Analytics continues to be a hot topic for lean startups, but many startups seem to focus on collecting data instead of searching for information. I’ve written about this difference before here, but I’d like to zoom in on it.
A simple example: Let’s say that I’m trying to figure out who won a particular US election so I can send a congratulatory note to the winner. (I won’t say which election until the end).
Metrics that Make You Ponder(ous)
Here’s some data to answer that question, visually presented in an intriguing way:
That’s a map of a US election. It’s pure data. I have no idea what it means. Perhaps it’s what the US looks like while it’s falling into a black hole while simultaneously dealing with a zombie apocalypse. I’m not sure.
This map was created by:
Looking at the votes county by county, scaling by population, and using a red-blue color scale to represent the percentage of voters voting for a particular party
Ok…that’s great, but so what? If we wanted to convey that America is a mixed country, not strictly dividing into “red and blue states”, this would be pretty cool.
Here’s a slightly better election map:
Ok…slightly better. On this map of the same election there are no distortions and the maps are chunked by electorate so I can at least tell who won which state.
But who won?
In order to answer the most relevant question we’d need to have a good understanding of US geography and also know the electoral votes of each of the states. Not knowledge most US citizens are likely to possess.
The context matters for metrics. As a foreigner, I might be inclined to say the red party won…whoever that might be. To present data as information, understand who will be viewing your dashboard is relevant, or you may find that people quickly draw the wrong conclusions.
So we’re still in the land of data without being able to parse much relevant information and we still can’t answer the question of who won without a lot of extra work or a considerable amount of expertise in the US electoral system.
A Single Focused Answer
Let’s try again:
Now we have answer very simply presented at the top. Clearly Obama won the election, and that explains why he keeps giving press conferences on the White House lawn.Information is data framed in such a way that makes it relevant to answering a question at hand and taking…ACTION.
The action in this case is sending our congratulatory note, 5 years too late. (But it’s the thought that counts.)
But let’s go further. If the only thing we’re concerned about is who won the election, then 90% of this image is unnecessary. It’s not information, it’s infotainment. (Unsurprisingly, the above image comes from network news.)
So we could crop this image a bit and get:
At this point some of you are thinking, “but all that data is really useful and I have a machine learning algorithm that will uncover amazing correlations that will make Nate Silver drool with envy!”
Ok sure…but let’s face it… that information is not really necessary. Many of you probably didn’t even notice that Alaska and Hawaii have been suspiciously absent from these maps. We set out to answer a specific question, so we should focus on finding the information relevant to answer that question and taking action.In the absence of a specific question or hypothesis, information devolves to pure data.
There’s nothing wrong with pure data. If you have enough of it, data mining can uncover all sorts of useful information. But if your startup is trying to figure out if you’ll make payroll at the end of the month, staring at 40 screens of Google Analytics data is not helping you.
Tips for Making Your Metrics Dashboard
- Information is data framed in such a way that makes it relevant to answering a question at hand and taking action.
- Avoid infotainment pie charts or other visual displays and focus on actionable information.
- Understand the audience for the dashboard and design correspondingly.
- Focus on answers to specific questions, keep interesting but distracting data out of your dashboard.
P.S.: I have avoided talking about vanity metrics vs. actionable metrics here. It’s a great subject and closely related. Sometime people think if something is not a vanity metric then it is de facto an actionable metric. This is not true and that’s what I’m focused on in this article. For workshops on stuff like this, go to new.luxr.co.
Metric system, metric tonne, metric schism, metric fun