Accuracy vs. precision in web analytics

When looking at web analytics reports, it’s easy to get sucked in by the numbers and start treating the metrics as precise counts. There is a difference between “accurate” and “precise,” and this difference is particularly useful when working with analytics data.

When looking at web analytics reports, it’s easy to get sucked in by the numbers and start treating the metrics as accurate counts. But, as we’ve said before, analytics tools are not 100% accurate and precise.

Accurate vs. Precise

accuracy-vs-precision
There is a difference between “accurate” and “precise,” and this difference is particularly useful when working with analytics data. Basically, accurate measurements return correct values and precise measurements return consistent values. Either, neither, or both can apply to a measurement. The ideal is accurate, precise measurements, but that’s not usually possible with analytics tools.

google analytics vs. proprietary analytics
google analytics vs. proprietary analytics
A quick comparison above shows results from two different analytics tools that vary widely, but that show similar trends across time when overlaid on top of each other. The blue line represents the Google Analytics data for number of visits and the red line represents the website’s content management system data. The actual visitor numbers are off by a huge amount, but the trends match. The measurements are precise (consistent), but we don’t know whether either one is accurate (correct).

The reason that the measurements are off is that each system measures visits to the site differently. It’s as if each is using a different yardstick to measure the same thing. Each time the measurements are made, they are precise, as measured by the yardstick. But, they differ from each other, as the yardsticks aren’t the same.
rulers

Since the data are consistent, you can analyze changes and trends, without having to focus on exact values. In this example, you can analyze how the number of visitors changes over time, either comparing numbers from year to year or spotting seasonal trends over a single year. You can read more about gaining insights through comparisons in “The Right Way to Use Analytics.”

Leave a Reply

Your email address will not be published. Required fields are marked *