Why build a custom dashboard?
Most growing businesses already have numbers in five places: the marketplace seller panel, the store admin, Ads Manager, a sales sheet and the bank. Each shows its own slice, with its own dates and definitions. Someone exports and pastes every week, and the report is out of date by the time it is shared. A custom dashboard does the exporting and joining for you and answers the questions that need data from more than one place, such as ad spend as a share of total sales.
What goes into a dashboard?
- Questions first. We write down the five to ten questions you want answered, such as "which products lost money after ads last month" or "which state grows fastest".
- Sources. For each question, where the data lives and how to get it: an API, a scheduled report, an export emailed to an inbox, or a sheet.
- A clean store. Data is copied into one place, often Google Sheets for small volumes or BigQuery or Postgres for larger ones, with dates, currencies and product IDs made consistent.
- Definitions. Each metric is written down once. "Revenue" means the same thing in every chart.
- The view. A Looker Studio report, a sheet with charts, or a small web dashboard when you need custom logic or logins for your team.
- Refresh and alerts. Data refreshes on a schedule, and a failed refresh sends an alert instead of quietly showing old numbers.
Which tool should the dashboard be built in?
| Tool | Good for | Watch out for |
|---|---|---|
| Google Sheets with charts | Small data, teams who live in sheets, quick changes | Slows down with large row counts; easy to break by editing |
| Looker Studio | Free shareable reports on top of Sheets or BigQuery | Connects to Meta Ads and Amazon only through connectors or a data store in between |
| A custom web dashboard | Your own logic, client logins, mixing data with actions | More to build and maintain |
Common traps in dashboard projects
- Too many charts. A dashboard with forty charts gets opened once. Start with one page that answers the weekly questions.
- Mismatched numbers. Amazon, Meta and your bank count sales on different dates and with different returns. We document the differences instead of hiding them, so nobody loses trust in the dashboard when two numbers disagree.
- Silent staleness. If a token expires, the dashboard keeps showing last week. Every source has a "last refreshed" time and an alert.
- Manual steps left in. If one report still has to be downloaded by hand, we automate that too, or the dashboard will be stale within a month.
Examples you can try
The Amazon seller dashboard turns order, business and ads reports into views by date, product and state. The Meta Ads report writes daily rows and a summary. The influencer campaign tracker ranks creators by results. All use fictional data.
Questions people ask about custom dashboards
Can you build a dashboard from Amazon Seller Central data?
Yes. Order, business and advertising data can be pulled through Amazon’s seller and ads APIs where your account has access, or from the reports you download today. The data is joined by product and date into one view.
How often does the data refresh?
As often as the sources allow and the questions need. Daily is enough for most sales and ads dashboards. Some sources update with a delay, and the dashboard shows when each one last refreshed.
Can my team or clients log in to see it?
Yes. Looker Studio reports can be shared with chosen people. For client logins with their own data only, a small custom web dashboard is the better fit.
Do we need a data warehouse?
Not at the start. Small volumes work well in Google Sheets. When data grows, or several sources must be joined often, moving to BigQuery or Postgres keeps the dashboard fast. We pick the simplest store that will hold your data for the next year.