Should we use a BI product or build custom dashboards?
If your data is already clean and in one warehouse, a BI product is usually the better buy — it gives you ad hoc exploration you would not want to build. Build custom when the hard part is upstream: joining systems, resolving the same customer across platforms, and defining what a metric means. A BI tool draws charts beautifully; it does not decide what a lead is.
Most dashboard problems are not dashboard problems
When a company says it needs a dashboard, the underlying complaint is usually that numbers from different systems disagree, or that nobody trusts the one number that exists. Neither is fixed by a charting layer. Point a BI product at inconsistent source data and you get inconsistency rendered in higher resolution.
The expensive work sits below the visualization: pulling from each platform, reconciling definitions, matching the same customer across a phone system and a field service platform, and deciding which system is authoritative for each fact. Do that work and almost any front end will look good.
Where each option genuinely wins
- BI products win at ad hoc exploration, pivoting, scheduled distribution, and giving analysts freedom without engineering involvement.
- BI products win when many people need to ask their own questions rather than read the same answer.
- Custom wins when the audience will not log into a separate tool — owners, general managers, technicians. Per-seat licensing also gets awkward when the audience is everyone.
- Custom wins when the view needs to be embedded in something else, sent as an email, or shown on a wall screen in a shop.
- Custom wins when the thing is not really a dashboard but an application — where users take actions, not just look.
The question that settles it
Ask what happens after someone reads the number. If the answer is "they think about it," a BI tool is fine. If the answer is "they assign a callback, flag a job, or message a manager," you need software, because the value is in the action and a report cannot take one.
This is the practical difference between reporting and operational systems. One describes the business; the other participates in it.
Dashboards get looked at less than anyone expects
A dashboard requires a person to remember to visit, to know what normal looks like, and to notice a change. In practice, most get heavy traffic for two weeks and then almost none. That is not a design failure, it is human attention.
The durable pattern is push, not pull: a short written summary or an alert that arrives where the person already is, with the dashboard available for when they want to dig. We build both for that reason — see custom dashboards and daily briefs — because the brief drives behavior and the dashboard answers the follow-up question.
Topics: dashboards · business intelligence · reporting · data modeling
Have a version of this question about your own business?
The useful answer usually depends on which systems you run and how they're connected. That's a conversation, not a blog post.