Data overload is not about human limitations; it’s about design failure
“Confusion and clutter are failures of design, not attributes of information”
– Edward Tufte
People throw around the term data overload all the time to describe a problem in a product. On the surface, this seems pretty direct. Users have trouble sorting through the data presented to them and routinely make bad decisions. Designers look at the UI, see how much data users are faced with and immediately declare that there was too much data for users to make the correct choice.
In some cases, there is even a phenomenon from Psychology that people cite to back this up: Hick’s Law. The simple version of Hick’s Law is that the more choices presented to a user, the longer it will take for the user to make a decision.
It’s a standard pattern and leads to pretty consistent design responses: Provide a control that lets the user filter the data (or filter for them it so the data is unavailable altogether), paginate, or implement a search. Perhaps the system designers introduce an analytic to make sure the user sees the important data first (like on your Twitter or Facebook feed). The designer tests the system and gets positive reviews and improved decision-making. The user sees a simpler UI with a manageable amount of data and declares that the system is much better.
This victory, while nice, is a bit hollow. The improvements are generally small and don’t address the root cause of what the user is experiencing. The problem is not that there was too much data (or too many choices), but as the quote from Tufte states, a problem with the design. Finding ways to remove data is almost definitely the worst of good alternatives.
Human Capabilities are Underrated
Despite what most people think, humans are amazing at information management. We have a tremendous capability to take in large quantities of data from a variety of sources and make good decisions about where to focus our attention and what actions to take. If not, we wouldn’t be able to function in our day-to-day lives. Just think about what life is like walking down the sidewalk on a busy street. You have store fronts with signs trying to catch your attention. Traffic lights are signaling cars and flashing the walk / don’t walk signs for pedestrians. Other people are walking all around you. You hear myriad noises from all directions. In short, your senses are being bombarded.
Despite the high sensory input, in these situations you rarely fail to get to your destination. Why? Because humans have evolved mechanisms to process information, filter out the irrelevant, and focus on what’s important. You ignore both the faces in the crowd you don’t recognize and the traffic signal. You don’t care about the walk sign until you get to the intersection. You don’t really notice the gentle slope in the sidewalk, even though your body has adjusted to start walking uphill. You focus on the task at hand — getting to that new restaurant you’ve been meaning to check out.
Good design utilizes these built-in mechanisms to help the user accomplish their goals successfully. They use the human’s innate capability to filter irrelevant information rather than try to guess and filter it for them.
What ‘Data Overload’ Really Is
When people talk about Data Overload, usually one of three problems is present. The first problem is that the system presents irrelevant data to the user. At least some, and possibly all, of the data fail to little decision-making value. This is a result of a product designer who has little idea what data is relevant, so they collect whatever data seems to make sense — or more likely, whatever data related to the problem that is easy to collect — and provides them in a table or grid. This seems like data overload because there is so much noise to wade through that the signal is impossible to find.
The second problem occurs when the product designers don’t know how to present the data to the user. Data visualizations are extremely important. There is a large body of research about the human perceptual system and how this can be leveraged in helping users understand their data. Not all charts can be used for all data sets. People are adept at identifying patterns in the world and data visualizations can be used to support this pattern recognition. But when the wrong visualizations are used (or no visualizations at all), the data can seem to be overwhelming.
The final problem when people talk about data overload is that sometimes there isn’t enough data. This seems counter-intuitive, but data is best understood in context. The simple demonstration above show how this is possible. The ‘Just Data’ and ‘Compare Data’ steps have the same number of data points. As we add in historical data in the ‘Trends’ step, our picture of the story (and decision-making) starts to become clearer. The sharp downward trend of Option 1 (the blue line) may make it a less idea option than options 2 or 3. Adding one more data point — the system goal demarcation (the dotted line) — in ‘Trends in context of goal’ helps clarify things further. Option 2 starts looking like perhaps the best option as it sits above the goal and has steady performance.
When you use data visualizations, as long as each additional element is placed with a purpose, more data can be more informative without increasing the cognitive load. As a designer you may be asking the system to gather and store more data, but this makes the user’s decision-making easier.
These problems are even more evident when they are found in combination. It’s really easy to use the wrong presentation approach when you don’t provide enough data. And when you don’t know what the right data to show your users, it’s really easy to give them the wrong data instead.
Helping Users Manage Their Data
How do you get things right? Rather than start with the data you have and figure out how to show it to the user, start with the decisions the user is making and figure out what data they need to make that decision and what context it should be in. This gets tricky, since sometimes the ideal data is not available or obtainable. Then you need to compromise, while maintaining a record of what you would like to get in case it becomes available. You should probably even inform your user that they are getting an approximation of what they really need, to help them understand the gaps in their knowledge.
Once the information is identified, the key is to then figure out the presentation device that best reveals to the user the relevant data patterns. Don’t pick a chart for your data that you think looks good, pick the right chart for the decision and make it as evocative as possible without reducing the decision-making capability of the user. User decision-making comes first; styling comes second.
The term data overload assumes humans are incapable and that designers have no responsibility.
I hate hearing the term data overload, because it assumes humans are incapable — a source of weakness and error — and that designers have no responsibility. Hick’s law is useful, but just talking about the raw count of data available is not. People assume the two are equivalent, but they are not. You can have thousands of data points on a screen and understand it with ease. The number of data points isn’t the issue — the presentation of that information is.
We designers have the duty to understand how humans think and to understand the decision-making needs of the user in the domain we are working in. We need to understand how human perceptual mechanisms work and how that impacts how we encode the data that users need. We need to give them the right data in the right way. We need to take advantage of the power of our users and design in a way that truly supports them.
Thanks for reading. For more of my UX ramblings, you can follow me on Twitter: @bkenna1 or here on Medium.
I posted an earlier version of this a few years ago but have updated it to include some more details, images, and ideas.
