Using Metabase to review coffee shop customer feedback

Customer feedback is useful when someone can find it, compare it and decide what to do next. In a small business, it often ends up in a spreadsheet or form that nobody has time to review.

We used Metabase to give a coffee shop owner a clearer view of the feedback they were collecting. This post explains the setup and the questions it was intended to help answer.

What Metabase does

Metabase connects to data sources and lets people explore them through queries, charts and dashboards. Its visual query builder can help colleagues work with data without writing SQL, while shared dashboards and reports give the team a common view.

You still need a suitable data source and someone to decide what each measure means. A chart is easier to read than a table of responses, but it can be misleading if the underlying question or calculation is unclear.

The coffee shop example

The owner wanted to understand changing customer preferences without losing what made the shop different from larger chains.

They had questions about popular menu items, service speed and the experience of visiting the shop. They did not have a repeatable way to collect and review that information, and they did not want customers to install an app.

Collecting the feedback

We built a short, optional feedback form that sent responses to a database. QR codes on tables and takeaway cups gave customers a way to open it after their visit.

Different QR codes prefilled the order type, so the owner could compare sit-in and takeaway responses without adding another question.

An early prototype of the feedback form and dashboard.

We later explored Formbricks as a way to make questions easier to change and adapt them to previous answers.

Building the dashboard

Metabase read the responses from the database and displayed them as charts. We organised the dashboard around questions the owner could act on:

• Which drinks and food items were attracting interest?
• Did comments about service speed vary through the day?
• What did people say about the seating and atmosphere?
• Why did regular customers choose to return?

The aim was to make those patterns visible during routine business discussions. Feedback could then inform decisions about stock, staffing, the space and local marketing.

This example does not establish how much any one change improved sales or satisfaction. Those outcomes need to be measured separately, with enough responses and context to support the comparison.

Connecting it to your own data

Start with a question before choosing a chart. "Are takeaway customers asking for different food options?" gives you something specific to investigate. "Build a business dashboard" does not tell you what information to collect.

A practical setup involves:

1. Identifying the records or responses that could answer the question.
2. Connecting Metabase to the database that holds them.
3. Building a query and checking a few results against the source records.
4. Choosing a chart or table that makes the comparison readable.
5. Sharing it with the people who will use it.

A dashboard brings related measures into one view.

Keep the first version small. A few measures that the team understands are more useful than a screen full of charts nobody trusts.

Make the reports part of the work

Agree who will check the dashboard and what would prompt a closer look. For example, repeated comments about slow service at a particular time might justify comparing staffing and demand during that period.

Look at the number of responses as well as the score. Check whether a change in the questions, collection method or customer mix could explain a difference. The dashboard should make it easier to ask these questions.

Shared reports also give colleagues a way to challenge an interpretation. Someone working the shift may know why an unusual result appeared. Keep that context alongside the numbers when deciding what to change.

Is it a suitable fit?

Metabase is worth considering if you already collect useful information and need a more repeatable way to inspect it. A visual query builder and shared reports can reduce the work of preparing the same spreadsheet summaries each week.

It will still take work to connect the data, check the calculations and maintain the installation. If you self-host, budget for updates, backups and support. Open-source software does not remove those responsibilities.

If your team is not yet collecting the information it needs, start there. A reporting tool cannot recover feedback that was never recorded.

Email us or book a call if you want help turning a particular business question into a form, data connection or report.