Call analysis

Every recorded call reaches TopClosers, gets transcribed and is scored against your agency’s sales manual. The result is a per-call report: a score from 0 to 10, the section-by-section breakdown, what went well, what to improve, the objections that came up and the moments worth listening to again.

The owner sees the analysis of every call in the agency; each closer sees only their own.

What it is

Overall score

From 0 to 10. The AI does not eyeball it: it is computed from the per-section scores and their weights.

Section breakdown

Every phase and every general criterion with its own score, its met criteria and its own coaching.

Lead context

Temperature, motivation, expressed pains and objections, backed by verbatim quotes from the conversation.

How it works

1. The call ends and the recorder (Fathom or Bluedot) sends the recording to TopClosers. It usually takes a few minutes.

2. The closer is identified by the email the call was recorded with. If it matches nobody on the team, the call lands as Needs attention and analysis waits until the owner assigns the right person.

3. It is scored against the active rubric. The call moves through the Analyzing… state, which normally takes a couple of minutes.

4. The report is ready. The call becomes Analyzed and the closer gets an email with the score and the points to improve.

You can trigger it yourself

If a call was never scored, the owner has “Analyze now” on the call’s detail view. If it was already scored and you want it done again, use “Reanalyze”.

What you see

The analysis is organized in tabs inside the call’s detail view.

Summary

The landing view: the overall score, the Scorecard with every section at a glance, What to improve (the most important part of the call, pulled to the top), What went well, the Key moments and the full transcript.

The closer

Performance by section: every phase of your manual and every general criterion, with its score out of 10 and its weight. Inside each section you get:

• What worked — what the closer did well in that stretch.
• To improve — the concrete advice, with phrasing they could have used. It is the longest part of the report on purpose.
• View criteria — the rule-by-rule breakdown of your success criteria, each marked Met, Partial or Not met, with the transcript quote that proves it.
• Metrics specific to the section, such as its duration.

General criteria carry the Whole call label, because they are judged over the entire conversation rather than one stretch of it.

The lead

The profile of the person on the other side: lead type, temperature (hot, warm or cold), awareness level before and after the call, their primary motivation, the buying trigger (what they said versus what really drives them) and the expressed pains with their severity. All of it backed by verbatim quotes.

Objections and plan

Every objection that surfaced — including the soft ones, the “I need to think about it” kind — with how it was handled, a handling quality score from 0 to 100 and what to say next time. Below that, the prioritized action plan for that specific lead.

Objections are named with short, reusable labels (“Price”, “Needs to check with partner”) so they can be aggregated later in Objections.

Key moments

Between two and six moments with a timestamp, the verbatim quote and why they matter. They let you listen to just the part you care about instead of the whole recording.

How the score is computed

The overall score is not an impression. Each section is scored from 0 to 10 by evaluating its criteria one by one, and the overall score is the average of those scores weighted by each section’s weight. If Discovery weighs 25% and Close weighs 30%, that is exactly what they weigh in the result.

Two practical consequences:

• Changing weights in the rubric changes the scores of future calls, not the ones already analyzed.
• The score you see and the weights you see always agree, because each analysis stores the weights it was computed with.

Analysis language

Every piece of text the AI produces — summary, per-section coaching, strengths, improvements, section names — is written in the agency language, even when the call or the manual is in another language. Verbatim transcript quotes stay in their original wording so the evidence remains faithful.

Change the language in agency settings.

Call states

• Needs attention — it is not known which closer took the call. Analysis is on hold until one is assigned.
• Waiting for analysis — the call arrived and is queued.
• Analyzing… — it is being scored. A couple of minutes, usually.
• Analyzed — the report is ready.
• Analysis failed — something went wrong while scoring.

You can filter the list by state from Calls.

Re-analyzing a call

The Reanalyze button scores the call again against the rubric that is active today and replaces the previous report. That is what you want when:

• You changed the criteria and want to see that call under the new yardstick.
• The analysis failed.
• The report came back incomplete — no lead profile or no objections, for example.

It replaces, it does not add

Reanalyzing overwrites the report: the previous analysis is not kept. And because it uses the current rubric, the score can go up or down without the closer having done anything different.

Calls left out of scoring

Sometimes a call should not count: an internal demo, a support conversation, a call that dropped after two minutes. That is what Don’t grade is for.

When you exclude a call from scoring:

• Its score is hidden and a Not graded badge appears.
• It stops counting toward averages, rankings and the closer’s trend.
• It keeps counting toward cash, outcomes and volume: if that call closed a deal, the deal stays in Statistics and Commissions.

It is reversible: Grade this call restores the score and its effect on averages.

Things to know

• Analysis never waits on the CRM appointment. The link between a call and its GoHighLevel or Calendly appointment only enriches outcome reporting. If it is missing, the call is analyzed anyway.
• It does depend on the manual. With no active manual there is no rubric to score against.
• It does depend on knowing who took the call. While it sits in Needs attention, nothing is scored.
• AI analysis is part of the paid plans; see Plans and billing.
• An analysis can fail on a transcript that is too short or unusable, or on a one-off error while scoring. It never affects the recording and is almost always fixed by retrying.

Best practices

• Read “View criteria” before the score. The rule-by-rule breakdown tells you whether the score dropped because the closer missed something or because the criterion is badly written.
• Use key moments in team sessions. A thirty-second clip with context convinces more than a whole report.
• Mark atypical calls as “Don’t grade” as soon as you spot them, so they do not drag a closer’s average for weeks.
• Do not re-analyze in bulk after every tweak. Change the rubric, let a few calls run, then compare.
• Work on repeated improvements, not one call’s. The pattern is easier to see in Coaching.