Skip to content

Generate a weekly customer-health digest across your accounts

Every Monday you piece together which accounts need attention — scanning emails, meeting notes, and ticket trends one customer at a time. Describe the digest you wish landed in your inbox, once, and let Cowork compile it across your whole book of business.

Stage: Cowork · For: End user, Manager, Champion · Level: Intermediate · Time: 25 min

When to use this

A CSM's hardest weekly question is "where do I spend my attention?" Answering it means looking across every account — recent activity, sentiment, open issues, renewal timing — and ranking by risk. That's a wide, repetitive, multi-source sweep: ideal Cowork territory. Describe the digest once and it becomes something you review, not something you build.

Use this when you manage several accounts and want a consistent Monday-morning read on which ones need you.

What you'll need

  • M365 Copilot license with access to Cowork (recurring/background automation needs Frontier — see the Stage 3 page)
  • Your account signals reachable by Copilot — emails, meeting notes, the account plan/CRM, support trends in OneDrive/SharePoint or Teams
  • A list of the accounts you cover (and, ideally, their renewal dates)
  • A clear sense of what "healthy" vs "at risk" means for you

Try it now — the prompt

Describe the digest as one outcome across all accounts:

Build me a weekly customer-health digest for these accounts: [list]. For each, pull
the last two weeks of activity — emails, meetings, and support trends — and give a
health rating (green/yellow/red) with a one-line reason, the single most important
open item, and any renewal in the next 90 days. Then sort the list worst-first and add
a short "where to spend your time this week" summary at the top. Save it and format it
so I can paste it into an email.

Why this works: you named the accounts and the signal window (grounds it), defined what a rating means and the per-account shape, and asked it to rank by risk with a "where to focus" header — so the output tells you what to do, not just what happened.

Step by step

  1. State the digest. Paste the task with your account list. Cowork plans the sweep — which sources it reads per account — before it runs.
  2. Let it compile across accounts. It works account by account, then ranks them. Watch the steps so you can spot an account where it found thin or stale data.
  3. Sanity-check the red and yellow accounts first. Those drive your week — confirm the reason behind each rating against what you know: 
    For every red or yellow account, show me the specific emails or notes that drove
the rating so I can verify it.
  4. Tune the rating logic. Make it match your judgment: 
    Treat "no contact in 3+ weeks" as an automatic yellow, and weight an open
escalation higher than sentiment.
  5. Make it stick. Once the format is right, schedule it: "Every Monday at 7am, regenerate this digest and email it to me." (Background automation requires Frontier.)

Screenshots

We deliberately don't ship screenshots that go stale — the Microsoft Copilot UI changes often. Follow the numbered steps above, which we keep current. Maintainers can regenerate fresh captures with the Playwright tool in tooling/screenshots/.

Make it better

Grow it from a digest into a workflow: - Add the next action. "For each red account, draft the outreach email I should send today." - Trend it. "Compare this week's ratings to last week's and flag any account that dropped a level." - Roll it up for leadership. Managers: "Summarize the team's book — total accounts by rating and the three biggest risks across all CSMs."

📚 Learn more. Recurring, process-level delegation like this is the core idea behind the community Cowork Cookbook by Sean Galliher (unofficial). See also the M365 Copilot resources hub.

Watch out for

  • A rating is a prompt, not a verdict. Cowork infers health from the signals it can read; the human read on the relationship is still yours. Use it to direct attention, not replace judgment.
  • Garbage signals, garbage ratings. If activity lives in tools Copilot can't see, the digest is partial. Note the blind spots so a quiet account isn't mistaken for a healthy one.
  • Keep the source window tight. "Last two weeks per account" grounds it; "everything about my customers" produces mush. Scope makes the difference.

Where this leads (the ramp)

You've delegated a recurring, multi-account process. When the whole team needs the same Monday digest in the same shape, you've outgrown describing it each time — Stage 4 · Agent Builder lets you package the accounts, signals, and rating rules into an agent everyone can run.

Next: Agent Builder → Build a team-knowledge agent over a SharePoint site

Was this page helpful?

Found an issue or want to suggest an improvement? Open a GitHub issue  ·  Star on GitHub