GAC logo

Case 17 · Auto retail · Saint Petersburg, 🇷🇺 Russia · 8 weeks

GAC: speech and end-to-end analytics for a car dealer in two months

GAC · Saint Petersburg, 🇷🇺 Russia

For the GAC car dealership we built AI analysis of every call and end-to-end analytics from channel to sale in two months, replacing selective listening with control of about 100% of target calls.


Stack

  • amoCRM
  • Whisper
  • PostgreSQL
  • Yandex DataLens
  • CallTouch

Before → After

Before

Calls were checked selectively

Management could not see which callers were serious, whether managers followed the script or where leads got lost. The off-the-shelf analytics did not keep the history of deal stages, and one marketplace multiplied in the reports.

After

~100% of target calls under analysis

Every call is transcribed and scored against 13 script criteria, each lead gets a grade of AAA, BBB or CCC, and cost per lead and sales by channel are visible live.


The numbers in detail

What moved — GAC


~100%

Target calls under analysis

Before, calls were checked selectively.

5 and 15 min

Lead response targets

Under control, with Telegram and Max alerts.

20 min

Refresh of the channel-to-sales view

A Yandex DataLens dashboard.


~8 h

Saved per week (team estimate)

For the head of sales and the marketer.


Section 01

The short version

  • The head of sales at the GAC dealership in Saint Petersburg could not see what happened on the sales team's calls. Control meant listening to a few recordings at random.
  • In two stages of about a month each we built a system that analyzes every call and shows the path from an ad channel to a sale.
  • The result: analysis covers about 100% of target calls, response targets are under control and cost per lead and sales by channel are visible live.

Section 02

Where did the project start?

The client already had a ready-made end-to-end analytics module from its phone provider, UIS. Almost nobody used it, and it had limits of its own.

  • It does not keep the history of deal stage changes, so a sales funnel cannot be built from it.
  • The same Avito marketplace arrived through several integrations with different tags and multiplied in the reports.
  • Deals were read from amoCRM fields, while ad spend came from contractors' reports. The two views never matched.

Section 03

The first stage: speech analytics

The first stage took about a month. We built a system we call "AI Head of Sales" that processes every call.

  1. Each call is transcribed with the Whisper model.
  2. The text goes through four separate analyses: is it a target call, how did the manager score on 13 script criteria, what facts did the client share and what insights does marketing get.
  3. Each lead gets a grade of AAA, BBB or CCC based on purchase timing and budget.
  4. The system fills in 12 deal fields in amoCRM and adds the call score to the card.

On top of this we set up Telegram and Max alerts (no answer within 5 or 15 minutes, no task, a failed script), a morning digest at 8:00 and PDF reports for the GAC and JAC brands.


Section 04

The second stage: end-to-end analytics

The second stage took another month. We set up a PostgreSQL database in Yandex Cloud and began loading the deal stage history from amoCRM, tasks, ad spend and call recordings into it.

A view from channel to leads, prepayments, sales and cost per lead refreshes every 20 minutes and feeds a Yandex DataLens dashboard. When the client moved from UIS to CallTouch, no report stopped: sources are a replaceable layer, and a fallback covers the numbers that stayed in UIS.


Section 05

Why did the off-the-shelf tool not fit?

  1. Our own source grouping. Every deal transition is recorded with dates from day one, and sources fold the way the business needs.
  2. A phone provider change without stopping the reports.
  3. Funnel specifics. Two types of refusal, one from the manager and one from the head of sales, sales counted by the date of the stage change and two funnels, one for GAC and one for JAC.

Section 06

What changed for the client?

  • Analysis covers about 100% of target calls instead of selective listening.
  • The 5 and 15 minute response targets are under control, and a finished report arrives every morning.
  • Cost per lead and sales by channel are visible live.
  • By the team's estimate, the head of sales and the marketer save about 8 hours a week.

The client keeps ordering improvements: a breakdown by manager, a list of deals without tasks and full quotes from calls.

Read the full write-up in the blog article.


Strategy session · 30 minutes

Want analytics like this? A free 30-minute strategy session.

We review your funnel, data sources and reports and show how to build the path from ad spend to a deal around your logic.