Customer Health Summaries in Slack
Use a playbook to pull customer account metrics from your database into Slack.
Connect the customer data sources
Connect Snowflake for usage and revenue data, Intercom for customer conversations, and Zendesk for support tickets. Opulent reads these to build a unified health view for any account.
Verify the connection by asking Opulent to list the tables or endpoints it can see and to run a one-account test query. If it returns the right schema, the health summary path is live.
Common sources: - Snowflake: users, accounts, events, subscriptions, MRR. - Intercom: conversations, tags, last contact, satisfaction. - Zendesk: tickets, resolution times, open issue count. Test: "For Acme Corp, list the tables you can use and the fields you would need for a health summary."
Use a read-only database user for Snowflake. Health summaries should never write back to the warehouse; if a summary needs to update a CRM, route that through a separate, reviewed playbook.
Create the !customer-health playbook
Create a playbook named !customer-health, the ! prefix turns it into a macro you can invoke from Slack. The playbook defines the metrics to pull and the format of the output.
Keep the playbook general: Opulent figures out the right tables and queries from your schema, but the metric list and format stay consistent across accounts.
Playbook: !customer-health When given a customer name, find and report: 1. Monthly active users, 6-month trend with month-over-month change. 2. Feature adoption, features used in the past 30 days and distinct users per feature. 3. Support tickets, open vs resolved in the past 30 days, with average resolution time. 4. Revenue, current MRR and 6-month trend. 5. Conversations, recent Intercom threads and any risk signals. Format: - One-line health verdict (Healthy / At Risk / Needs Attention). - A table with MAU trend, feature adoption count, open tickets, and current MRR. - A short list of risk factors or positive signals from the data.
Run it from Slack on demand
Make sure Opulent is invited to the channel where the team discusses accounts, then invoke the playbook with a customer name.
Opulent queries the connected sources, runs the playbook, and replies in-thread with a formatted summary. Follow-ups stay in the same thread.
The sharp edge: a common name like 'Acme' can match multiple accounts. Always use the exact account name or domain from the CRM so Opulent resolves the right record.
@Opulent !customer-health Give me an account summary for Acme Corp. Follow-up: @Opulent For Acme Corp, show me the daily active users for the past 30 days. Flag any days where DAU dropped below 50% of the 30-day average. @Opulent Compare Acme Corp's feature adoption and MAU to the median for all Enterprise-plan accounts. Where are they lagging?
Make it routine
Once the playbook works for ad-hoc lookups, schedule it to run automatically: weekly summaries for top accounts posted to #customer-success, or pre-meeting packets before quarterly business reviews.
Scheduled summaries keep the team ahead of churn instead of reactive. The same playbook runs; only the trigger changes.
Schedules to add: - Monday morning: top 10 account health summaries to #customer-success. - Pre-QBR: summaries for all accounts in the meeting calendar. - Renewal risk: weekly summaries for accounts in the final quarter of their contract.
Sharpen the health summary loop
When the health verdict repeatedly misses something (a support escalation, a product change, a seasonal dip) add the signal to the playbook or write the correction into memory (the notes a run recalls next time).
Tighten the thresholds as you learn: a 5% MAU drop may be noise for a small account but a warning for an enterprise account. Segment the rules so the verdict matches the account tier.
The natural chain: when a health summary flags an at-risk account, hand it to Autonomous Triage for Knowledge Work Queues to route the right follow-up, or to Data Analysis as an Agent-Run Workflow for a deeper root-cause investigation.