What is wrong with sampled quality review?
Sampled quality review, where a lead reads a handful of tickets per agent each month, tells you about those tickets. It misses patterns that appear across hundreds, such as a category every agent handles inconsistently or a policy that is applied three different ways, and it arrives too late to be coaching.
The limitation is capacity, not intent. Reading every ticket is impossible for a lead; scoring every ticket against a rubric with notes is exactly the kind of consistent, evidence-based work a Digital FTE does well.
How does the Digital FTE score tickets?
The Digital FTE scores each resolved ticket against the rubric you define. A typical rubric covers accuracy against the knowledge base, tone against your guidelines, completeness of the answer, and adherence to policy on things like refunds and escalation. Each dimension gets a score and an itemised note quoting the relevant part of the reply.
Because the notes quote the reply and reference the rubric, a score can be challenged specifically. An agent or lead who disagrees can point to the line and the criterion, and repeated disagreements refine the rubric.
What patterns does it surface?
It surfaces patterns by agent, by category, and by time: a category where accuracy scores are low for everyone, which is usually a knowledge base problem; an agent whose tone scores drop late in the shift; a policy applied inconsistently across the team, which is a training or documentation problem.
These patterns go to the team lead in the performance view, with the evidence. The FTE does not deliver feedback to agents, and it does not rank agents publicly. Its output is for the person who coaches.
- By category: knowledge and documentation gaps
- By agent: coaching opportunities, drafted for the lead
- By policy: inconsistency that needs a decision or training
- By time: fatigue and staffing patterns
How is coaching handled?
Coaching notes are drafted for the team lead to approve, edit, and deliver in person. A note might summarise the pattern, quote two or three examples, and suggest a specific practice, but the conversation is the lead's and happens between people.
This is fixed in the design. Feedback about a person's work is a decision about a person, and those are permanently human on the ftes.ai platform. The FTE prepares; the lead coaches.
What else does the review reveal?
Full-coverage review reveals as much about the system as about the people. A category with low accuracy across all agents is a knowledge base gap, which becomes a proposed article. A policy applied inconsistently is a documentation gap or an unclear policy, which becomes a question for the support lead.
Measure with the rubric scores over time, by category and overall, and with the number of process fixes the review generated. Record the baseline from a first full month before drawing conclusions, and expect the rubric itself to need refinement from the first round of challenges.
What does the rubric look like, and how does it get better?
As a labelled example, a support rubric might score four dimensions: accuracy against the knowledge base, with the article the reply should have cited named; tone against the team's guidelines, with the phrases to use and avoid listed; completeness, meaning whether the reply answered everything the customer asked; and policy adherence, meaning whether refunds, credits, and escalations followed the rules. Each dimension has a short scale and a note explaining what each level looks like, written by the support lead in the remit.
The rubric gets better from challenges. When an agent or lead disagrees with a score, they point to the line in the reply and the criterion, and the disagreement is logged. Repeated disagreements on the same criterion mean the criterion is unclear, and the lead rewrites it. Scores from before and after a rubric change are kept apart in the performance view so trends are not distorted. Over a quarter the rubric converges on something the team recognises as fair, and the review's output shifts from scores to the knowledge and process fixes the patterns reveal.