answer is usually, “You’re about to waste that money unless you stop and think first.”
We have watched a dozen founders make this exact mistake. They wire up the engine, turn on the data feed, and let the algorithm loose. Within two weeks they get the toilet seat email. You know the one. The email that asks someone who just bought a house to buy another toilet seat. That is not a technology failure. That is a thinking failure. The engine did exactly what you told it to do, and you never told it what not to do.
The fix is not more data. The fix is a prepersonalization workshop, a half day where you decide what personalization is actually for before you let it touch a single customer. Your engine is a Ferrari with no driver. The workshop is the driving lesson.
What this means for founders
Here is the verdict: do not let your engineering team configure personalization on their own. Do not let marketing pick a use case from a vendor demo. You run the workshop yourself, with three people maximum, and you answer one question only: what decision are we helping the user make faster?
That is the entire game. Personalization is not about knowing the user’s name or their birthday. It is about removing friction from a decision they already want to make. A returning customer deciding whether to reorder cat food. A new visitor deciding whether to trust your pricing page. Those are decisions. “Show them a different hero image based on their location” is not a decision, it is decoration.
In the workshop, write down every user moment where you currently lose people. Rank them by revenue impact and by how easy they are to personalize. Pick one. Just one. Then define what a good outcome looks like in numbers, like a 10 percent lift in repeat purchase rate, not in vibes. That becomes your pilot. You run it for two weeks, you measure it, and only then do you expand.
The second thing the workshop forces you to do is write the failure list. What should the system never do? For a furniture store, that is never suggesting a second sofa to someone who bought a sofa last month. For a SaaS tool, that is never suggesting a beginner feature to a power user. Write those rules down in plain language. That list is your guardrail, and it matters more than any machine learning model you will ever train.
Key takeaways
- Run a half day workshop before you configure any personalization tool, with no more than three stakeholders in the room.
- Define personalization as helping one specific user decision, not as showing different content for its own sake.
- Pick a single pilot use case tied to a measurable business outcome, like conversion rate or repeat purchases.
- Write an explicit “never do this” list for the system before you launch, and treat it as a product requirement, not a nice to have.
This post builds on research originally published by A List Apart on August 11, 2026. We adapt established industry research into practical, first person guidance for founders.