Coach your team with call scores and sentiment

Work the coachable-calls queue, compare extensions, and read sentiment trends without misreading a blank average.

Insights > Coaching turns finished call analysis into a short list of conversations worth reviewing with a person. Start from the coachable-calls queue, listen to the call, then use the per-extension views to decide whether one conversation was an outlier or a pattern.

Who can see coaching data

Coaching is a manager-level view. Signed-in colleagues with a narrower role can open the page, but the per-extension leaderboard and the sentiment scatter tell them the view requires manager access, the sentiment cards never resolve, and the low-scored list reports that it could not load. That is the access boundary doing its job, not an outage — ask an administrator to review the role rather than retrying.

Read the coaching cards

Choose a range first (Last hour through Last 90 days; the page opens on Last 30 days). The selection is written to the page address so a shared link reopens the same window.

Sentiment averages only include calls whose sentiment value carries a numeric score. A call coded with a sentiment value that has no score still appears in the distribution charts but cannot move an average.

Work the coachable-calls queue

1. Select the Coachable calls card. The list opens with the range you were already using, and the range stays in the address so you can share the exact queue.
2. Review the rows worst-score-first. Each row shows the score, the caller, the one-line call summary, and the call length.
3. Select Listen to open the call. The Scoring tab holds the classification badges, the coaching tips, and — behind View detailed scores — the per-criterion score, the reasoning, and the quotes from the conversation the score was based on.
4. Coach from the criterion that lost the most weight, not from the overall number. The detailed view shows each criterion’s weight, so a 20-point criterion missed is a different conversation from a 5-point one.
5. Return to the queue and repeat. The list shows up to 50 calls at a time; when more match, the page says so — narrow the range to work through the rest.

Coaching tips and per-criterion reasoning are produced by the same analysis pass that scores the call, so the tips a call carries follow from the scorecard that matched it. If the tips are not useful, change the scorecard’s criteria rather than the call.

Compare extensions

• Low-scored calls is the short version of the same queue, capped at ten rows, for a quick scan without leaving the page.
• Leaderboard lists each extension with calls handled, talk time, average employee sentiment, average customer sentiment, and conversion rate. Select any column header to sort. An extension with no sentiment-bearing calls shows a dash rather than a zero, so sorting by sentiment does not bury it at the bottom.
• Calls by extension stacks each extension’s calls by overall sentiment, for the busiest extensions in the range.
• Sentiment trend plots employee sentiment as a solid line and customer sentiment as a dashed line over the range. Days with no sentiment data render as gaps, never as zeros.
• Sentiment vs. duration plots one point per call, up to the 500 most recent in the range. Long calls low on sentiment usually mean friction; short calls high on sentiment usually mean an efficient answer. The panel also calls out the longest calls and the lowest-sentiment calls directly, and each of those opens the call.

Expected result

The coachable queue names specific calls with specific reasons, the leaderboard shows where those calls concentrate, and every number on the page reflects the range shown in the selector.

Troubleshooting

• The leaderboard or scatter says it requires manager access: Your role does not include manager-level reporting. Ask an administrator for that phone system.
• Coachable calls is far smaller than low-scored calls: Only low-scored calls that produced at least one coaching tip are coachable. A call scored low with no tip stays in the low-scored count.
• Both counts are zero but calls are clearly being answered: No scorecard matched those calls, or none has finished analyzing. Check that an active scorecard targets that direction, duration, and category.
• A sentiment average is blank while the charts show sentiment: Those calls are coded with a sentiment value that has no numeric score. Values loaded from the standard set carry scores; values created by hand do not.
• The coachable count shows a dash: The count failed to load rather than returning zero. Use Refresh; the card intentionally withholds its link until it can trust the number.
• A call has a score but no coaching tips: Tips are produced only when the analysis has something concrete to suggest. Nothing is missing.

Related guidance

• Score calls with scorecards.
• Read the Today, Overview, and Revenue dashboards.
• Configure call categories, sentiment, and handlers.
• Review call recordings, transcripts, and AI analysis.
• Use call history and insights.