How to leverage event anomalies in your analytics: use cases and examples
Go back
30.8.2021

Event anomalies can be a powerful tool in your analytics. If you’re not familiar with the term, event anomaly is when an event occurs at a time that doesn’t seem to fit the pattern of other events occurring around it. For example, if you’ve noticed some unusual spikes on your website’s conversion rate for days where there are no holidays or special events happening, then this may be because of an event anomaly.

We’ll look more into how to find these anomalies and what they mean for your business below!

What are event anomalies?

Event anomalies are defined as “a sudden and unexpected change in the number of events.”

And here at Smartlook, we believe that nothing speaks louder than data – so we’ll illustrate it with some numbers.

For example, let’s say you have 500 monthly visitors that visit each day for one minute on average. One day there are 600 visits but they only stay on the site for half a minute. The event anomaly would then be something like this: 600 (visits) – 200 (average minutes per visit) = 400 minutes or 7 hours of additional time spent on your website. 

The more time visitors spend on your website, the more purchases, clicks on ads, and interactions you get, the more revenue you earn. And yet, you’re losing it. 

By analyzing this topic thoroughly, you can determine if anything is wrong with your online store, website, or app.

How to tackle event anomalies?

Event anomalies are not necessarily bad; they just require some extra attention if the frequency of occurrence suddenly increases significantly.

If you have several spikes on your conversion funnel, it means that there might be something wrong with your website’s UX flow or conversion rate optimization strategy. In this case, user behavior analysis will help identify the issue and fix it accordingly.

On the other hand, if one spike exists near another one in time – say two weeks apart – then it may indicate a certain trend in user behavior.

If this is confirmed by other events such as declining revenue and decreased engagement rates then there might be an issue with your SEO strategy which needs to be resolved immediately.

It’s important to remember, too, that once you notice an event anomaly, the next step is to find out what caused it and how long has this been happening – and then take actions to prevent it or minimize its impact now and in the future.

Event anomalies are a powerful way to track and predict customer behaviors. Let’s explore how it can be used in your analytics.

Event anomalies use cases

#1 Missing conversions

With event anomalies, you can find out you’ve been missing conversions from someone visiting your blog page by clicking through to different pages than expected.

If someone visits your blog page and clicks through to other pages than expected (instead of taking the conversion), then this is considered an anomaly that can be detected by analytics.

In addition to discovering bugs, anomalies detection can also identify errors (minor bugs such as broken links, major bugs like SSL encryption causes users to bounce) that are most likely easy and quick to fix. The data is there for you to work on.

#2 Abandoned shopping carts

You will be notified when the rate of abandonment rises through anomaly detection. By watching the recordings of the funnel or event, you can easily determine the reason why your shopping carts are abandoned.

The key is to find out what’s happening right before the abandonment.

For example, if you notice that a lot of carts are abandoned when people go through your checkout process, then it can mean there’s something in your design or copy on this page that makes them think they need to log into their account first. However, with anomalies detection, you can also find out that something has only just bugged on the site, causing even higher abandonment of shopping carts in your online store.

The Smartlook funnel can be built as follows:

Add item to Cart > View Cart > Submit order

You can set up these anomalies so that you are notified of spikes in conversion rates. With one click you can view the drop off between the “View Cart” and “Submit order” steps.

This will lead you to make some changes and quick fixes that will help keep more shoppers from abandoning their cart and allow you to reach a higher conversion rate.

#3 Integration issues

You find out about a problem with one of your third-party integrations before it becomes widespread and causes damage to your site’s performance or reputation. For example, you may find out that a lot of people select one of the packages of your SaaS, but cannot get into payment because of the issues with the payment gate. Anomalies detection features will notify you about that so you can look into this matter a bit deeper.

If a spike in abandoned carts was due to payment gate problems, then your task would be to resolve this issue as soon as possible and provide new promotional offers for the cart abandonment problem you might have with future sales.

#4 Unexpected user behavior pattern

For example, if you start seeing a spike of three-star reviews from customers who did not purchase the product before that it may indicate a change (lower quality) in your product or service. Your anomaly detection features will show you that something is not working as it should.

Here’s how it could work in practice: set up a series of custom events in Smartlook. Let’s put them as follows:

  • 1_star_feedback, 
  • 2_star_feedback, 
  • 3_star_feedback, 
  • 4_star_feedback, 
  • 5_star_feedback.

You can then set up anomaly detection over these events, and you will be notified if the number of ratings increases. Additionally, you can easily watch all the recordings of users who have recently posted 3_star_feedback. 

It appears that the 3 stars in the 5-star grade just feel neutral, so you should investigate the cause of the spike in neutral feedback and drop in positive ratings. 

That might mean one of these things:

  • there is something wrong in our checkout process 
  • there are some bugs in our holiday sale where the benefits promised are not applied correctly
  • customer service team is not efficient and can’t resolve all problems
  • there are some problems with payment gateways, causing potential clients to cancel purchases

But there may also be other reasons, and session recordings can help you discover them. 

If a spike in conversion was due to holiday season promotion then your task would be to find out why conversion rates are low during non-promotional periods. You can simply accommodate this fact or take action to grow sales, but identifying the reason is the very first step in this process.

#5 Traffic anomalies

You can also gain insight into traffic patterns through anomaly detection. Let’s say you run an online store operating only in the United States, but you receive a lot of traffic from Canada. It could be caused by a variety of reasons: a spike caused by a misplaced ad (displayed in Canada, for example), or a pleasant blog post praising your product from an influencer, which resulted in a spike in traffic. Detecting anomalies and analyzing them afterward are the only ways to prevent guesswork.

Now it’s your turn

Anomalies detection can be used by both marketers and analysts alike for a better understanding of how users engage with content or products.

Some analytics tools offer event anomaly detection features. Smartlook is one of them.

With Smartlook’s anomaly detection, the process is automated and you’re informed in real-time about spikes or surges in your events – and now – also funnels.

It’s free with all Smartlook packages.

Go Back

Recommended articles