Daily Snowflake KPI digest in Slack | Narrative executive summary

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.

Additional information

What does this prompt do?

Pulls your core executive metrics from Snowflake every weekday morning: new signups, activations, MRR and ARR, churn, paid conversion, and top customers by usage.
Writes a short narrative summary instead of a table dump, so leaders read the why behind the numbers, not just the numbers.
Flags anomalies, week over week deltas, and any metric crossing a threshold so nothing important slips by.
Posts a single, scannable Slack message to the channel of your choice at 9am ET, in your team's tone.

What do I need to use this?

A Snowflake account with permission to read the tables behind your business metrics.
A Slack workspace and a channel where the digest should post.
Your own SQL for each metric you want included, or a starting set you can tweak.

How can I customize it?

Swap in your own SQL for each metric, including the trailing comparison window (yesterday, prior day, same day last week).
Change the time, frequency, or Slack channel the digest posts to.
Tune the anomaly thresholds and which metrics get a callout when they move.

Frequently asked questions

What happens if one of my SQL queries fails?

The digest still goes out. Any metric that errors is skipped, and the message notes at the bottom which ones did not run. A typo or a renamed table will never block the whole report.

Do I need a specific Snowflake schema or table layout?

No. You provide the SQL for each metric, so you can point it at whatever signups, revenue, churn, or usage tables you already have.

Will this work if my warehouse is BigQuery or Postgres?

This prompt is built for Snowflake. If your data lives in BigQuery, Postgres, or Redshift, look for the equivalent digest prompt for that database.

Can it post to more than one Slack channel?

Yes. Add the channels in the workflow settings and the same digest will post to each one.

Is it really a summary, or just numbers with extra steps?

It is written as a short narrative, like a chief of staff note. The headline gets a sentence or two of context, supporting metrics sit in bullets, and anomalies are flagged inline.