If you run customer success at a SaaS company, you already know the problem: you have data showing which customers are healthy and which aren't — it's sitting in your database — but actually getting to it requires a SQL query, a data analyst, or a BI tool that takes three weeks to set up.
Customer health scores shouldn't be this hard. Here's how to build one directly from your database without writing a single line of SQL.
What Is a Customer Health Score?
A customer health score is a single metric that tells you, at a glance, how likely a customer is to renew, expand, or churn. It typically rolls up several signals into one number or color (red/yellow/green).
Common inputs for a health score:
- Product usage frequency (logins, feature activations, API calls) - Support ticket volume and open issues - Payment history and plan tier - Onboarding completion - Time since last active session - NPS or satisfaction scores (if stored in your DB)
The challenge isn't knowing what to measure. It's pulling those numbers without a full-time analyst.
Why Customer Health Scoring Usually Fails
Most CS teams hit one of three walls:
1. They depend on engineers. Every time they want to add a new signal or adjust a threshold, they need to file a ticket and wait.
2. They use a BI tool that requires SQL. Metabase, Tableau, Looker — all powerful, all requiring someone who can write queries. That's not most CS managers.
3. They export data to spreadsheets. Which are stale the moment you open them.
The result: CS teams fly blind or wait on data that arrives too late to actually act on it.
The Better Approach: Query Your Database in Plain English
With a natural language database tool like aifordatabase.com, you connect directly to your PostgreSQL, MySQL, Supabase, or other database and ask questions the way you'd ask a colleague.
No SQL. No waiting on engineers. The query runs against your live data.
Here's how to build a customer health score step by step.
Step 1: Connect Your Database
Connect aifordatabase to your production or analytics replica database. It supports PostgreSQL, MySQL, Supabase, PlanetScale, MongoDB, BigQuery, MS SQL Server, SQLite, and others. The connection takes a few minutes — just a connection string or credentials.
Step 2: Ask Questions About Your Data
Once connected, start pulling the signals you care about. You don't need to know your schema upfront — just ask in plain English and the tool figures out the right tables and columns.
Examples of what you'd ask:
"How many times did each customer log in last 30 days?" "Which accounts haven't used the product in over 14 days?" "List customers with more than 2 open support tickets." "Show me accounts that haven't completed onboarding." "Which customers on the Pro plan have API usage below 10 calls this month?"
Each question returns a table of results instantly. No code, no SQL, no waiting.
Step 3: Build a Health Score Dashboard
Once you've confirmed which queries give you the right signals, turn them into a self-refreshing dashboard. Each widget on the dashboard runs a live query every time it loads — so your team always sees current data, not last week's export.
A typical customer health dashboard might show:
- Accounts by last active date (sorted oldest first) - Customers by login frequency (30-day trend) - Open support ticket counts by account - Onboarding completion rate by signup cohort - Accounts at risk (below your usage threshold)
This becomes the CS team's daily working view. No refresh button, no spreadsheet, no ticket to engineering.
Step 4: Set Up Automated Alerts for At-Risk Customers
The dashboard tells you who is at risk. But waiting for someone to check it is still reactive. The better move: set up automated workflows that fire when a customer crosses a threshold.
With aifordatabase's action workflows, you can configure rules like:
"If a customer hasn't logged in for 10 days, send a Slack alert to their CS owner." "If a Pro plan account drops below 5 logins in 7 days, trigger a re-engagement email." "If open support tickets for an account exceeds 3, notify the account manager." "If an account's API usage drops more than 50% week-over-week, create a Slack thread."
These workflows run continuously against your database. The CS team gets proactive signals instead of discovering churn after the fact.
What Signals to Prioritize
Not all health score inputs are equally predictive. If you're starting from scratch, prioritize these in order:
1. Login frequency — the strongest leading indicator for most SaaS products. A customer who logs in regularly is not churning. 2. Core feature activation — did they use the feature that delivers your product's core value? If not, they haven't seen the point yet. 3. Breadth of usage — are they using 1 feature or 5? More breadth = more embedded. 4. Support burden — high ticket volume is a yellow flag; high tickets + low usage is a red flag. 5. Time since onboarding completed — customers who didn't finish onboarding churn at much higher rates.
You don't need to build a weighted scoring algorithm on day one. Start with a simple "at risk" filter: customers who fail 2 or more of these signals need attention.
Who This Is For
This approach works best for:
- CS leads at SaaS companies with 50–500 customers - Ops managers who own customer data but don't have analytics support - Founders doing their own CS before they can hire a dedicated team - Teams that already have a database but no BI tool or data analyst
If you're using Mixpanel, Amplitude, or Segment and just want event-level product analytics, those tools have their place. But if your source of truth is a PostgreSQL or MySQL database — and most SaaS companies' is — querying it directly is faster and more accurate than syncing to a third-party tool.
Common Questions
What databases does this work with?
aifordatabase.com supports PostgreSQL, MySQL, SQLite, Supabase, MongoDB, PlanetScale, MS SQL Server, BigQuery, Amazon Redshift, Snowflake, and Neon. If your data is in any of these, you can build a health score dashboard without SQL.
Do I need to set up a data warehouse?
No. You connect directly to your existing database. If you have a read replica (recommended for production databases), use that — otherwise a direct connection works fine for most query volumes.
How is this different from using Metabase or Looker?
Metabase and Looker both require SQL knowledge to build queries. You can use their UI for simple bar charts, but anything involving filtering by customer, joining tables, or calculating trends requires SQL. aifordatabase is built for non-technical users — the SQL is generated automatically from plain English.
Can I export the health score data?
Yes. Query results can be exported to CSV. You can also use the action workflow feature to push alerts to Slack, send emails, or trigger webhooks to tools like HubSpot or Salesforce.
How long does it take to set up?
The database connection takes 5 minutes. Building your first dashboard — with 3–4 health signal queries — typically takes under an hour, even if you've never queried a database before.
Get Started
If your customer data is in a database, you already have everything you need to build a health score. You just need the right interface to get it out without SQL.
Connect your database at aifordatabase.com and build your first health score dashboard today. No SQL required, no analyst needed, no spreadsheet exports.