the job description is the wrong unit of analysis
job titles average away the variation that matters, and the decisions enterprises now face need a smaller unit than the job.
Every large organisation runs internal mobility on two different clocks. Most only wind one of them.
The first clock is fast. A role opens, a requisition goes out, and somebody has thirty or sixty days to close it. The measure is days-to-fill. The owner is talent acquisition. The question is narrow and answerable: who can do this job now.
The second clock is slow. It runs for three years or longer. It asks where a person is heading, what stands between them and that destination, and which assignment closes the distance. The measure is contested. The owner is frequently nobody.
Programmes get built around the fast clock and then judged by the slow one. That mismatch explains much of what goes wrong with internal mobility, and most of why employees stop believing in it.
both clocks pay, and the evidence is not subtle
LinkedIn's 2022 Workplace Learning Report found that companies which excel at internal mobility retain employees for an average of 5.4 years. At companies that struggle with it, the average is 2.9 years. That is close to double the tenure for the same headcount.
The mechanism shows up in movement data. LinkedIn analysed 32 million member profiles at companies with more than 500 employees. Among people who made an internal move within three years of being hired, 70 per cent of those promoted were still there, and 62 per cent of those who made a lateral move. Among those who made no internal move at all, 45 per cent remained.
Read the lateral number again. It sits far closer to the promotion number than to the no-move number. Sideways movement retains people almost as well as upward movement does. That matters in flat structures and in markets where headcount growth has stopped, because lateral moves are the supply you already have.
the fast clock is failing on its own terms
Before adding anything, check whether the clock you already run is working. Gartner surveyed 3,000 candidates in June 2021 and found that only 33 per cent of people who looked for a new job in the previous twelve months searched inside their own organisation first.
The same research found only 51 per cent of candidates were aware of the internal openings at their employer. Only 17 per cent said their manager helped them apply for an internal role. Only 20 per cent felt supported by their peers and team in doing so.
Those four numbers describe a market with poor information and an active tax on participation. Publishing more roles does not fix it. The binding constraints are visibility and manager behaviour, and only one of those is a technology problem.
the slow clock has almost stopped
LinkedIn's 2025 Workplace Learning Report found that only 15 per cent of employees said a manager had helped them build a career plan in the previous six months. That figure fell five points from the year before. The survey covered 937 learning and HR professionals and 679 learners across four regions.
The consequence appears at the moment organisations think they are winning. ADP Research Institute studied the job histories of 1.2 million US workers between 2019 and 2022. Within one month of receiving a first promotion, 29 per cent had left their employer. Among comparable people who were not promoted, 18 per cent left.
ADP's own reading is that a promotion raises a person's value on the open market. There is a second reading worth testing inside your own data. A promotion is a fast-clock event. If nothing on the slow clock explains what the promotion was a step towards, the employee is left to interpret it alone, and the most available interpretation is a price signal.
one profile, two read speeds
The usual response is to buy a talent marketplace for the fast clock and run career conversations for the slow one. Two systems, two data sets, two owners. The marketplace knows what people can do and does not know where they want to go. The career conversation knows the aspiration and cannot see the openings.
The alternative is one record read at two speeds. Same evidence, two layers, one owner per layer.
the fill layer
- Proven now. Skills with evidence attached and a date. A skill claimed in 2021 is history, not capability.
- Available from. A real date, set by the current assignment, not by the manager's preference.
- Constraints. Location, visa, notice period, family commitments. Recorded once so nobody has to guess.
the path layer
- Target role at 24 to 36 months. One role, named. Not a level, not a family.
- The two or three gaps. What separates the person from that role, written as observable capability rather than as a competency score.
- The next assignment. Which gap it closes, and by when. An assignment, not a course.
- Review date. Fixed, so the layer expires if nobody touches it.
the rule that connects the two
Two layers on one profile still gives you two clocks running independently. One rule joins them, and it is the whole design.
Every fill decision writes back to the path layer. When a person is placed, record which named gap the move closes. When a person is shortlisted and not selected, record what was missing and what would change it. When a person is not shortlisted at all, record why.
This costs a hiring manager about four minutes per candidate. It converts the fast clock from a consumer of the slow clock into its main source of data. Every requisition becomes a calibration event for dozens of career plans, at no additional meeting cost.
Add one placement rule to keep the loop honest. For every open internal role, at least one shortlisted candidate should be a stretch placement whose path layer says this role closes a named gap. If no such candidate exists, the path layers are stale and you have found that out cheaply.
the two numbers to put in front of the executive committee
Report a pair, never one alone.
- Internal fill rate. Share of open roles filled internally, by quarter. This is the fast clock, and it is already measured almost everywhere.
- Path movement rate. Share of employees whose path layer changed in the last twelve months because of an assignment, not because of a rating or a conversation. This is the slow clock.
The pair works because each one constrains the other. A high fill rate with a flat path movement rate means you are recycling the same available people. A high path movement rate with a low fill rate means development is running without a destination.
the diagnostic to run first
Before changing any process, run the twenty-leaver test. It takes a morning.
Take the last twenty voluntary resignations from your target population. For each one, answer three questions from records you already hold. Was there an open internal role at the time of resignation that this person met the requirements for? Had they applied for any internal role in the previous twelve months? Had a manager recorded a career conversation with them in the previous six months?
The pattern of answers tells you which clock is broken. Mostly yes to the first and no to the second means you have a visibility problem. No to the third across the board means the slow clock is not running, and no marketplace will substitute for it. Mixed answers concentrated under three or four managers means you have a talent-hoarding problem, which is a management issue rather than a systems one.
what this asks of you
The two-clock model does not require new technology. It requires one record instead of two, a four-minute write-back after every shortlist, and a standing agenda item where the person who owns days-to-fill and the person who owns capability sit in the same review.
Most organisations already hold the evidence. It is scattered across an applicant tracking system, a learning platform, a performance record and a set of manager notes that never meet. Bringing those into a single profile that both clocks can read is the practical work, and it is where entomo's internal mobility approach starts.
The test of an internal mobility programme is not how many roles you filled internally last quarter. It is whether a capable person, three years in, can name where they are going and point to the assignment that is taking them there. If they cannot, the fast clock is the only one running, and the retention data says what that costs.