HomeFind your caseNobody has time to review our sales calls
Case · Enterprise
A common situation · not a client storyNobody has time to review our sales calls.
Sales and support teams that record their calls but only ever listen to a handful, so coaching stays generic and patterns across the team go unnoticed.
What you're probably seeing
If this sounds familiar, you're in the right place.
- Calls are recorded, but managers listen to a few a week at most.
- Coaching is generic, because nobody can point to the moment a call went wrong.
- Patterns across the team, like a common objection or a weak opening, go unnoticed.
- Call notes in the CRM are short, late or missing.
- Recordings contain personal data, so sharing them for review feels risky.
The usual ways out
What teams usually try. Each works somewhere. None works everywhere.
| The way | Works when | The catch |
|---|---|---|
| Managers listen to more calls | The team is small | Doesn't scale, and the sample is never random |
| Buy a conversation intelligence tool | Your process fits the tool's view of a sales call | Per-seat cost, and scoring you can't change |
| A transcription script and a spreadsheet | You want a quick look at a few calls | Masks names in text but not in the audio, and one failure stalls everything |
| Peer review sessions | You want the team to learn from each other | Takes selling time, and depends on who shows up |
| A scoring system built on your own rubric | You know what a good call looks like for you | Needs your rubric written down first, and care with personal data |
| Score only the calls that matter | A few call types drive most revenue | Needs a way to pick those calls reliably |
What other companies did. Their published results, not ours.
We read each company's own write-up and asked one question: would our diagnostic's tests have led to the same call? We judge the approach, not the result.
| Company | What they did | What they published | Source |
|---|---|---|---|
| Intact FinancialCanadian property and casualty insurer. Support call centres handle up to 20,000 calls a day in English and Canadian French. Quality agents previously picked calls to audit by hand. | Built an in-house call-quality system on AWS: transcription with Amazon Transcribe, chosen partly for its accuracy in English and Canadian French, then their own models for speaker roles, sentiment, removing personal data from transcripts, script adherence and call outcome. Results go to a search tool and dashboards for auditors. | 1,500% more calls reviewed (15 times as many per auditor). 1,500% faster auditing. Auditors 65% more efficient. 10% less agent time per call and 10% shorter hold time, credited to targeted coaching. No timeframe stated. | Unlocking insights and enhancing customer service: Intact's transformative AI journey with AWSOct 2024 · cloud provider story |
| Rocket MortgageUS retail mortgage lender with over 2.6 million clients. 65 million voice interactions totalling 147 million minutes across banking, operations and servicing. | Built a call transcription and summary tool on AWS: transcription with personal data removed from transcripts, sentiment and entity extraction, a fine-tuned model for call classification, and a separate LLM for summaries, shown in an internal app. | Automated post-call wrap-up projected to save about 40,000 team hours a year. First-call resolution up nearly 10% (about 20,000 team hours a year). Built in 10 days. First rollout covered 30,000 servicing calls, then scaled 4x for operations and 6x for banking.They noted: They name scaling and bursts of call volume as challenges, handled with serverless components. Their stated approach was to start small and then scale. | Transforming home ownership with Amazon Transcribe Call Analytics, Amazon Comprehend, and Amazon Bedrock: Rocket Mortgage's journey with AWSSep 2024 · cloud provider story |
| MintelLondon market research firm with 1,000-5,000 employees and over 6,000 clients. Sales team size not stated. | Bought Gong. Reps tag managers on calls so managers can see what is happening, reps self-coach from talk-ratio and listening stats, and alerts flag stalled deals. Call snippets can also be sent to clients to confirm agreement on a project. | 34% increase in win rate 'since adopting Gong'. No baseline or period stated.They noted: Call data showed reps were reaching too few buyer contacts per deal, so they set a policy of about four contacts per deal. | Research, recommendations, and reality: How Gong helped Mintel increase win rates by 34%Undated · vendor story |
| ZapierZapier's own sales and success teams, which already record calls in Gong. Team size not stated. | The enablement team built an automation that sends each Gong transcript to ChatGPT after the call. It flags the call as needing improvement or effective, rates up to three objections and product knowledge with reasons, and logs the results in a table. | Not stated. | Use AI to provide sales coaching and call prepMar 2025 · vendor story |
Our read
Reasonable call. One thing showed up late. These systems scaled call review, but the write-ups describe masking personal data in transcripts and say nothing about the audio.
Intact went from hand-picked audits to reviewing 15 times as many calls per auditor, and chose its transcription engine for accuracy in English and Canadian French, which matches our transcription test. Intact and Rocket Mortgage both describe removing personal data from transcripts, and neither says whether the recordings themselves are masked. Mintel's managers review reps' recorded calls, and Mintel can also send call snippets to clients, so where personal data sits in the audio matters there too. Zapier kept Gong for recording but wrote its own scoring criteria on top, which is the 'own rubric' option in the table above, though the post does not publish the rubric.
What we would test first: Find where personal data appears in your recordings, and whether it must be masked in the audio as well as the text.
What the diagnostic would test
The tests that rule the wrong ways out.
On your own systems and data, read-only first. The number of tests sets the diagnostic's price, quoted before it starts.
- 1
How many calls you record, and how many anyone reviews today.
- 2
What a good call looks like for you, written as a rubric.
- 3
Transcription quality on your real recordings, in your languages and accents.
- 4
Where personal data appears, and whether it must be masked in the audio as well as the text.
- 5
Which tools the results must reach: your CRM, Slack or a coaching view.
This week, without us
Three things you can do yourself.
- 1
Pick five recent calls, one won and four lost, and listen for where each one turned.
- 2
Write down the five things a good call must do, in plain words. That's the start of a rubric.
- 3
Check what your recording tool exports: audio, transcripts, or both.
Sound like you? Bring it to a 30-minute call. We'll tell you which tests your case needs, and what they cost.
This page describes what teams in this situation commonly face. It is not a client story.