Reading your skills heat-map

Before you read it: turn on the right overlays

A naked heat-map shows current rating only. That's not enough. Before you start reading, turn on:

• Role targets, what each role should be at for each skill. Without this, you can't tell capability from over-investment.
• Date freshness, flag any rating older than its cadence (typically 12 months).
• Criticality, mark business-critical skills (loss-of-licence, single-customer-dependency).

The Excel template and PulseAI both ship with these overlays pre-built.

Pattern 1, The red column (skill gap)

What it looks like: An entire skill column is mostly red (0-1) or amber (2).

What it means: The team as a whole cannot do this skill. If the skill is critical, you have an immediate capability problem; if it is emerging, you have a training problem.

What to do: Decide whether to train (existing team can absorb), hire (no foundation to build on), or de-scope (the work doesn't need to be in this team). Pick one, set a date.

Pattern 2, The lonely 5 (single point of failure)

What it looks like: A skill column where exactly one person is at 4 or 5 and everyone else is at 0-2.

What it means: The whole team's ability to do this skill rests on one person. That person leaves, gets ill, or moves on, and the skill goes with them.

What to do: Build a cross-training plan. Aim for two people at level 3+ within 90 days. Document the steps so the 5 isn't the only one who knows them.

Pattern 3, The capability cliff

What it looks like: A row where capability drops sharply between two related skills (e.g. someone is a 5 in Skill A and a 1 in Skill B, when Skill B is the logical next stage).

What it means: Career progression is blocked. The person can't move into the next role without bridging the cliff.

What to do: A targeted development plan, almost always mentor-based not course-based. Cliffs respond to coaching faster than to courses.

Pattern 4, Hidden depth

What it looks like: A person who is at 4+ in skills that aren't in their job description.

What it means: You have under-deployed talent. Often a quiet senior with a previous career, or someone who has self-taught.

What to do: Have the career conversation. The risk of doing nothing is they leave for a role that uses the depth.

Pattern 5, Surplus capability

What it looks like: A column where every cell is at 4 or 5.

What it means: Either the skill is foundational (everyone genuinely needs to be expert, fine), or you have over-invested in training, or rater inflation has set in.

What to do: If it's foundational, do nothing. If it's over-investment, redirect future training spend. If it's inflation, run a calibration session, the descriptor at level 4/5 probably needs sharpening.

Pattern 6, Quiet decay (data quality)

What it looks like: Half the cells haven't been refreshed in 18+ months. The freshness flag is everywhere.

What it means: The matrix is no longer trustworthy enough to base decisions on. You are looking at last year's team.

What to do: Block out a re-rating sprint. Don't make new decisions on stale data; the cost of acting on noise is higher than the cost of pausing.

Pattern 7, Rater drift (data quality)

What it looks like: Two managers' teams have suspiciously different distributions for the same skill, one is mostly 4s and 5s, the other is mostly 2s and 3s.

What it means: The descriptor at each level is interpreted differently by each manager. The capability gap might not be real; the rating gap is.

What to do: Calibration session. Tighten the descriptor wording. Re-rate only where the descriptor genuinely changed. The full pattern is covered in the competency policy guide.

A repeatable reading order

The fastest way to read a heat-map for the first time:

1. Columns first. Scan for red columns and lonely 5s. These are the operational risks.
2. Diagonal next. Compare this quarter's heat-map to last. Anything that has decayed by a level needs a why.
3. Rows last. Read each person's row only for those you're about to have a conversation with. Don't scan rows for risk, scan columns.

What not to do with a heat-map

• Don't show it as a leaderboard. Capability ≠ performance. Treating the heat-map as a league table destroys honesty.
• Don't merge it into pay decisions directly. Pay belongs in a separate process; capability informs it but does not replace it.
• Don't act on a 1-cell signal. A single 2 next to four 4s is noise; a column of 2s is a signal.