01Manifesto
The Insights Engineer
Every Thursday, somewhere inside a consumer brand, an insights manager rebuilds the same deck. The tracker moved two points. The marketing lead wants to know why. The answer takes a day and a half, and by the time it lands the media plan has already shipped.
That deck is the whole job in miniature. A question comes in, a study goes out, a deck comes back. The deck was the product. What happened to the decision was somebody else's problem.
Software has done this to other jobs before. Sales became go-to-market. The solutions architect became the forward-deployed engineer. The analyst who could write SQL became the analytics engineer, and within five years there were conferences for it.
Consumer insights is next. The job that replaces the insights manager is the insights engineer.
- SalesGo-to-market engineer
- Solutions architectForward-deployed engineer
- AnalystAnalytics engineer
- Consumer insights managerInsights engineer
What changed
A language model can now transcribe the interviews, code the open ends, write the summary and draft the deck before lunch. The six weeks are gone, and so is the reason to charge for them.
What is left is the part the job was always supposed to be. Knowing which question matters. Getting the answer to the person who can act on it. Checking whether they did.
That part did not shrink. One person can now do it every week instead of once a quarter.
What the job looks like
Decision
They start from a decision, not a brief
Before anyone opens a dataset, they can name the decision, who makes it, and what would change their mind. A question with no decision behind it waits.
Hover or tap a point on the loop.
They start from a decision, not a brief. Before anyone opens a dataset, the insights engineer can say which decision is on the table, who makes it, and what would change their mind. A question with no decision behind it waits.
They start with the work already on the desk. The first project is the Thursday deck. They rebuild the thing the team already does by hand, time it, and show the hours it gave back. That buys the next project.
They build it the second time. The first time a question is asked, they answer it. The second time, they build the thing that answers it. A monthly question becomes a report that arrives before anyone asks.
They send the recommendation with the finding. Nothing goes to a sponsor as a chart with the "so what" saved for the meeting. The finding and the suggested move travel together, in a form someone can put straight into a plan.
They sit with the business. Inside product, marketing and customer teams, in the room where the decision gets made, rather than in a function the business sends a brief to.
Their scorecard has two lines. Hours given back, measured against what the team used to spend. Decisions changed, with the decisions named.
Who does this
Mostly not data scientists.
The people taking these jobs come from strategy consulting and from inside insights teams. They already know how to frame a problem and tell a chief marketing officer what to do. What they add is a habit: when they do something twice, they build it.
They know enough SQL to join two tables. They write prompts the way a researcher writes a discussion guide, and they know where the model gets it wrong. They can turn a pattern into a recommendation someone will pay for.
And they check whether the decision moved. The insights function always skipped that step. It is the only step that counts.
What we think
The study was never the product. The decision was.
If you do it twice, build it.
Prove it on real work. Nobody funds a hypothetical.
A chart without a recommendation is homework for the reader.
The person closest to the problem should have the best tools, and usually has the worst.
AI is not taking the insights job. It is taking the part of the job that was in the way.
Why write it down
The title does not exist yet. No job board lists it and no course teaches it. People are doing the work anyway, under older titles, in teams that have not noticed what they have.
Titles get decided by whoever says them first and keeps saying them. Analytics engineering was a blog post before it was a profession.
This is the blog post. If someone on your team already rebuilds the Thursday deck as a system instead of a chore, that person is an insights engineer. Give them the title.