Everyone knows that you can’t improve what you don’t test and measure, which is why I’ve always been a huge proponent of A/B testing and analytics. Having a lot of data, however, isn’t enough. To correctly determine the return on investment (ROI) of your marketing efforts, it’s important to not only collect data efficiently, but to analyze that data correctly.
Unfortunately, many businesses make some fundamental mistakes when analyzing the data they’ve collected, which can lead to a miscalculation of ROI and sometimes even a misuse of data. As a result, companies may keep programs that are underperforming, or even end programs that are doing well.
Here are five mistakes businesses make when they analyze their analytics, as well as how to solve them.
1. Coming to the Data With Specific Biases
No one likes to see a program they’ve championed for months fail. On the other hand, some people get annoyed with the pace of a certain campaign and will use any poor result they can find to axe it. When you come to your analytics data with a specific bias you’re trying to confirm—intentionally or subconsciously—you’re guaranteed to approach the data incorrectly and draw false conclusions.
Even if you love a program, a disappointing number can be a learning opportunity rather than something to be glossed over in favor of other, better results. And if you really dislike a campaign, solid numbers can mean it’s bringing the results you’re working for. Leave your biases at the door, and let the numbers tell you the truth about ROI and performance.
2. Confusing Correlation With Causation
Did you know that the divorce rate in Maine is closely correlated with the per capita consumption of margarine? It really is, but of course, that doesn’t mean that eating more margarine causes people in Maine to divorce. It’s a spurious correlation.
When you’re reviewing data from your analytics, be careful not to draw conclusions about cause and effect based simply on correlated trends. You need to dig deeper and find out if there’s actually a link between the two items, or if there are other outside factors that influence the variables in play. You may need to run additional tests to find this out.
3. Confusing Statistical Significance With Actual Significance
Because analytics creates extremely large data sets, it’s possible to see very small differences in a trend that are statistically significant. Your program may even automatically flag any correlations, changes or trends that are considered statistically significant.
However, the fact that something is statistically significant in a large data set does not automatically mean that it’s actually significant to the day-to-day operation of your business. Will a 1% difference in conversion rate cause enough actual difference to warrant a change in your campaign? Businesses have to weigh the cost and trouble involved in a change against the actual significance of the result.
4. Failing to Clean and Format Data Correctly Before Analyzing
A lot of the time that goes into data analysis is spent on “pre-work,” which is ensuring the accuracy of the data and formatting it correctly so it can easily be analyzed and shared with others. It may be tempting to simply eyeball the data and draw conclusions from it, rather than taking the time to set everything up. Unfortunately, doing this can lead to significant mistakes in analysis, and cause more work revising and reworking the data on the back end.
Instead, select a portion of data, and check it for accuracy against other sources. Make sure names and purchase dates are lined up correctly, for instance, and that there’s no missing information. Then, take the time to lay out your data correctly so that you can find the trends and create reports for others. You’ll end up minimizing mistakes and saving yourself time overall.
5. Confusing Visits With Views
When you’re analyzing your web marketing results, keep at the top of your mind that visits and views are not the same thing. A visit occurs when someone comes to your website from a URL outside your site. Depending on your analytics program, a visit ends after a certain amount of inactivity or it ends when someone closes the browser or leaves the page. Either way, the visitor can generate multiple page views during their single visit.
Be sure not to confuse these two metrics when reviewing your marketing data, or you’ll draw incorrect conclusions about how well your campaigns are working. When you keep the distinction clear, you can focus on turning those visitors into customers.
Reviewing analytics data isn’t easy. It requires the ability to leave your personal biases behind, and to do whatever is necessary to avoid important mistakes. However, when you take these steps, you’ll reap valuable insights about your marketing campaigns and your overall ROI that will help you grow your business and profits as you go forward.