One of the big differences between KISSmetrics and most analytics products is that KISSmetrics does it’s best to track people, where most systems focus on tracking just events. For example, imagine that Adam, Bob, and Charlie all visit your website over the course of a week. Now imagine you are measuring an event “viewed homepage” and the following happens:
If you are just looking at the number of times that your homepage was viewed (looking at just the event, not the people) then you would see that your homepage was viewed 14 times in the week. However, as you can see, those 14 page views were generated by just 3 people. So KISSmetrics will report “3 people viewed your homepage” while other analytics systems may say “your homepage was viewed 14 times”. Both statements are accurate, but the subtle differences can lead to confusion.
At KISSmetrics we’ve made the decision to focus on people as we feel strongly that it’s a better metric. However, if you want to see your data the other way you can click on any step in your funnel and scroll down to see an Event Histogram. This Event Histogram how show many frequently people did your given event, the total event count, and the min, max, average and median number of times people did your given event.
One thing to keep in mind when viewing reports is the date range you select. When you pull up a KISSmetrics report for Monday it will only show the data for Monday. However, imagine you have the following situation:
If you look at the data for just Monday you will see that 3 people visited your site, but it will say that only 1 person (Alice) converted for a conversion rate of 33.3%. This is because Bob actually converted on Tuesday. If you look at just Tuesday you will say then Bob will appear as if he skipped the previous steps (even though he did them on Monday). If you have a funnel that can span multiple days we recommend you look at at least a week view. In this example if you looked the whole week you would see that you 3 people visited your site and that 2 people converted for a conversion rate of 66.6%. Even if you are sure your funnel has to be completed within 24 hours, a small number of “midnight” customers may follow across multiple days.
KISSmetrics like all analytics vendors make their best effort to achieve 100% accuracy. However, the reality is that despite all of our best efforts there are many things that not within our control. These include:
If you are really interested in all of the challenges there are whole whitepapers dedicated to this very topic. However, the main point to take away from this is that no analytics product can provide 100% accuracy. However, that doesn’t mean we won’t continue to try to innovate and get closer and close to that goal.
So if analytics products aren’t 100% accurate does that mean that they aren’t useful? Hardly. When scientists do analysis they don’t measure the whole data set, they measure a significant sample. When a field biologist announces that there are only X Wildebeest left in Africa he didn’t actually count all of them. Almost all analysis, be it scientific, financial, demographic, military, or historical uses significant data sets, not the whole data set. Most web analytics systems offer enough accuracy to provide confident insights. However, if you have any large discrepancy between us and another vendor or your own internal metrics we want to know about it.
If tracking every event is really important to you, we do provide server-side libraries (see APIs). These require more work for your developers to integrate, but because they eliminate a lot of the issues with client-side tracking they may be more accurate. However, our server-side libraries do offer less features due to limitations with server-side tracking.
Imagine you have the following funnel:
You might notice that there are a large number of people who “skipped” the “Viewed Home Page” step. These are people who went directly to the “Viewed Signup Page” without first going to the “Viewed Home Page” step. You can control whether or not to show these people with the checkbox in the upper-right hand corner of your reports.
If you are certain that your funnel is completely enclosed and that there is no way a person can do step 3 without doing step 2 first, you may be experiencing one of the issues listed earlier (click tracking issues, date ranges, or just the general challenges of providing 100% accuracy in analytics).
If you have any questions or concerns or feel that this document doesn’t address your issue, please feel free to contact us.
Filed under Troubleshooting