Interactive reporting formats: focused information delivery
Research gets ignored at the reporting stage more often than at any other. What the data-visualisation literature says about dashboards that are actually read.
The commonest fate of a research project is not to be contradicted. It is to be filed. A 280-page appendix and a deck of cross-tabulations satisfy the contract and change nothing, because nobody with the authority to act had the time to find the four numbers that mattered.
A dashboard is a definition, not a technology
The useful definition is narrow: a visual display of the most important information needed to achieve one or more objectives, consolidated and arranged on a single screen so it can be monitored at a glance. Every clause is doing work. Most important means selection has happened. Single screen means scrolling is a failure. At a glance means the reader should not have to decode anything.
Most things called dashboards fail the definition on the first clause. They show everything that was measured, which is a table with a colour scheme.
Ink that does not carry information is noise
The governing principle is the data-ink ratio: maximise the proportion of the display that encodes data, and remove, de-emphasise or regularise everything that does not. Gridlines, gradients, drop shadows, three-dimensional bars and decorative frames all cost attention and return nothing.
Alongside it sits graphical integrity: axes that do not start where they should, inconsistent scales, and pie charts sliced too finely all mislead, whether or not anyone intended it. A chart that overstates a difference is a reporting error in the same way a mistyped number is.
Design rules that survive contact with a real audience
- Eliminate clutter and distraction. If removing an element loses no information, remove it.
- Group data into logical sections, so the eye can navigate without a legend.
- Highlight what is most important — deliberately, and only a few things. If everything is emphasised, nothing is.
- Support meaningful comparisons and discourage meaningless ones. Putting two measures on one axis invites a comparison; make sure it is one you want made.
Interactivity is for the second question
Interactive delivery earns its cost when different readers need different cuts of the same study — a country manager wanting one market, a category manager wanting one product. It stops earning it when interactivity replaces editorial judgement. A filterable dataset with no stated finding hands the analysis back to the reader, which is the job they commissioned out.
The test is simple. If the top of the screen does not say what the research found, no amount of drill-down will rescue it.
