"I worked hard this quarter. I just don't know how to show it."
I've heard some version of that line from every IC who ever sat across from me in a review. The problem isn't effort. The problem isn't even outcome — most of the people saying it actually hit their number. The problem is evidence. The work happened, the result happened, and the line connecting the two is invisible.
That invisibility is what KPIcons exists to fix. Not with a brighter dashboard or a denser report. With two stages of the loop most platforms don't have at all: Credit and Adjust. This essay is about why those two stages matter more than the three that came before them, and what we built to make them real.
Activity vs. motion
Most performance tools measure activity. Calls made. Demos booked. Tickets closed. Lessons completed. The implicit theory is: do more activity, get a better outcome. So the dashboard counts the activity.
The flaw is that activity and outcome aren't the same thing, and the dashboard can't tell you which activity caused which outcome. So at review season, you get one of two stories: "I did a lot of stuff and the number went up" (correlation that nobody can verify) or "I did a lot of stuff and the number didn't go up" (and now the activity itself is the defense).
What we wanted to build was a platform that measured motion — the specific, attributable move that caused the result. Not "I made 47 calls." But "I ran the three-question reframe play on April 14, suggested by Marie, on a discovery call with Acme, and the demo conversion went from 62% to 84% the following week." That's a motion. It has an actor, a play, a persona, a context, and an outcome. It can be credited.
Why dashboards fail at this
Dashboards aren't designed for attribution. They're designed for aggregation. They show what happened, not what caused it. You can layer correlation analysis on top, but correlation isn't credit — and at review season, you still have to argue your case from circumstantial evidence.
The closed-loop tools that came before us — and to be fair, we pitched as one of them in our first deck — fixed three stages of the problem:
- See — the gap surfaces in real time
- Nudge — the AI suggests a fix
- Deploy — the move gets executed
That's the part the industry knows how to do. What's missing is what happens after the move runs. The result lands. Some moves work, some don't. Memory is short. Attribution is fuzzy. And the same person who hit their number last quarter has to argue from scratch about why they deserved the credit.
What Credit and Adjust actually do
The fourth stage of our loop is Credit. The moment a play runs, we open a ledger entry: who ran it, which KPI it targeted, which persona suggested it, what the result was, when it landed. The ledger is signed at write time and immutable after. Every motion has a receipt — not a Slack thread you have to dig up at review season, but a row in the same ledger your manager and your IT team can see.
The fifth stage is Adjust. The result of every play feeds back into the model. Marie's discovery-reframe worked seven times in a row for Sarah at noon meetings? The next nudge for Sarah at noon ranks Marie higher and the discovery-reframe higher. Annie's direct-pitch failed three times for David in afternoon demos? The model down-ranks both for that context. The coach you have today is not the coach you started with.
Those two stages — attributing the motion and using the attribution to retune the next move — turn a dashboard into a coach. Without them, you have a system that detects gaps and fires advice but never learns whose advice worked for whom. With them, you have something that gets sharper over time and tells the truth about the work.
What goes in the receipt
I get this question a lot from CIOs in due-diligence calls. The receipt schema is small on purpose:
- Subject — the person who ran the motion
- KPI — the metric the motion targeted
- Persona — which of the four AI voices fired the nudge (George, Duke, Annie, Marie)
- Play — the specific named action ("three-question reframe", "thirty-second voicemail", "patch-Tuesday escalation")
- Context — time, channel, day-of-week, deal stage, etc.
- Outcome — the delta on the targeted KPI inside a configurable attribution window
- Signature — Ed25519, key-rotated quarterly, Merkle-chained to the previous receipt
That's it. Seven fields. The whole point is that a person should be able to point at any motion they ran in the last twelve months and trace the line forward to a number that moved.
The career attribution graph
Stack a year of receipts and what falls out is something I think is genuinely new: a graph of every motion you ran, attributed to specific outcomes. Not a leaderboard. Not a dashboard. A map of your own work — what you did, when, at whose suggestion, and what it caused.
This is the part of the platform that gets the longest "huh" from prospects. Career attribution isn't a new concept — recruiters and managers have been trying to reconstruct it from LinkedIn bullet points and gut feel for decades. We just believe it should live where the work actually happens, not in someone's narrative reconstruction at review time.
If your growth doesn't have a paper trail, you're trusting someone else's memory. We don't think that's good enough.
The cost of getting this wrong
Attribution sounds like a feature until you realize it's the foundation. Without it:
- The LLM never gets sharper, because there's no signal to retune on.
- Reviews stay political, because the strongest argument wins instead of the strongest evidence.
- Managers can't tell which plays are landing, so they can't tell which ones to scale.
- And ICs keep saying "I worked hard this quarter, I just don't know how to show it."
I don't think any of those problems get fixed by a better dashboard. They get fixed when every motion has a receipt and the receipt is the source of truth.
What we're still figuring out
Two honest admissions before I close.
Attribution windows are hard. Some motions land in two days; some take a quarter. We let customers configure the window per KPI category, but the calibration takes a few cycles to feel right. We're working on adaptive windows that learn the typical lag for each play.
Negative attribution is socially uncomfortable. If the system says "Annie's pitch tanked David's last three demos," that's actionable and useful — but we have to surface it carefully, because the same data point read by the wrong person becomes performance theater. We're still iterating on the manager-side UX for this.
The covenant
The promise of KPIcons is small enough to fit on the back of a napkin: every motion attributed. If we can't credit your work, we shouldn't ask you to do it inside our platform. If we can't tell you what moved the number, we shouldn't be charging you to look at the number.
That's the difference between a dashboard and a coach. The dashboard tells you the score. The coach knows what you did to earn it — and credits you when it lands.