Most managers find out a rep is done the same week they hand in their notice.
Not because the signs weren't there. Because no one was watching for them.
Call volume doesn't drop overnight. It drifts. Talk time shrinks a little each week. Outbound activity slows while the rep keeps logging in and looking busy. The data is telling you something for weeks before they say a word. You just have to know what to look for and where to look.
Here's exactly how to track it.
What you're looking for
Three signals show up in the data before a rep checks out:
Volume drop. Total calls placed and total time on the phone start declining week over week. Not a bad week. A direction.
Activity pattern shift. They're logging in. They're technically present. But outbound contacts, follow-up attempts, and new conversations are all shrinking while admin activity holds steady or climbs. They're doing the minimum to look busy.
Response time creep. Time between inbound lead and first contact starts stretching. A rep who's engaged moves fast. A rep who's checked out lets the queue back up.
None of these is a firing offense on its own. Together, trending in the same direction over three or four weeks, they're a conversation you need to have.
The quick version
Export your activity data from your CRM as a CSV. You want at minimum: rep name, date, calls placed, talk time, outbound contacts, and inbound leads worked.
Open a new chat in Claude or ChatGPT.
Upload the file.
Run one of the prompts below.
Volume trend
Analyze this data by rep. For each rep, show me their total calls placed and total talk time by week. Flag any rep whose call volume or talk time has dropped for two or more consecutive weeks.Activity pattern
For each rep, compare their outbound contact activity to their total logged activity. Show me the ratio by week. Flag any rep where outbound contacts are declining while total activity holds steady or increases.Response time
Show me the average time between inbound lead assignment and first contact attempt, broken down by rep and by week. Flag any rep whose average response time has increased over the last three weeks.Any one of these gives you something to bring into a one-on-one. All three together tell you who needs that conversation before the decision is already made.
Option 2: track it week over week in Cowork
The quick version works. But a single export only tells you where someone is right now. Catching the trend means reading the same data week after week with the same context. That's what the Cowork setup is for.
What you're building
A Cowork project connected to a folder where you drop your weekly CRM exports. Claude reads the most recent file by default. Once you have a few weeks stacked, it can compare across all of them and show you the direction, not just the snapshot.
Step 1: Create the folder
Create a folder on your computer or in Google Drive called something like "Rep Activity Exports." Drop all your existing weekly exports in. Name them consistently by date: 2026-05-25-activity.csv works. Consistent naming is how Claude knows which file is the most recent without you specifying it every time.
Step 2: Create the Cowork project and connect the folder
Open the Cowork desktop app and create a new project. When it asks for a folder, point it at the Rep Activity Exports folder you just created. If you're using Google Drive, go to Settings, connect your Google account, and grant access to that folder specifically.
Step 3: Add the project instructions
Paste the block below into the project instructions field at the top of your project:
You are here to help me track rep engagement signals using weekly activity exports from our CRM. Files are named by date and follow a consistent format. Always read from the most recent file unless I ask otherwise. When I ask about trends, compare across all files within the time range I specify.
The signals I care about: declining call volume week over week, outbound activity shrinking while total logged activity holds steady, and response time to inbound leads increasing.
The first time we work together, ask me what the columns represent so you have that context for every future session. Once I've told you, do not ask again.
Do not invent or estimate data that isn't in the file. If a column is missing or unclear, flag it rather than filling it in.Step 4: Drop in your existing exports
Add every weekly export you have. The more history you start with, the more useful the trend view is from day one.
What you ask from here
For the weekly check:
Pull the latest export. Show me any rep whose call volume, outbound activity, or lead response time is trending in the wrong direction over the last three weeks.For a deeper look at a specific rep:
Pull all data for [rep name] across every file in the folder. Show me their week-over-week trend on calls placed, talk time, and outbound contacts. When did the drop start?For the full team view before your one-on-ones:
Read all files from the last 30 days. Rank each rep by how their activity has trended. Who's moving in the right direction and who's slipping?Claude already has the column context from your first session. Every future check is one prompt. No re-explaining, no rebuilding context.
Run it before your next one-on-one
Pull your last two exports. Run the volume trend prompt. See if anything comes back that you weren't expecting.
If it does, that's the conversation you need to have this week, not next month.
Boots & Bots is a weekly newsletter for home service operators and sales managers. Every issue is one practical AI workflow you can run inside your business. No theory, no generic tips, just the actual setup. If you're not on the list, signup is at the top of the page.
— Liam