Support leaders need a competency matrix of their own because their work with agents differs from their teams'. Advisers review held replies; leaders set the refund and goodwill limits those replies run under, calibrate reviewers, staff for certificate lapses, and decide when evidence justifies more autonomy. This template sets those skills by leadership role.
How is a leadership matrix different from a team matrix?
A frontline matrix asks whether people can decide held items well. A leadership matrix asks whether leaders can keep a whole system of agents and reviewers working well. Different skills, sometimes held by the same person on the same day.
Four stand out. Setting limits: support leaders usually own the organisation's own limits on refunds, credits and goodwill, which are one of the five bounds every agent action must meet. Set them too low and the queue floods; too high and the agent acts where judgement was needed. Calibration: getting ten reviewers to decide the same held refund the same way. Capacity: forecasting held volume, including the day an agent's certificate lapses after 90 days and its routine work returns to people. Escalation: knowing when a run of complaints or refusals means the agent's scope is wrong, not the customers.
The frontline counterpart is the support team's competency matrix. The full picture is in AI agent skills by role.
How do you fill it in?
- List the leadership activities in the left column. The template gives a starting set.
- For each, mark the RACI letter for each leadership role, using the logic of a RACI matrix: Responsible, Accountable, Consulted, Informed.
- Beside each letter, set the competency level needed: E (emerging, learning under guidance), P (proficient, does it alone), X (expert, teaches it).
- A role marked Accountable must be at P or X. A role marked Responsible at E needs a named coach.
The template
| Leadership activity | Team lead (RACI, level) | Support manager (RACI, level) | Head of support (RACI, level) | Evidence |
|---|---|---|---|---|
| Set refund and goodwill limits for agents | ||||
| Run weekly calibration on held replies | ||||
| Coach reviewers from corrections | ||||
| Forecast held volume, including lapse dates | ||||
| Staff the queue during a lapse or downgrade | ||||
| Read refusal patterns for scope gaps | ||||
| Request a depth change for an agent | ||||
| Handle a complaint about an agent's reply | ||||
| Report agent performance to leadership |
A worked example
A hypothetical subscription business with six team leads, two support managers and a head of support, running agents that answer billing questions and issue small credits at depth 3.
| Activity | Team lead | Support manager | Head of support | Evidence |
|---|---|---|---|---|
| Set limits | C, P | R, X | A, P | Limit decision log |
| Calibration | R, X | A, P | I, E | Monthly calibration record |
| Coaching | R, P | A, P | I, E | Coaching notes |
| Forecast held volume | C, P | R, X | A, P | Capacity plan |
| Staff during a lapse | R, P | A, X | I, P | Drill report |
| Read refusal patterns | R, P | A, X | I, P | Weekly pattern review |
| Request depth change | C, P | R, P | A, P | Owner request on file |
| Complaint about a reply | R, P | A, P | I, E | Complaint file |
| Report performance | C, E | R, P | A, X | Leadership pack |
Team leads are expert at calibration, ahead of managers. They run the session each week, pick the cases, and see which reviewer approves more and explains less. Managers hold accountability and check the records.
The head of support is only emerging on calibration and coaching. That is acceptable for someone Informed, but the head is Accountable for limits and for depth requests. Before signing either, they should be proficient at reading what the agent's credits look like in the record: the size, frequency and reasons. Limits are set by evidence, not by instinct about customers.
Approvals for limit changes belong in the approver authority matrix for support leadership, and who covers the queue on which shift in the reviewer rota for support leadership. Leaders who want certified evidence of these skills can take Pryme Intelligence Academy's Operator track, which includes running the rhythm and reading the workspace.
Questions people ask
Should team leads still review held replies themselves?
Yes, regularly. A team lead who stops reviewing loses the feel for what the agent gets wrong, and calibration sessions become abstract. Many teams have leads take a share of the queue each week.
Who sets the agent's refund limit?
The organisation sets its own limits; within support, the manager usually proposes and the head of support approves. The agent's effective limit is then the lowest of all five bounds, so a generous limit cannot override a lapsed certificate or a scope restriction.
What is the first leadership skill to build?
Forecasting held volume. Most early problems come from queues no one planned for, especially around lapse dates.