People and skills
Every org has a skills matrix somebody filled in once and nobody trusts. This one is read from work that was actually delivered and confirmed by the people around it, and what you say about yourself counts for a tenth of it.
Confidence in a skill is assembled from five things, and the one people usually build these systems on is the smallest. Most of the weight sits with who confirmed it and how the work turned out.
Weights
Confirmed by others
35%
How the work turned out
25%
Read from work delivered
15%
Who they worked with
15%
What they say themselves
10%
Warehouse operations
Carrier negotiation
Demand forecasting
Vendor onboarding
Four levels, and a skill can go backwards if the evidence stops arriving.
Nobody has answered for themselves? The other four are reweighted, rather than the score being penalised.
Confirmation from above, across, below, sideways and self, counted separately rather than averaged into one nod.
How the work turned out feeds the score, so a skill is not proven by having been assigned once.
A claimed skill with no evidence says so, instead of sitting in a matrix looking identical to a proven one.
Skills are extracted from what people actually did, then matched against the skills your organisation really has. Nothing invents a skill to make a profile look fuller, and anything unmatched is dropped rather than quietly created.
You worked with Jovan on
Re-slot the Halifax picking route
Did he bring this?
Counted as a downline confirmation.
Signals
Delivered actions and projects
Who they worked alongside
How those outcomes were rated
Confirmations from five perspectives
A suggested skill your org does not have is dropped. No stubs get created to make the extraction look successful.
Agents carry skills on the same model, so the work can be routed to whichever can actually do it.
An extracted skill that does not exist in your organisation is discarded rather than created.
Agent capabilities were migrated onto the same skill model, so work routes on one language.
Confidence scoring adjusts as more evidence arrives, so the picture stays current without an audit.
Not what somebody typed into a form in onboarding. What they have delivered, what the people around them confirmed, and how it turned out.