Every review here comes from the form below. Nothing is written on anyone's behalf, and names stay private unless the reviewer asks otherwise.
Nothing is published yet. When a review arrives it is read before it goes up, and only genuine submissions from people who have actually used the tools are shown.
Titles had been negotiated one at a time during a hiring push, leaving several teams with three seniority labels for the same work. The exercise was to separate the question of what the work actually was from what people had been called, then rebuild titles from the level structure rather than the other way round.
A taxonomy had been drafted twice and abandoned twice, both times because the team could not agree how granular a family should be. The useful move was a test applied consistently: would this be hired for as a distinct capability, and does it have its own progression logic. Anything failing both was a sub-family, not a family.
Every previous attempt had been read by the business as a re-grading exercise, which meant managers argued level in order to protect pay. Sequencing the work as a work-content exercise first, with any pay implications handled separately and later, removed the incentive to game the mapping.
A technical track existed on paper but stopped two levels below the management track and held nobody above its midpoint. Engineers had read that correctly. The work was structural rather than communicational: raise the ceiling to parity, and populate it through promotion rather than external hiring.
The concern was less the mapping itself than the disruption of running it. Sequencing by function, using calibration anchors instead of case-by-case debate, and keeping a visible exceptions log meant most roles resolved quickly and the genuinely ambiguous ones were handled deliberately rather than argued repeatedly.
Ten professional levels had accumulated, and adjacent descriptors used nearly identical language. Two reviewers reading the same person reached different answers. Reducing the count until every level had a genuinely distinguishable descriptor made calibration tractable again.
Two organisations arrived with incompatible level structures, different titling conventions and no shared taxonomy. Mapping both onto a common global grade spine — rather than choosing one side's structure and imposing it — gave a neutral basis for comparison before any structural decisions were taken.
The framework had been designed carefully four years earlier and then left ungoverned. Most of the population had drifted into the top two levels, entry levels sat empty, and several levels existed for a single person. Re-anchoring meant restoring calibration anchors and putting a real approval gate on creating a level.
Several thousand distinct job titles existed for a few thousand people, so no question beginning "how many people do we have doing…" could be answered without manual work. Establishing a stable job layer beneath the titles made the underlying data queryable without touching what anyone was called.
Employees asking about progression were being told, in effect, to wait for a vacancy. There was no ladder to publish because none existed. Building the level structure first gave something concrete to publish, and made the adjacent moves between families visible for the first time.
Six management levels had been defined against four actual reporting layers, so two levels described work that structurally could not exist. Sense-checking the level design against the real structure before publication avoided promoting people into levels that would have meant nothing.
At just under three hundred people the instinct was to build the architecture a much larger organisation would run. The more useful conclusion was a minimum viable version — three tracks, broad bands, one page of governance — with an explicit list of what to defer until scale genuinely required it.