Book a Demo Try Speak AI Free Login
Case study · Insurance & financial services · An insurance brokerage

How an Insurance Brokerage Coaches Every Call Without a Sales Manager

An insurance brokerage with no sales manager moved call grading out of a general AI chat, now records client calls, and scores each one automatically, including how it was delivered, so owners can coach without listening to every call.

Book a DemoTry Speak AI Free
As of October 2026Measured from Speak usage records. The account is young, so this reflects early use.
  • 98+
    Calls recorded and scored
  • About 33
    Hours of calls
  • 89
    Fields applied to each call
  • 12
    Automations running

Summary

The brokerage had built its own short scorecard and was grading call transcripts by pasting them into a general AI chat. Supervisors still listened to the same calls, and there was no budget for a sales manager.

The team moved to Speak. Calls are recorded or uploaded, transcribed, and scored through fields and automations, with a score that reflects delivery as well as the words spoken.

In the first two months the team recorded close to a hundred calls and ran hundreds of automation runs. The scorecard now runs the same way on every call.

A smiling woman wearing a headset sits at a tidy desk with a laptop in a bright small office.

Three lines to take away

Before

Calls were graded by pasting transcripts into a general chat while supervisors listened to the same calls again.

With Speak

Each call is recorded, transcribed and scored by a repeatable automation that also reads delivery.

After

The owners get a sales-management function from the record, with no extra hire.

A scorecard that only worked on transcripts

The brokerage cared about how calls sounded, but its tools could only see what was said.

An older man in plain over-ear headphones frowns in concentration at a cluttered desk by lamp light.

Words without tone

A transcript cannot tell a warm greeting from a flat one. Two calls with identical wording could land very differently with a client and still look the same on paper.

Grading by copy and paste

Each transcript was pasted into a general chat by hand. Results shifted with the wording of the request, so scores were hard to compare week to week.

Supervisors relistening

Supervisors listened to the same calls the chat had already graded, so the review effort was doubled.

The cost: The old way cost supervisor listening time, and the owners still had no steady view of how calls were delivered.

The turning point

Warm and flat, scored differently

The turning point came when the team ran two calls with the same script, one delivered warmly and one delivered flatly. The old method scored them the same. The new score separated them.

Before that, the owners needed to see the score line up with their own judgment. They compared Speak's scores with their own sense of which calls had gone well, and kept adjusting until the pattern fit. The question was never whether a machine could grade a call. It was whether the grade would match the way the people who know their producers would grade it.

How they use Speak AI

Calls come in, get scored the same way every time, and sit in a folder the owners can ask questions of.

  1. Step 1

    Capture

    Calls are uploaded to the web app or collected through recorder links, so capture does not depend on one phone system.

  2. Step 2

    Transcribe

    Each call is transcribed with speakers labelled so the producer and the client are kept apart.

  3. Step 3

    Score

    A large set of fields and a dozen automations apply the scorecard to every call, including delivery as well as wording.

  4. Step 4

    Coach

    Owners review scores and flagged calls, then use chat across a folder to look for patterns before a coaching conversation.

Honest friction: Fields and automations took guided sessions to get right, and the scorecard is still being tuned. The account is young and one person does the uploading, so the team has not yet spread the habit across everyone. Early numbers show a working routine, not a finished programme.

A young man in over-ear headphones reviews a call on a laptop at a wooden desk with a glass of tea.

Book a Demo

Results

A working routine in the first two months

In about two months the brokerage recorded close to a hundred calls, roughly a third of an hour of audio each on average, and has run hundreds of automation runs on them.

950+
Automation runs in the last 90 days
13
Recorder links set up
19
Calls captured through recorder links
About 2 months
Since the account started
How we calculated this

An estimate, not a measurement

The usage numbers on this page are measured from Speak records. This block is a separate estimate of the manual effort the same workload would take. It uses published time benchmarks at the low end of their ranges and official wage statistics at the 25th percentile. Most teams already have transcripts, so Speak’s value here is the analysis layer.

Headline for this study

Starting from: already using a transcription service

Counts only the analysis, grading and notes effort. The conservative baseline.

At least 15 hours
about US$570 at the low end

Up to about 45 hours and US$2,100 at the high end of the published ranges.

Alternative baseline

Starting from: no transcription today

Baseline (a) plus manual transcription. The upper bound.

At least 120 hours
about US$3,200 at the low end

Up to about 240 hours and US$7,700 at the high end of the published ranges.

Which baseline we lead with: The team was already pasting call transcripts into a general AI chat, so transcripts existed and the conservative baseline (a) leads.

Task Manual time (published range) With Speak (our assumption)
Listening to and scoring each call 1 to 2 hours per hour of calls
Survey Results (scoring time vs AHT)		 0.5 hours per hour of calls
Our assumption: 0.5 hours per hour of calls to check and correct an AI score.
Manual transcription (baseline b only) 4 to 7 hours per hour of calls
Realities Toolkit #08: Transcribing your own qualitative data		 0.25 hours per hour of calls
Our assumption: 0.25 hours to check an AI transcript.

Workload (measured): 30 hours of calls. Live, non-deleted recordings for the account, with a 12-hour cap per recording and zero-length rows ignored, pulled 2026-10-11.

Hourly cost: US$26.90 (25th percentile) to US$33.41 (median) for First-line supervisors of office and administrative support workers; transcription is costed at US$16.42 to US$19.43. Both are multiplied by 1.43 to include benefits and employer costs. US wages are used as a neutral reference.

Hours saved = workload x (manual hours – hours with Speak). Value = hours saved x hourly wage x 1.43.

Young account with about two months of data. Equivalent manual effort, not a measured saving, and no accuracy claim.

As of October 2026. This estimate is not customer-reported, is not audited, and does not include software costs.

What the team found, in short

Paraphrased from our own conversations with the team. Not quotes.

Tone and warmth change how a client feels about a call, and a score that only reads words will miss both.

Paraphrased from our calls with the team

A score earns trust when owners can check it against their own judgment of the calls they already know.

Paraphrased from our calls with the team

Small brokerages can run a sales-management routine from a repeatable score instead of waiting until they can afford a manager.

Paraphrased from our calls with the team

What is next

An older woman and a younger woman chat over coffee on a sofa in a bright office lounge.
  • Coaching by producerTeams like this group scores by producer to see who needs help with openings, objections or follow-ups.
  • More producers recordingRecorder links make it simple to extend capture beyond the first person uploading.
  • Tighter scorecardsWith more calls in the folder, owners can refine fields and prompts against the calls they trust most.

More customer stories

Customer name withheld at their request. Figures are from Speak usage records, as of October 2026.

Coaching producers without listening to every call?

Talk to us about scoring client calls on delivery as well as wording.

Book a DemoTry Speak AI Free