Nine rules for building a subscription commerce reporting dashboard

What to take away

This guide sets out nine rules for a subscription commerce reporting dashboard, in the order they should be built, from the definitions everyone agrees to the quarterly check that keeps the numbers honest.

Most teams build the chart before agreeing the definition, so two people quote two churn figures in the same meeting.
Write the metric dictionary first, then wire the dashboard to it. Every tile traces back to one rule.
Keep personal data out of the reporting layer. Pseudonymised customer IDs are usually enough.
Tag sign-up conversions properly, or paid channel reporting breaks when subscription events are mis-mapped.
Review the dashboard quarterly against your wider measurement and reporting guide so it does not drift.

Rule 1: Agree the definitions before you build anything

Two people can quote two different churn figures in the same meeting because nobody wrote down what churn means. Before anyone opens a chart tool, bring finance, marketing and support together and agree the wording for the handful of terms the dashboard will carry.

Record that decision with a date and a named owner, so the next debate starts from the same page rather than from memory.

Rule 2: Write the metric dictionary first

A dashboard is a set of promises about what words mean. If active subscribers means a live billing mandate to one person and a payment this month to another, board packs become arguments.

Metric dictionary first page

Name each metric
Write the calculation
Name the owner
Set the cadence
Ten to fifteen metrics only
Fix reporting grain for a year

Rule 3: Choose the reporting grain

Decide whether you report daily, weekly or monthly, and whether cohorts run monthly or weekly. Subscription revenue behaves differently from one-off retail, so weekly cohorts show early churn sooner.

Keep the grain stable for at least a year. Changing it mid-year makes historic charts unusable.

Rule 4: Get the data plumbing right

Start where subscription truth already lives: Stripe Billing, Chargebee, Recurly or Zuora. Land that data in a warehouse you control (BigQuery, Snowflake or Redshift) and model it there, rather than pointing the dashboard straight at a billing API.

Move the data with an ELT tool such as Fivetran, Airbyte or dbt, and consume webhooks for the events that matter between batches, like a plan change, a cancellation or a failed invoice. Every tile should read from one modelled table with a known refresh time.

Rule 5: Separate personal data from reporting data

The ICO's explanation of what counts as personal information is the right starting point. If a row can identify a living person, treat it as personal data even in a warehouse table.

Most dashboards do not need names, emails or addresses. Pseudonymised IDs, plan codes and dates cover retention, churn and revenue reporting.

Rule 6: Map sign-up events to the right conversion

Paid acquisition reporting depends on a clean conversion definition. Google's guidance on Quality Score for Search campaigns explains how landing page relevance shapes what you pay, and your sign-up event feeds that loop.

Count trial starts and paid starts separately. Blending them inflates early performance and hides the drop-off that follows.

Rule 7: Build the tiles people actually use

Choose the tool that matches the job. ChartMogul, Baremetrics and ProfitWell arrive with subscription metrics already defined and connect straight to billing systems, which suits a small team that needs answers this month. Looker Studio, Power BI, Tableau and Metabase give you a general-purpose layer over the warehouse once the questions get more specific.

A tile earns its place only if someone opens it before a decision. Build the ones people actually use and retire the rest.

Keep the top row for money and retention

Put recurring revenue, net revenue retention and logo churn along the top. They answer most trading meeting questions.

Below them, add acquisition cost per subscriber, average revenue per subscriber and failed payment recovery. Each tile needs a definition, a source table and a named owner.

Rule 8: Add a glossary to the dashboard itself

A short glossary stops the same argument recurring. Keep it visible rather than buried in a wiki.

Glossary

Active subscriber:: a customer with a live billing mandate who has paid at least once.
Net revenue retention:: revenue from existing subscribers this period, including upgrades and downgrades, divided by the same group last period.
Logo churn:: subscribers lost in a period divided by subscribers at the start.
Failed payment recovery:: payments collected after an initial failure, divided by all failed attempts.
Reporting grain:: the time unit each figure is calculated over.

If your team is still settling which numbers matter most, key metrics for UK operators is a sensible next read before you lock the schema.

Rule 9: Keep it defensible

Defensibility is what separates a dashboard from a folder of screenshots. Every figure needs a source, an owner and a date, and the checks below should be routine rather than heroic.

Check consent and marketing claims

Some reporting feeds marketing decisions that fall under consumer law. The CMA's consumer protection guidance for businesses covers unfair commercial practices, which matters when dashboard segments drive renewal messaging.

Keep consent flags in the source system, not the dashboard. Report counts of consented subscribers rather than exporting contactable lists.

Schedule a quarterly rebuild check

Each quarter, pick three tiles and rebuild them by hand from source data. If the manual figure differs by more than a rounding tolerance, fix the pipeline first.

Record who checked, when and what changed. That log turns a dashboard into evidence a board can rely on.

Common questions

How many metrics should a first dashboard have?

Ten to fifteen. More than that and nobody reads the tiles. Add metrics when a decision needs them.

Should churn be reported monthly or weekly?

Whichever matches how you acquire subscribers, then held constant. Weekly cohorts surface early churn faster.

Can we keep customer emails in the reporting layer?

Usually unnecessary and adds risk. Use pseudonymised IDs and join back to personal data only for a documented reason.

Who owns the dashboard?

One named person, normally in finance or commercial operations, not a committee. They own the dictionary, the refresh and the quarterly check.