Customer dashboard development

What customers often want to see

Showing each customer only their data

In a customer dashboard, every query must be limited to the signed-in customer's data, and that limit must be enforced on the server, never left to the screen hiding other customers' figures. Business customers add another layer: several users at one company, sometimes with different permissions, such as a manager seeing all sites and a site user seeing one. Automated tests that try to read another customer's data, and fail, should run with every change.

A single isolation mistake can expose one customer's information to another, so this deserves more attention than chart styling.

Designing a useful customer dashboard

Keeping it fast as data grows

• Pre-calculated summaries

  Daily or hourly totals prepared in advance.

• Sensible default periods

  Recent months first; longer history on request.

• Caching

  Recently viewed figures reused for a short time.

• Separate reporting store

  Dashboard queries kept off the live transaction system.

Building a customer dashboard

1. 01

   Ask customers

   Which questions they ask support about their account.

   Output: Priority measures

2. 02

   Prepare data

   Per-customer summaries built and checked.

   Output: Reliable figures

3. 03

   Design

   Prototype tested with a few customers.

   Output: Useful layout

4. 04

   Build and isolate

   Dashboard built with access tests.

   Output: Safe dashboard

5. 05

   Release gradually

   Offered to a group of customers first.

   Output: Feedback

Figures customers will quote back

Customers may use dashboard figures in their own reports or disputes, so they must match invoices and other records. Define each figure carefully, show when data was last updated and explain any delay. Where figures are estimates, say so. A difference between a dashboard and an invoice damages trust faster than having no dashboard at all.

Alerts as well as dashboards

Customers rarely check a dashboard every day. Notifying them when something notable happens, such as usage approaching a plan limit, a spend threshold reached or a delivery running late, brings them to the dashboard when it matters. Let customers choose which alerts they receive and how, and keep the number modest so alerts are not ignored. Each alert should link straight to the relevant figure.

Review which alerts customers switch off; it shows which ones are not useful.

Building charts or embedding a reporting tool

Customer dashboards can be built directly into your product with a charting library, or by embedding a reporting tool that supports per-customer filtering. Building gives full control over design and performance; embedding can be quicker for many charts but brings licence costs that may grow with the number of customers who view it, and needs careful setup of data isolation. The services and cost pages for customer dashboards compare these routes in more detail.

Measuring whether it helps

Track how many customers use the dashboard and whether related support questions fall. Ask a few customers which figures they rely on and which they ignore, and adjust.

Personal data in customer dashboards

If dashboards show individuals' activity, for example usage by named users at a business customer, check that showing it is appropriate and explained in your privacy information.