News

How to Create a Dashboard: Excel, BI or Custom

· DigifyNow

Auf Deutsch lesen

You build a dashboard in four steps. Define each KPI with a formula and an owner. Connect the data sources by export or API. Calculate the KPIs from one clean data table. Then show them on a single overview page with comparison values and filters. Excel is enough for exports and a few readers. A BI tool like Power BI suits several sources and many viewers. A custom web app only pays off if people also work inside the dashboard.

The short answer: to create a dashboard, first decide which KPIs each person sees and how each one is calculated. Then pull the data into a single data table, by export or through an API, and build one overview page from it with comparison values and filters. Choose the tool last, whether that's Excel, a BI tool or a custom web app, and base the choice on your data sources and on who will use the dashboard.

What is a dashboard, and what else is it called?

The word comes from the instrument panel in a car. In software, a dashboard is an overview page that shows a team's or department's most important numbers in one place, such as open quotes, revenue this month or overdue invoices. You'll also hear it called a KPI dashboard, a scorecard or a cockpit. A dashboard isn't the same as a report. A report answers one question in depth and often runs to several pages. A dashboard shows at a glance whether something is going off track, and lets you drill into the detail when you need it. One point matters here: a dashboard doesn't create numbers. It reads them from a source, which might be a spreadsheet, your accounting system, your CRM or a database. If that source is incomplete, the dashboard shows the same error, just in a nicer layout.

Illustration for the question "What is a dashboard, and what else is it called?"
AI-generated image

What do you need to settle before you build a dashboard?

Most dashboards don't fail because of the tool. They fail because nobody ever agreed on what a number means. Get these four points down on paper before you pick any tool:

  • Users and decisions: Who looks at it, and what does that person do differently when a number turns red? If a KPI doesn't lead to any action, it doesn't belong on the dashboard.
  • A definition for every KPI: Is revenue net or gross? Is it counted by invoice date or by payment date? Does it include credit notes? Write down the formula and the source for each one.
  • Data source and access: Where does each number come from? Is there an API, or only a manual export as a CSV or Excel file?
  • Refresh rate: Daily, hourly, or once a month at month-end close? This decides whether a manual export is enough or whether you need an automated connection.

Taken together, these decisions are a small requirements document. If someone builds the dashboard for you later, that document is what the quote and the sign-off are based on. Our guide on how to write a requirements document, with steps and a template, shows how to put one together.

Example: which KPIs belong on a dashboard for a trades business?

Here's an example of the kind of conversation we have in projects. A trades business with an office, a crew of field technicians and a managing director wants to see where money is getting stuck. It doesn't need twenty charts. Four KPIs are enough, each with a definition, a source and a person who acts on it. The table is a template you can adapt. It isn't an analysis of a real client. The Action column is the test: if you can't fill it in, the KPI goes.

Sample dashboard for a trades business: each KPI with its definition, source, refresh rate and action.
KPIDefinitionSourceRefreshAction
Open quotes with no responseQuotes with status open that are older than the agreed follow-up periodQuoting module of the field service software, or the CRMdailyOffice calls the customer
Completed jobs not yet invoicedJobs with status completed that have no matching invoice numberJob management systemdailyOffice sends the invoice
Overdue receivablesInvoices past their payment due date, net totalAccounting or invoicing softwareweeklyManaging director decides on sending a payment reminder
Capacity for the coming weeksScheduled hours per calendar week divided by the technicians' available hoursScheduling board or calendardailyManaging director accepts jobs or reschedules them

What the example depends on in practice: the second KPI only works if jobs and invoices are linked by a shared number. Without that key, no tool can calculate the figure automatically. This is usually the first point where a dashboard project turns into a process project.

Excel, Power BI, Looker Studio, Tableau or a custom web app?

Search results are dominated by Excel and Microsoft Power BI, with Google Looker Studio and Tableau close behind. The table measures every option against the same criteria: how data gets in, how you share it, when the option fits and where it stops working. Licensing models change often, so before you decide, check with each vendor what creators need and what viewers need.

As of November 2026. Check vendors' features and licensing in their documentation before you decide.
OptionHow data gets inSharing and permissionsGood fit whenHits its limits when
ExcelFile imports, queries via Power Query, PivotTables built on one data tableSend the file around or open it from shared storage, permissions at file levelfew readers, data arrives as exports, weekly or monthly refreshseveral people edit at once, copies multiply, exports have to be run by hand every day
Microsoft Power BIModeling in the desktop app, publishing to the online service, connectors for many sourcesSharing through workspaces in the online service, clarify viewers' licensing needs up frontthe company already runs on Microsoft's Office and cloud services, several sources need to come togethernobody in-house maintains the data model, sources without an API still have to be loaded manually
Google Looker StudioBrowser-based, connectors to Google services such as Analytics, Ads and SheetsSharing works much like Google Docsmarketing and web data already live in Googledata from inventory or accounting systems needs an extra step, for example via Google Sheets or a database
TableauDesktop and cloud versions, connects to files and databasesPublishing to a server or the cloud with rolesextensive visual analysis, analysts on staffsmall user base, no time to learn it, a simple red-yellow-green overview is all you need
Custom web appDirectly from your own database or through the APIs of your business systemsRoles and visibility fully configurable, down to individual recordspeople should also work in the dashboard: approve, assign, set statusesit's only about viewing numbers that a BI tool can already read

One detail about Power BI that often causes confusion: according to Microsoft's documentation on dashboards in the Power BI service (checked November 2026), dashboards are a feature of the online service. In Power BI Desktop you build reports, and once you publish them, you pin their visuals to a dashboard. So if you only work in the desktop app, you have reports, but not yet a shared dashboard.

An honest rule of thumb: if your data already sits in a system with an API and you only want to look at numbers, a BI tool is usually the shorter path. Custom software only becomes interesting when the dashboard is part of a workflow. Our article Custom software vs. SaaS: when each one pays off explains why.

Illustration for the question "Excel, Power BI, Looker Studio, Tableau or a custom web app?"
AI-generated image

How do I build a dashboard in Excel?

Excel is often enough to get started. Work in this order and you won't have to rebuild everything later:

  1. Put the raw data on its own sheet: paste the export in unchanged, or load it with Power Query so the next export goes through exactly the same steps.
  2. Format the data as a table (Ctrl+T): one row per transaction, one column per attribute, no merged cells, no subtotals. PivotTables then grow automatically as new rows come in.
  3. Calculate the KPIs on a separate calculation sheet: use PivotTables or formulas that refer to the table and its column names, not to fixed cell ranges that will be too short after the next export.
  4. Create the dashboard sheet: charts built from the PivotTables (PivotCharts), the most important number in the top left, with a comparison value next to it, such as last year or target.
  5. Add filters: connect slicers and a timeline to all the relevant PivotTables so one click filters every chart. According to Microsoft Support on filtering with slicers (checked November 2026), you do this through the slicer's Report Connections, and the PivotTables must be based on the same data source.
  6. Protect and document: hide or lock the calculation sheets, put the KPI definitions on an info sheet, and show the date of the last refresh.

Where Excel stops working: as soon as someone copies the file and keeps updating it locally, you have two versions of the truth. And if the daily export takes half an hour of manual work, that's a process problem, not a presentation problem.

Why dashboards fail in practice

  • The refresh breaks: the source renames a column or changes its export format, and the query can no longer find the field. Fix: reference columns by name instead of position, and show the time of the last refresh where everyone can see it.
  • The numbers don't match accounting: the definitions differ, for example invoice date versus payment date. Fix: a written definition for every KPI and a check against a known monthly total.
  • Too many charts: if you show everything, you show nothing. Every tile needs a person who acts on it.
  • Expiring credentials: automated connections often run on the tokens or passwords of individual employees. When that person leaves the company, the dashboard stops. Fix: use service accounts instead of personal logins.
  • Permissions that are too broad: payroll costs or margins end up in front of people who shouldn't see them. You have to assess for yourself which data flows into which tool and to which vendor, especially when personal data is involved.
  • A data source with no API: then a manual export is the only option, and the dashboard is never more current than the last export.

Our article What is an API: definition and example explains how systems hand over data through endpoints, tokens and pagination.

Illustration for the section "Why dashboards fail in practice"
AI-generated image

How much does it cost to have a dashboard built?

You can only get a serious number once the scope is clear. Four things drive the effort. First, the number and type of data sources: a clean API connects quickly, while a legacy system that only exports files needs an extra step in between. Second, the state of the data. Duplicate customers, inconsistent item numbers or free-text fields have to be cleaned up first or converted during loading. Third, permissions and roles: a dashboard just for the managing director is simpler than one where every sales rep sees only their own customers. Fourth, ongoing operation, meaning viewer licenses for a BI tool, hosting for a custom application, and someone who responds when a connection breaks. A number becomes binding through an effort estimate based on the KPI definitions and a list of sources. Our article on software development cost and what drives it shows how such estimates are put together.

How DigifyNow builds dashboards

DigifyNow doesn't build dashboards as a standalone product. We build them as part of custom web apps and back-office systems, for example as the home screen of a CRM or a job management system. Here's how it works: first we go through the KPIs and the sources. Then we check whether an existing BI tool already answers the question. If it does, we tell you, because then custom development isn't worth it. If people should also work in the dashboard, we build it as a web app with its own database and connections to your business systems. We typically start with an MVP in 4 to 8 weeks, which then keeps growing once it's in use. The DigifyNow services page gives an overview of our app, web and AI projects.

One example of this kind of project, where data and workflows live in one dedicated system instead of scattered spreadsheets, is the custom back office with CRM for Mundo Marketing. The overview there isn't a tool bolted on afterwards. It reads from the same database people work in, so there's no export step in between.

Illustration for the section "How DigifyNow builds dashboards"
AI-generated image

Tablify: analytics for restaurants, a product by DigifyNow

Tablify is a product by DigifyNow for restaurants and hospitality: digital menus, vouchers and analytics via QR code. The analytics are part of the product. A restaurant owner who wants to know how their menu and vouchers are being used doesn't have to combine exports. They see the numbers right in the product. For KPIs outside of that, for example from the POS system or accounting, the steps above apply.

Related questions

How to create a process map

A process landscape map sorts a company's workflows into management, core and support processes and shows how they connect. It's a good starting point for a dashboard, because every KPI should belong to a process, for example lead time to the order process. You'll find the structure and an example in our article process landscape map: layout, example and steps.

Getting an online store built in the Westerwald region

For an online store, you first decide on the route: a website builder, a hosted store platform, open source or custom development. That choice decides who takes care of servers and updates. Most store platforms provide store KPIs such as orders or return rate through an API, so you can pull them into a dashboard. DigifyNow is based in Unnau in the Westerwald region of Germany, works across Germany and describes its offer on the page online store development. The routes are compared in How to create an online store: options, cost drivers, pitfalls.

Having software built

If you're having software built, you start by describing the workflow and the acceptance criteria. After that come the estimate, development in iterations, acceptance and operation. A dashboard is often just one part of this. The real work lies in the data model, the interfaces and the permissions. The article Custom software development: how a project works describes the process.

When outside help is worth it

Outside help is worth it when several systems without shared keys need to come together, or when exports are run by hand every day. The same goes when permissions have to be broken down to individual customers or employees, or when people should also process transactions in the dashboard. For a simple monthly overview from one export, you usually don't need anyone. For the conversation, prepare a list of KPIs with their definitions, the source systems with a note on whether they have an API, the number of users and a sample export. In a free assessment of your project, we'll look at it and tell you which route fits, even if that's an off-the-shelf tool.

Illustration for the section "When outside help is worth it"
AI-generated image

Frequently asked questions

How do I create a dashboard?

In four steps. First, define your KPIs with a formula, a source and a person responsible. Second, load the data by export or API into one data table with one row per transaction. Third, calculate the KPIs from it, in Excel for example with PivotTables. Fourth, build an overview page with the most important number in the top left, a comparison value next to it such as target or last year, plus filters and the date of the last refresh.

Where do I find the dashboard?

That depends on the software. In most web applications, the dashboard is the home screen after you log in, or a menu item with that name. In BI tools like Power BI, dashboards live in a workspace in the online service. In Excel it's usually a separate sheet in the file. If you can't find it, often your account just hasn't been given access.

Excel or Power BI: which is better for a dashboard?

Excel is a file and fits when a few people read an analysis built from exports. Power BI separates the data model from publishing and fits when several sources need to come together and many viewers should see the same version. So the difference lies in sharing and refreshing, not in how it looks. With Power BI, viewers may also need licenses, which you should check beforehand.

How many KPIs should a dashboard have?

As many as a decision needs. Every KPI needs an owner, a written definition and a comparison value, such as target or last year. A number nobody acts on is decoration. Several focused dashboards per role, such as sales, accounting and management, work better than one page for everyone.

What's the difference between a dashboard and a report?

A report answers one question in depth, often across several pages with tables and filters. A dashboard is a single overview that shows whether action is needed and links to the details. According to Microsoft's documentation (checked November 2026), the tiles of a Power BI dashboard are created from pinned visuals of existing reports. So the dashboard is the summary, and the report is the analysis.

Request a free assessment of your project