You can see who attended.
We can see who changed.
Every talent system in your stack reports participation. None of them can tell you whether one person actually got better, where development stops, or why your best people leave. Tradecraft can — and it does it without ever seeing an individual.
This is a sponsorship problem, not a capability problem. Your development spend is aimed at the wrong thing.
That is your single largest retention leak, and it costs nothing to fix.
HR designs the people system.
The manager is the people system.
Hiring, onboarding, reviews, comp, promotion, development, retention, conflict, exits. Every one of them is designed by you and executed by a manager who was handed a process document, a deadline, and no capability at all. That's where your talent investment dies.
Tradecraft fixes the delivery mechanism, not just the output. Every manager — team lead, first-line, middle, senior — gets their own private coach for the parts of the job nobody taught them: the calibration meeting where careers are actually decided, the "no" that has to retain someone, the stay conversation nobody has, and the news they didn't decide and don't agree with.
You will never see an individual.
That's exactly why what you see is true.
Every competitor routes insight upward. One of them monitors flight risk for the employer, continuously. Employees know. And so their signal is fiction — people stop being honest with a system that reports on them.
Our data is real because your people know it can't reach you. The privacy isn't a constraint on the product. It's the data-quality guarantee.
Minimum cohort of five. Any view that would resolve to fewer than five people returns nothing. No admin override exists.
Structured computation only. Insights are built from state — goal status, competency level, blocking conditions. Never from summarising what anyone wrote. Verbatims never leave the individual.
No path from a chart to a person. Not by filtering, not by export, not by asking. The schema has no column for it.
Every query is logged. Hand the audit trail to your works council.
Managers are protected too. A manager's own coaching is private from their boss. The chain does not cascade — because nobody admits weakness to a system their boss can read.
The questions you've never been able to answer
Do our programs actually work?
Observed capability movement over time — not course completions. Before-and-after competency change for a training cohort. The closest thing to real training ROI that exists, with the confidence limits stated honestly.
Where does development break down?
The stall analysis. Not just that people stop — but exactly what's blocking them, ranked. Computed from the shape of the system, never from anyone's notes.
Are we losing people we could have kept?
Cohort-level flight risk with its structural cause. Never an individual flagged — ever. Because the moment we did, your people would stop being honest and the signal would vanish.
What should we actually teach?
The real competency gaps in your population, ranked by how often they block someone's goal — not by what a vendor's catalogue happens to contain.
Who is genuinely ready?
Readiness backed by observed evidence, not manager nomination. And for the ones who aren't: exactly which condition is blocking them. Usually it's sponsorship, not capability.
Are we using what we already pay for?
Tuition reimbursement, EAP, mentorship — surfaced to the right person at the moment they need it. Utilisation of spend you've already committed.
Everyone else infers. We observe.
The talent intelligence category has a data problem it can't solve. These platforms guess at skills from résumés, job descriptions and click behaviour — and a platform purchase doesn't manufacture the data it needs. Buy one before you have that data and you get an expensive dashboard that changes no decisions.
Tradecraft generates observed capability data as a by-product of coaching your people. Every level traceable to a real thing someone did, in a real moment at work. It maps into your existing skills architecture. It decays honestly, because skills do.
See the stall analysis on your own data.
A 30-minute demo. We'll show you the one artefact nobody else can produce — and be honest about what we can't yet prove.