Seeing in Depth

Why three imperfect views beat one clean one

By now you have evidence. You gathered what was already known and generated what wasn’t, and you are holding a handful of findings: a market number, a few interviews, a scrappy test, a competitor’s move. Here is the uncomfortable truth about all of it. Every piece is partial. Every source is biased in its own way, blind to something, true only from one angle. The founder’s skill is not finding the one perfect source, because there isn’t one. It is learning to see what several imperfect sources show when you hold them up together.

That is triangulation, and it is how you turn a scatter of flawed findings into a picture you can act on.

Triangulation — reading several imperfect sources together, so their overlap points somewhere no single source could point on its own.

Three angles, one picture

Evidence comes in three kinds, and each sees something the others miss:

  • Numbers (the quantitative sources, public and private) show you scale and pattern, but never why.
  • People (interviews and experts) show you meaning and motive, but from a small and subjective sample.
  • Behavior (what customers actually do, in a test or in your own data) shows you the truth, but only in the narrow slice you managed to watch.

The ladder you just climbed handed you all three: the low rungs gave you numbers and the voices of people, and the top rungs gave you behavior. Any one of them alone is a flat photograph. Hold all three up at once and they give you depth, the way two slightly different views give your eyes distance. When the numbers, the people, and the behavior all point the same way, you are no longer guessing. You are seeing in three dimensions.

When they agree, and when they don’t

Convergence is what you are hoping for. Three weak signals pointing the same direction are worth more than one loud headline, because a single source can be wrong in a big way while three independent ones rarely fail together. So look for the pattern across your sources, not the strongest single point, and when they line up, trust the direction even if no one of them is airtight.

But the more useful moment is the one most founders shrink from: when the sources disagree. A survey says customers love the product; the renewal numbers say they quietly leave. That contradiction is not noise to be smoothed away. It is the finding. The gap between what people say and what they do is almost always where the next question lives, and triangulation is what makes that gap visible. Disagreement is not the failure of your evidence. It is your evidence telling you where to look next.

Trap to Avoid — The comfortable source

Everyone has a favorite kind of evidence. One founder trusts the spreadsheet, another the customer quote, another the seasoned expert. When the sources conflict, the temptation is to believe the one that feels most like home and quietly discount the rest. That is not confidence; it is bias wearing confidence’s clothes. The source you least want to believe is often the one with the most to teach you.

Halo Alert — reading three angles

The Halo Alert team had gathered three kinds of evidence, and they read them together. The numbers said the fear was real and widespread. The people, their interviews and an hour with a former dispatcher, said the same, and added when and where the fear spikes. So far, convergence, and they could trust it. The problem was real. But the behavior they had watched in the field sat against what people said. Women who told them they wanted a dedicated device still, in the moment, reached for the phone already in their hand, the very workaround the ring was meant to replace. That gap between the saying and the doing did not sink the idea. It handed them their next question, the one worth testing: not whether people wanted safety, but what would make them reach for this.

Working with your AI

Working with your AI — where you step in

Your AI is good at laying sources side by side. You are the one who reads what their disagreement means.

  • Line them up. Have it collect your numbers, your interviews, and your behavioral results in one place and mark where they agree and where they clash.
  • Weigh the convergence. Ask which conclusions hold across several independent sources, rather than resting on the loudest single one.
  • Mine the conflict. When two sources disagree, don’t let it average them into a bland middle. Have it name the contradiction sharply, because that is your next question.

Ask Your AI

Here is the evidence I’ve gathered on this decision, from different sources: [paste your numbers, your interview notes, your test results]. Lay them side by side and tell me where they converge and where they contradict each other. For each contradiction, tell me what it might mean and what single test would resolve it.

Putting It to Work

Try This — Triangulate one belief

Take a belief you’re leaning on to make this call.

  1. Find three sources for it: one number, one person, one piece of real behavior.
  2. Write what each one says, plainly, side by side.
  3. Mark where they agree. That agreement is where your confidence belongs.
  4. Mark where they disagree. That gap is your next experiment, and probably the sharpest one you could run.

The move: Never rest a decision on one source. Read your numbers, your people, and your behavior together, trust what they agree on, and treat what they contradict not as a mess to tidy but as the next question to test.

This is where the work of gathering ends. You came into this part with a prior and a decision to make. You gathered what the world already knew, generated what it didn’t, and now you have read the results in depth, from every angle you could afford. You have a picture. What you do not yet have is a verdict on how far to trust it, or on what exactly it should do to your belief. That is the next gate. Having gathered your evidence and seen it in the round, you turn now to weighing it.