How I Built a Fully Automated KPI Dashboard in 4 Hours (Stack + Prompts Inside)
Most dashboards are lying to you.
Not because the data is wrong ā but because it's stale, unconnected, and manual. You refresh it once a week, you copy-paste from Stripe, you summarize it in a doc nobody reads.
Here's how to actually fix that. This is the exact stack I use, and it's running for multiple clients right now.
---
The Stack (All Free or Low Cost)
Tool | Role |
|---|---|
Airbyte | Pull raw data from Stripe, GA4, HubSpot, etc. |
BigQuery | Warehouse ā store and query everything |
dbt | Transform raw data into clean KPI tables |
Looker Studio | Render the live dashboard |
n8n | Automation orchestration (cron ā trigger ā alert) |
OpenAI API | Generate the KPI narrative summary automatically |
Total cost to run this for a $1M ARR business: ~$80ā120/month.
---
The KPI Narrative Prompt (Copy This)
Paste this into your n8n ā OpenAI node to auto-generate your weekly summary:
You are a business intelligence analyst. Given the following KPI data for the week of [DATE], write a 3-paragraph executive summary.
Paragraph 1: Performance vs. last week (highlight wins and drops above 10%)
Paragraph 2: Root cause analysis for any metric that moved significantly
Paragraph 3: One specific action the team should take this week
KPI Data:
[INSERT JSON FROM BIGQUERY]
Tone: direct, data-driven, no fluff.Drop this into your weekly automation run and you'll have a board-ready summary in your inbox every Monday.
---
The 3 Automation Failures That Kill Dashboards
1. Schema drift ā your source changes a field name and your pipeline silently breaks. Fix: add schema validation in Airbyte with alerting on.
2. Credential rot ā API keys expire and n8n fails silently. Fix: set calendar reminders to rotate keys, and add a "last successful run" monitor.
3. Looker Studio caching ā your dashboard shows yesterday's data. Fix: force 15-min refresh intervals and cache-bust with a timestamp field.
---
This is the kind of content we ship every week inside the AI Analytics Dashboard Playbook.
If you're building automated KPI infrastructure, drop a comment with your current stack ā happy to give feedback.
