Why more data on screen can mean less understanding in the room

Picture a presenter clicking to a slide covered edge to edge with a bar chart: twelve colors, a wood-grain 3-D background, a company logo in the corner, gridlines running through every bar. Somewhere in that slide is a genuinely useful piece of information. Nobody in the room can find it. This is the scenario Garr Reynolds returns to again and again, and he borrows a term from radio engineering to explain exactly what's going wrong: the signal-to-noise ratio.

On an AM radio, the signal is the broadcast you actually want to hear, and the noise is the static that creeps in the farther you drive from the station. Turn that idea toward a chart or graph, and the signal is the concrete point you're trying to make, the clearest form of the data itself, while the noise is anything, a texture, a stray color, an unnecessary decoration, that gets between your audience and that point. Reynolds's goal for any quantitative display is simple: push the signal-to-noise ratio as high as it can go, using three moves he calls restrain, reduce, and emphasize. Restrain yourself from adding more than the audience needs. Reduce whatever is already there but isn't earning its place. Emphasize the one thing that actually matters, through contrast or a title written as a plain sentence rather than a vague label.

Reynolds points to the public health researcher Hans Rosling as someone who mastered this at scale, not by simplifying his data but by finally giving it a way to be heard. Rosling had spent years working with detailed statistics, respected in his academic field, before he and his collaborators built the Gapminder software, which turned dense, boring numbers into animated, interactive graphics that could be understood by a general audience. In an interview, Rosling compared ordinary data presentation to handing someone sheet music and expecting them to appreciate a song just by reading the notes. Most people, he said, need to actually hear the music to understand how beautiful it is; presenters usually just show the notes, and never play them. The chart itself hadn't changed. What changed was whether anyone could actually hear what it was saying.

Next time you build a chart, decide up front whether your audience needs exact numbers or just needs to see a trend, because that single choice tells you whether to reach for a table or a simple graph. Then go through your slide like an editor and strip out anything that isn't the data itself: the drop shadows, the extra colors, the logo in the corner, any 3-D effect that distorts more than it clarifies. Swap a vague title like a date range for a full declarative sentence that states the finding outright, so the point lands before anyone has finished reading it. Use color or contrast on exactly one part of the chart, the part you most want noticed, and let the rest sit quietly in the background. Do this consistently, and your numbers will finally get to speak instead of just sitting on the screen looking busy.