Daily executive KPI digest in Slack from Snowflake

Every weekday at 9am ET, post an executive KPI digest to a Slack channel I configure. The output should read like a short chief of staff note, not a wall of numbers. The audience is the leadership team scanning Slack with coffee.

Use the Snowflake Execute SQL Statement operation to pull the core metrics I care about over the last 24 hours and a trailing comparison window (the prior 24 hours and the same day of the previous week). Default metrics: new signups, activations, MRR and ARR, gross and net churn, paid conversion rate, and the top 3 to 5 customers by usage. Each metric is its own named query in workflow settings so I can swap in my own tables and metric definitions without editing the agent.

For every metric, compute the 24 hour value plus day over day and week over week deltas. Then reason about what moved and why it matters. Flag anomalies: anything that crossed a configured threshold, anything that moved more than 2x the recent norm, or anything where the direction changed (for example churn flipping from negative to positive, or paid conversion dropping below floor). Call out the top customers by usage and any newly large accounts.

If any individual query errors (warehouse cold, table renamed, permission denied, syntax error), skip that metric and keep going. Do not fail the whole digest because one query broke. At the bottom of the Slack message, list the metric names that failed so I can fix them, but never paste raw SQL errors or stack traces into Slack.

Use the Slack Bot Send a Message operation to post a single message to the configured channel. Format with Slack mrkdwn. Structure: a one line headline with the most important number, two or three sentences of narrative on what changed and why it matters, then a short bullet list for supporting metrics with their deltas, then an anomalies section if anything is worth flagging, then the top customers section, then the skipped-metrics footer if any. Keep the whole thing under roughly 25 lines. If everything is steady, say so in one line and stop. Do not paste tables or raw rows.

Configuration to expose at setup time: the Slack channel ID, the list of named SQL queries (with the comparison window encoded inside each query), per-metric anomaly thresholds, the Snowflake warehouse and role to run the queries under, and the timezone for the schedule. Default to America/New_York and weekdays only.