What is a KPI dashboard?

A KPI dashboard needs three things to be useful: four to eight headline metrics, a comparison against a previous period, and a breakdown showing what drove the change. Most dashboards ship the first, skip the third, and leave the reader knowing a number moved without knowing why it moved.

1. The headline figures. Four to eight numbers, no more. Past roughly eight, nobody scans the whole screen, so the extras are not read and the important ones get diluted. If a number would not change a decision this month, it belongs in a detail view.

2. The comparison. Revenue of $48,000 means nothing on its own. Against $61,000 last month it means something specific and urgent. A figure without a prior period is decoration.

3. The breakdown. Knowing revenue fell is not actionable. Knowing it fell because one product line dropped 40% while everything else held steady is. This is the part that gets cut most often, usually because the underlying tool made it hard to build.

Most example galleries show polished mockups built from invented data. Those are useful for choosing a colour scheme and useless for working out what you can actually build. These four are organised by the columns a real export contains, because that is what decides whether a dashboard is possible at all.

There are four types. Strategic dashboards review performance monthly or quarterly against targets. Operational dashboards track what is happening today and need a live connection. Tactical dashboards sit in between, tracking a specific team, project or campaign. Analytical dashboards support investigating why something changed. Picking the wrong one is why dashboards go unread.

Strategic. Monthly or quarterly, for owners and boards. Few metrics, long time ranges, always compared against targets. This is the one a spreadsheet export serves best, because monthly data does not need a real-time feed.

Operational. Daily, for the person running things. Needs near real-time data, so it usually needs a live connection rather than a file. If you want to watch orders arrive today, this is not the tool.

Tactical. Mid-level, for a specific team, project or campaign rather than the whole business. Usually weekly or monthly, which a periodic export serves well.

Analytical. Ad hoc, for working out why something moved. Wider data, more slicing, less fixed layout. Our filters and breakdowns cover the light version of this; a warehouse and a BI platform cover the deep one.

A KPI dashboard gives everyone one agreed set of numbers, removes the manual work of assembling them each week, and shortens the gap between something changing and someone noticing. The benefit is not the chart. It is that a question which used to take an afternoon now takes a glance.

One agreed set of numbers. Most arguments about performance are really arguments about whose spreadsheet is correct. A single computed view ends that, provided the figures reconcile to source, which is why every number here traces back to your file.

Time back every week. The recurring cost of manual reporting is not the analysis, it is the reassembling: exporting, pasting, reformatting, rechecking. That is the part worth removing.

Noticing sooner. A monthly report tells you about a problem up to a month late. A dashboard you actually open shortens that, which is usually worth more than any single insight in it.

Knowing when not to trust it. Less commonly listed, and the one we would argue matters most: a dashboard that is confidently wrong is worse than none, so the Data Readiness score sits beside the numbers rather than behind them.

Every dashboard you have seen in a gallery was built from clean data. Real exports have blank cells, three date formats in one column, product names spelled four ways, and totals that do not reconcile.

That matters because a dashboard built on a file with 30% missing costs will still render a confident margin figure, and there is nothing on screen to tell you it is wrong. A dashboard that is quietly wrong is worse than no dashboard, because you act on it.

This is why every Dashlytics dashboard ships with a Data Readiness score: a rule-based audit of completeness and consistency, shown next to the numbers rather than buried. It tells you how far to trust the output before you act on it. No competitor we have found publishes anything equivalent.

Worth being direct, because the search results for templates are enormous and mostly a detour.

A template is a fixed layout waiting for data shaped exactly like the template author's. Real exports almost never match, so the work becomes reshaping your file to fit the template rather than answering your question. People often spend an afternoon on that and end up with a dashboard describing someone else's business.

We do not publish downloadable templates, deliberately. The alternative is to read the columns your file actually has and build around them. If your data supports margin analysis, you get margin analysis. If it has no cost column, you are told that instead of being shown an empty widget.

No data model, no measures, no query language. If you can export a spreadsheet, you have everything needed.

A dashboard answers "what is happening and what changed". It is the wrong shape for three other jobs.

A one-off question you will never ask again is a query, not a dashboard. Build the answer and throw it away.

A deep investigation into why something happened needs exploration, not a fixed layout. Dashboards are for noticing; analysis is for explaining.

A formal document for a client or a board wants narrative, structure and prose. That is an AI report generator job rather than a dashboard job, and the two are genuinely different outputs.