SaaS MRR Analytics Dashboard: dbt + Stripe Integration

Results at a Glance

90%

Reduction in reporting time

8 hours → 45 minutes

100%

Confidence in numbers

Automated & tested

$12k

Expansion MRR identified

In first month

18.5x

Growth visibility

$270 → $5,010 MRR tracked

DuffDash Analytics team of Simpsons characters including Mr. Burns, Lisa Simpson, Homer Simpson, and other Springfield residents

Meet the DuffDash Analytics customer base: 35 Springfield businesses, from Mr. Burns' nuclear plant to Lisa's education nonprofit

"What's our MRR this month?"

It was a simple question from the CEO. But for DuffDash Analytics, a fast-growing SaaS startup in Springfield, answering it meant:

Logging into Stripe dashboard
Exporting subscription CSV files
Opening Excel and praying formulas work
Manually categorizing seat changes as "expansion" or "contraction"
Hoping the VLOOKUP didn't break again
Texting the answer 2+ hours later: "Approximately $4,800... I think?"

"We were growing 30% month-over-month, but had zero visibility into why. Was it new customers? Expansions? Reactivations? We were flying blind."
— CEO, DuffDash Analytics

The Breaking Point

The startup had hit 35 customers across three pricing tiers (Starter, Team, Enterprise) with constant seat changes. Their Stripe account was a firehose of subscription events:

New Signups

Principal Skinner's school district, Dr. Hibbert's medical practice, Mayor Quimby's city hall...

Seat Expansions

Lisa Simpson grew from 2 to 10 seats. Mr. Burns expanded from 10 to 30 Enterprise seats...

Churn & Contraction

Homer canceled (too confusing). Krusty's studio had budget cuts (12 → 6 seats)...

The leadership team needed answers: Which customers are expanding? Where's the churn coming from? What's driving growth? Excel couldn't keep up.

The Solution: dbt-Powered MRR Analytics

I built a subscription analytics system using dbt (data build tool) to transform raw Stripe data into business-ready MRR dashboards.

The Data Architecture

Raw Stripe Data (The Starting Point)

Customers: 35 businesses, all Simpsons-themed
Products: Starter ($10/seat), Team ($30/seat), Enterprise ($50/seat)
Subscriptions: Current quantity and status per customer
Subscription Events: 75+ events tracking every seat change, cancellation, reactivation

Data lived in Postgres but was a mess of timestamps and event logs. Unusable for analytics.

Analytics-Ready Output (The Goal)

Monthly snapshots: What was each subscription's quantity in each month?
MRR calculations: Quantity × price = revenue per customer
MRR bridge: Why did MRR change? (New, Expansion, Contraction, Churn)
Cohort analysis: Which signup months have the best retention?

Clean, tested, dashboard-ready data refreshing hourly.

The dbt Transformation Layer

Built with SQL and dbt, creating a maintainable, version-controlled analytics pipeline:

Stage 1: Staging Models

Clean and standardize raw Stripe tables:

stg_customers
stg_products
stg_prices
stg_subscriptions
stg_subscription_events

1-to-1 with source, type casting, renaming

Stage 2: Intermediate Models

Convert event logs into time-series:

int_subscription_monthly_quantity

Generates a row for each subscription for each month showing quantity at that time

The magic: Cross-join subscriptions × months, then use window functions to find latest event per month.

Stage 3: Fact Tables (Marts)

Business-ready analytics tables:

fct_subscription_monthly
fct_mrr_bridge_monthly

Joins everything together, calculates MRR, categorizes changes (new, expansion, contraction, churn)

Why dbt?

Version Control: Every change tracked in git. Rollback anytime.

Testing: Automated tests ensure no duplicate IDs, no nulls in critical fields, valid MRR categories.

Documentation: Self-documenting models. New team members understand the logic instantly.

Dependency Management: dbt knows the order to run models. Change one, rebuild downstream automatically.

The Dashboards: Real-Time MRR Visibility

Built with Streamlit (Python), these dashboards connect directly to the dbt-modeled data. Refresh on-demand or schedule hourly updates.

DuffDash Analytics MRR dashboard showing revenue growth from $270 to $5,010 over 14 months with customer count, MRR trends, and product tier breakdown

Executive dashboard showing 18.5x MRR growth (Jan 2025 → Feb 2026), active customer count, and revenue breakdown by product tier

What Leadership Can Now See

Current MRR: $5,010 (vs $270 at launch)
Monthly growth %: Track acceleration or deceleration
Active customers: 35 businesses, growing steadily
MRR by product: Enterprise driving 45% of revenue
Top customers: Mr. Burns ($1,500/mo), Dr. Hibbert ($600/mo), Mayor Quimby ($450/mo)

Actionable Insights

MRR waterfall: See exactly what changed each month (new, expansion, churn)
Churn alerts: Spot customers who downgraded or canceled
Expansion opportunities: Identify Starter → Team upgrade targets
Cohort retention: Jan 2025 customers still strong, June cohort had issues

Detailed MRR bridge analysis showing monthly changes categorized as new revenue, expansion, contraction, churn, and reactivation with top customer breakdown

MRR bridge waterfall chart revealing $530 in expansion MRR from Lisa Simpson, Mr. Burns, Dr. Hibbert, and Comic Book Guy in January 2026

Measured Business Impact

Within the first month of dashboard deployment, the DuffDash team made data-informed decisions that directly impacted revenue.

90%

Time Savings on Reporting

Went from 8+ hours per month (exporting CSVs, building Excel reports) to 45 minutes (checking dashboard, exporting screenshots for board deck). CFO got nights and weekends back.

$12k

Expansion MRR Identified

Dashboard revealed Lisa Simpson and Mr. Burns were Enterprise-tier usage on Team plans. Sales reached out → both upgraded → $12k additional annual revenue locked in.

6

Churn Save Attempts

Contraction alerts flagged Krusty's studio (12 → 6 seats) and Mayor Quimby (12 → 9 seats) budget cuts. Customer success intervened with discounts. Retained $180/mo MRR.

Cultural Shift: Data-Informed Decision Making

Before this system, revenue discussions were guesswork. Now:

Weekly executive reviews

CEO, CFO, Head of Sales review dashboard every Monday. MRR growth, churn, expansion opportunities discussed with real data.

Board confidence

"Investors stopped questioning our numbers. Dashboard screenshots in the board deck show professionalism and control." — CEO

Technical Stack & Implementation

Technologies Used

dbt (data build tool): SQL transformation layer, version controlled
PostgreSQL: Data warehouse for Stripe data
Streamlit: Python-based dashboard framework
Plotly: Interactive charts (MRR growth, waterfall, product breakdown)
Docker: Containerized Postgres for local development

Why This Approach Works

Maintainable: SQL models, not brittle Excel formulas
Testable: dbt tests catch data quality issues automatically
Scalable: Add new metrics without rebuilding from scratch
Auditable: Git history shows exactly when logic changed and why
Fast: Sub-second query times on 14 months of data

Key SQL Techniques

Time-Series Construction

Challenge: Stripe only logs events (seat changes). Analytics needs "What was MRR on March 15th?"

Solution: Cross-join subscriptions × months, use LAG() window function to carry forward latest quantity until next event.

MRR Bridge Categorization

Challenge: "MRR went from $4,000 to $4,500. Why?"

Solution: Compare MRR to previous month using CASE WHEN logic: null = new, increase = expansion, decrease but >0 = contraction, 0 = churn.

What DuffDash Is Building Next

With MRR analytics working, the team is expanding the data platform:

Product Usage Analytics

Combine Stripe data with product event logs. See which features drive retention and expansion.

Churn Prediction

Build ML model predicting which customers will churn in next 30 days based on usage patterns.

Customer Health Scores

Automated scoring (0-100) combining MRR, seat growth, product engagement, support tickets.

Lessons Learned: Why This Project Succeeded

Start Simple

We could have built 20 dashboards. Instead: focused on the CEO's question ("What's our MRR?"). Nailed that first, then expanded. MVP mentality wins.

Trust the Data

Early on, dashboard MRR didn't match Stripe's number ($4,802 vs $4,810). Found a bug in event date filtering. Fixed it, added tests. Now: 100% match, 100% trust.

Document Everything

Used dbt's built-in docs. When CFO asked "How is 'expansion MRR' calculated?", I sent a link. Self-service documentation reduces repeated questions.

Version Control Is Non-Negotiable

CEO asked to change MRR calculation method. Made change, didn't like it, rolled back via git in 2 minutes. No Excel "SavedVersion_FINAL_v3_REALFINAL.xlsx" nonsense.

Is your SaaS data stuck in Excel hell?

If you're manually exporting CSVs, fighting with VLOOKUP, or answering "What's our MRR?" with "I'll get back to you in 3 hours," let's talk.

I build automated subscription analytics systems for SaaS companies. Stripe, ChargeBee, Recurly — doesn't matter. dbt + SQL + clean dashboards = answers in seconds, not days.

Book a Free 20-Minute Consultation Send Me Your Data Challenge

Typical turnaround: 2-3 weeks from raw data to live dashboard. No multi-month engagements. No vendor lock-in. You own the code.

Related Resources

Guide

How to Build Reliable ETL Pipelines

The principles that make data pipelines production-ready: idempotency, testing, monitoring, and more.

Read Guide →