Somebody has to go and look.
Past the Job Description
When the work still isn't right
You've shipped work that counted as finished. It still didn't do what it was meant to do.
Marketing saw qualified leads; sales saw noise in the same records. The buyers were real, but weren't ready to buy that quarter. The two reports answered different questions.
You've checked a dashboard and found its figures accurate. The business still wasn't working as those figures suggested.
You've had more people on the work and less progress to show for it. What was meant to help left you moving more slowly.
You did your job. The job was defined to end before the outcome.
The line between jobs is drawn where work counts as finished: shipped, sent, approved, accurate.
The outcome lies past that edge, in someone else's part of the work. Even your certainty can close a question before the answer has been checked.
And you've been certain. You've felt no need to check a call, then found it was wrong.
The certainty hadn't felt any different from the times you were right. You had the feeling, but it didn't tell you which kind of call you'd made.
We are Market-Side Builders. We investigate demand failures from inside marketing operations, using SQL, Python and site code across marketing, sales and operations.
We stay responsible for the outcome when its cause takes us outside the channel we started with. That doesn't make us experts in every department. It means looking with the people whose work connects to ours.
We work inside the workflow long enough for trust to matter. That trust has to leave room for an answer people would rather not hear.
How the work connects
AI makes coding and plans cheaper; the checks remain. We look at how a problem gets framed, how work passes between people, and whether what reaches the customer matches what the numbers say.
Before work moves
The Context Flow. We frame the problem before AI drafts, so we have our own judgment to compare against. The First Move →
The Information Flow. We check what others understood, because sending a document does not prove its meaning arrived. A Certain Answer →
As work moves
The Momentum Engine. We trace coordination, handoffs and feedback before adding help that could make the work slower. The Extra Chair →
Shadow Strategy. We follow budgets, calendars and rewards to see whether the stated strategy can change the work. The Version That Runs →
Where work lands
Dashboard Blindness. We compare metrics with customer outcomes and name who can stop the work when they disagree. Accurate and Still Wrong →
The Perception Engine. Did we build this message for the people who might buy, or the room that approves it? The Perception Engine: Why Marketing Fails by Design →
These aren't separate subjects just because departments give them different names.
A message depends on what people understood. A plan depends on the budget and the handoffs that carry it. A metric needs checking against what the customer gets.
Following those connections doesn't mean taking over someone else's job. It means looking with them when the cause crosses the line between jobs.
My view of AI follows the same logic: AI changes work. The mechanisms underneath don't. A cheaper plan still needs someone to check whether it can reach the customer.
Start with your own calls
Where do you start? Look at the edge of your own work. Which question do you stop asking because the answer belongs to someone else?
Feeling sure didn't tell you whether you were right. Write down five calls you feel sure about, put a percentage beside each, and return to score them. Leave yourself something more exact than a memory of having been right.
That's how Field Notes begins. I write five short emails, one a day, making one argument. The first asks you to record your calls; the fifth asks you to score them.
After that, Field Notes arrives on Sunday mornings, most weeks. It's free, with no course behind it and no lead magnet. You can join us in that work.
The map helps. The market decides. I build accordingly.
