With every year that passes, data becomes more and more important to businesses of every size in every industry. Consumer profiles, usage metrics, web traffic — everything that can be quantized into discrete data points is totted up and gathered into silos, where huge computational resources are used, milling it into useful metrics.
Trying to get a grasp on these data processes is daunting: Here are three common pitfalls to avoid as you begin using data in your business.
1. Poor Quality.
The first and most obvious issue you may have with your data is its quality. Data is an odd resource because one way — the simplest way, in fact — to improve its quality is to improve its quantity.
The laws of probability indicate that the more information you gather on a population, the closer your basic metrics (like averages) are to the real values for the population. That said, some thought should go into the sources for your data as well. Customer surveys, for instance, tend to capture the extremes: very satisfied customers or very upset ones. Usage data gathered anonymously is biased away from users who opt-out of reporting it — in software, this tends to be the most enfranchised users.
In general, look for outliers in the data and ask questions about why they are arising. Then, take steps to eliminate them.
2. Bad Metrics.
Once you have good data, it’s equally important that you’re doing the right things with it. Your data might be impeccable, experimentally sound and voluminous, but if you’re asking the wrong questions about it, all the effort that went into collecting it is wasted.
For example, the tendency is to default to averages when working with data, as it’s an easy metric to understand, but don’t underestimate the median when working with certain populations. Say you have a service where most users are spending a few dollars, while some users are spending thousands. Your average revenue per user may be in the tens of dollars, and from that, you might be tempted to conclude that, if you select a user at random, they are probably spending around that amount on your service.
However, this is not the case; if you select a user at random from this population, it’s vastly more likely that they’re spending only a few dollars, a fact that the median would have captured. This is only one of the ways common metrics like the average can be deceptive, so always take time to think through the literal meaning of your statistics and be careful of jumping to conclusions.
3. Inconvenient Access.
Finally, you can have wonderful data and excellent metrics and still not be able to use any of it. Particularly at the enterprise level, the sheer amount of data provides a challenge of its own, as straightforward computing methods for sifting through it begin to break down.
The science of data usage at this level is its own academic and professional discipline, and there are myriad approaches to storage and access of data. While you don’t necessarily need to know everything about knowledge graphs, it’s worth learning a thing or two about enterprise data management. At the very least, you should understand that at a certain enterprise size, working with data gets more complicated than running a few commands in Excel. Keep this in mind when working with outside sources for your data management. Be patient and willing to listen, and your consultants will thank you.
There is a lot to take in when it comes to enterprise data and many subtle distinctions that can lead you down the wrong path. Avoid these common mistakes when working with your data, and you can make your data work for you.