A support team lead's practicum log covers the decisions only a lead makes: goodwill above reviewers' limits, escalated complaints, quality checks on replies the AI agent and the team sent, and proposed changes to the AI agent's knowledge. Each entry records the decision, the reason, the coaching or knowledge action that follows, and a support manager's independent call.
What is a team lead actually tested on?
Two things reviewers are not. First, decisions with precedent: a goodwill payment above the usual limit tells the team what is acceptable next time. Second, the loop back into the system: when a reply goes wrong, the lead decides whether the fix is coaching a person, correcting the AI agent's knowledge, or asking for its bounds to change.
So the log has a "follow-up" column that front-line logs do not. A lead whose decisions are right but who never follows up is clearing items, not leading a team. For the front-line view of the same queue, see the practicum log for customer support agents. The setup for leads is in how to run a two-week practicum for support team leads.
Which items reach the lead?
- Goodwill and refunds above a reviewer's limit.
- Complaints escalated by a reviewer or the AI agent.
- Quality samples: a set number of the AI agent's and reviewers' sent replies each day, marked pass or fail.
- Knowledge proposals: reviewers suggest new or corrected answers for the AI agent.
- Disputed refusals: a reviewer thinks a refusal was wrong.
How do you fill it in?
- Record the item and what was proposed.
- Record the decision and a reason a reviewer could learn from.
- Record the follow-up: coaching (who, on what), knowledge (which article, from which source), bounds (a request to the agent's owner), or none.
- The support manager records a decision before reading the lead's.
- At day ten, count follow-ups as well as matches.
The practicum log template
Header: team lead; support manager with a current certificate; team and queues; lead's goodwill limit; daily quality sample size; dates; bar of at least 95 correct in every 100 and no missed red line.
Red lines
- Approving a knowledge change that has no sealed or signed source.
- Approving a reply that promises something outside policy.
- Closing a complaint without a response that addresses it.
- Approving goodwill above the lead's own authority.
| Field | Entry |
|---|---|
| No. and date | |
| Item type | Goodwill, complaint, quality sample, knowledge, refusal |
| Raised by | AI agent / named reviewer |
| What was proposed or sent | |
| Lead's decision | |
| Reason | |
| Follow-up | Coaching / knowledge / bounds / none, with detail |
| Support manager's decision | |
| Match | Yes / no / defensible |
| Red line | None / caught / missed |
A worked example
Take a sample online electronics retailer. Values are illustrative.
Entry 22. Quality sample: the AI agent told a customer their laptop warranty covers accidental damage. The policy library (sealed) says it does not. Lead: fail; send a correction to the customer. Follow-up: knowledge, the agent's warranty answer cites an old product page; replace the source with the policy library. Manager: same. Match: yes.
Entry 23. Knowledge proposal from a reviewer: "Customers can return opened headphones within 30 days." Source: the reviewer's memory of a previous policy (declared). Lead: approve; it sounds right. Manager: return; no source, and current policy excludes opened in-ear headphones for hygiene. Match: no. Red line 1: missed.
Entry 24. Complaint: a customer received a damaged television twice. Reviewer's draft offers a third replacement. Lead: approve replacement plus goodwill of 50, within the lead's limit of 100. Follow-up: coaching, the reviewer should offer a choice of refund or replacement after a second failure. Manager: same. Match: yes.
Entry 25. Disputed refusal: the AI agent refused to change a delivery address after dispatch. Reviewer asks the lead to override. Lead: uphold; contact the courier instead. Follow-up: none. Manager: same. Match: yes.
Entry 23 fails the bar. Knowledge changes reach every customer the AI agent speaks to, so a lead's approval there carries more weight than any single refund.
How does this compare with other logs?
The structure carries into other fields where a person decides with authority over others' work. Underwriters, for example, approve referrals within limits and their decisions set precedent too; compare the practicum log for underwriters. The Operator track at Pryme Intelligence Academy names team leads among its audience and covers teaching the agent and keeping knowledge current, the two skills entries 22 and 23 test. More in AI literacy and certification for people who work with agents.
Questions people ask
How many quality samples should the lead mark each day?
Enough to include several AI agent replies and several from each reviewer, often ten to twenty. Keep the number fixed for the fortnight so days can be compared.
Can the lead approve their own knowledge proposals?
No. Another lead or the support manager approves them, and the log records who.
Is a team lead accountable for the AI agent's replies?
For items they approve and knowledge they change, yes. See are support team leads accountable for an AI agent's mistakes and do support team leads need a certificate to approve AI agent actions.