The three confidence tiers: an honest attribution primer

Most marketing numbers are reported as if they're all equally true. They aren't. A conversion proven by a promo code and a conversion guessed from a view-through window are both printed in the same black font, and that's the lie.

GrowthCrew grades every number by how much proof stands behind it, using three tiers: exact, strong, and inferred. This primer explains each one, with examples, so you can grade your own numbers the same way.

The three tiers, defined

Exact — a receipt exists. A promo code was redeemed. The link between the touch and the outcome is unambiguous because the customer performed a specific, unique action that only that campaign could produce.

Strong — a tracked link was clicked and the same session converted. You have a real trail, but it's circumstantial: the click happened, the signup happened, and they're tied together by a session and a timestamp rather than a unique redemption.

Inferred — a self-report. "Saw it on YouTube." "A friend sent it to me." Real signal, no hard trail. It's the customer telling you a story, and people misremember, round up, or credit the last thing they recall.

Same event — "this creator drove a signup" — can land in any of the three tiers depending on how you know it. The tier isn't about how important the number is. It's about how much you can prove.

Why a promo code beats a click beats a self-report

Walk one customer through all three tiers and it's obvious.

Exact. A creator posts code ADITYA20. Someone types it at checkout. That code appears in exactly one place in the world — that creator's post. There is no other explanation for how it got typed. This is as close to proof as attribution gets. It's a receipt.

Strong. The same creator shares a tracked link. Someone clicks it and signs up in that session. Very likely the creator drove it — but the person could have had the link open in a tab for a reason unrelated to this session, or clicked out of idle curiosity and converted for a different reason. The trail is real; it's just not a receipt.

Inferred. No code, no click. At signup, the person picks "YouTube" from a dropdown. Could be this creator. Could be a different video. Could be misremembered. It's the softest signal, but for the 70%+ of the buyer journey that happens in dark social — the DMs, the WhatsApp forwards, the AI-chat recommendations no pixel can see — inferred is often the only signal you'll ever get. So we count it. We just count it honestly, as inferred.

The ranking is exact > strong > inferred, and it's about the strength of the evidence, not the size of the win.

Why ranges beat point estimates

Once every touch is tiered, something better than a single number falls out: an honest range.

Instead of reporting a confident "19 signups from this creator," GrowthCrew can report "4 exact, plus 8 strong, plus a handful of inferred — call it 14 to 22 verified signups." That range is more useful than the point estimate, for a blunt reason:

A truthful range beats a flattering point estimate. The point estimate hides its own uncertainty. It looks precise, so you trust it more than you should, and you make a scaling decision on a number that was always shakier than it appeared. The range shows you exactly how firm the ground is. The width of the range is information — a tight range means scale with confidence; a wide one means get more proof before you bet.

Any tool can give you a clean-looking number. Very few will tell you how much that number can be trusted. The range is where the honesty lives.

How to grade your own numbers

You don't need our software to start thinking this way. For any marketing result you're about to report, ask one question: how do I know?

• If the answer is "a code was redeemed" or "an event fired that only this campaign could cause" → exact.
• If it's "a tracked link was clicked and the session converted" → strong.
• If it's "they said so" or "signups went up around then" → inferred.

Then report the tier alongside the number. "12 signups (exact)" and "12 signups (inferred)" describe very different levels of confidence, and once you say the tier out loud, you'll make better decisions about which one to bet money on.

Why we built it this way

Confidence tiers are the honest core of GrowthCrew. Every touch we record carries one, stored right on the data. Attribution is then triangulated across all three — the exact receipts, the strong trails, and the inferred self-reports — instead of pretending one source is the whole truth.

Nothing gets promoted a tier to look better. A self-report never becomes a "strong" touch because it would make a campaign look successful. An inferred signup stays inferred.

[FACT-CHECK: real distribution of exact/strong/inferred touches across our design-partner campaigns, if we want to show a live example]

The goal isn't to make your numbers bigger. It's to tell you, for every single number, exactly how much you should trust it — and then let you decide.