Imagine your analytics tell you that you had 100k sessions on your site today. Can you say how many of those sessions were a customer’s first session? Or how many were sessions started by customers who had been on your site earlier that day?
We’ll show you how to classify your sessions into seven segments, based on how long ago a customer’s last session occurred.
We will be comparing the time between session_start events by looking at the number of days since the last session_start for a particular customer.
The analysis consists of an Expression and an Event segmentation which are then applied to a Report. The expression is used to calculate the time between sessions, the event segmentation divides the session_start events, and the report visualises the segmentation. At the end of the article, we’ll show you three examples of e-commerce traffic we’ve analysed and interpreted with this analysis.
Not all sessions are equal. Some customers are first-time visitors, others have visited many times in the past. First-time customers are more likely to just want to inform themselves about the brand, while returning customers are more likely to purchase. Our research in travel, electronics, and fashion e-commerce verticals shows that customers that return to your site within 0.1 days (2.4 hours) tend to have the highest conversion rate among other returning segments.
This analysis gives you visual feedback on the effectiveness of campaigns, showing you if a peak in traffic should be attributed to new or certain segments of returning traffic.
Keeping an eye on the ratio of new to returning customers can tell you a lot about healthy your traffic is. Our research shows that e-commerce brands whose traffic is primarily new visitors (>50% of traffic are first visits) are likely to have issues with returning visits and building a loyal customer base.
For customers returning to the site after more than 30 days of inactivity, you can trigger a campaign showing them what’s changed since their last visit.
Retarget using Facebook Ads / Google Adwords to customers who have returned within 0.1 days. Our analyses show that this segment of customers is most likely to convert.
Customer behaviour varies based on which segment they fall into – customers who are returning after less than 0.1 days are more likely looking to purchase, while those returning after >30 days will behave differently. With Exponea you can personalise the web experience based on a customers segment, e.g. showing a “Welcome back” pop-up for visitors returning to your site after 7 days of not visiting.
What is Returning Visitors Time-Based Analysis
Session_start is an event that is tracked by Exponea to the customer when a device connects to a website with Exponea’s tracking. In this analysis, we break down the time of “how long ago” into “number of days since the last sesion_start”.
This analytics use-case is in essence an advancement on the industry standard “new versus returning” way of viewing your session visits to the website. Returning Visitors Time-Based Analysis classify the event session_start into custom time intervals by the variable # of days since last session_start or in other words “how long ago, in days, has the previous session_start occurred”.
Why you should use this analysis
The primary value of doing this analysis is that it is a more in-depth version of new vs. returning session_starts. This advanced way of viewing your sessions will effectively break down the “returning” customers into more detailed segments (in this example six/seven). Understanding how to perform this will allow you to answer the following business questions:
The sessions that occur on my website, how many of them are returning customers? How many of them or 1st sessions? How many of the sessions are returns within the day, of leaving the website?
How can I initiate a Return Welcome Email Scenario to a customer who has returned after more than 30 days of engagement? With this classification of session_starts it is also possible to use it as a basis for personalised triggers in Exponea Scenarios.
How can I target Welcome Back Web-Layer with the last products viewed by the customer to customers who have returned after 2 or more days? With this classification of session_starts it is also possible to use it as a basis for Audience selection in Exponea’s Web-Layers, Experiments and the Tag Manager.
How can I retarget on Facebook Ads / Google Adwords to customers who have returned after more than 7 days, with a personalised message? With this classification of session_starts it is also possible to use it as a basis for a Segmentation attribute that you can enrich Facebook Pixels and Google g-tags with.
This analysis is useful to monitor when spikes occur in new and returning traffic segments as well as overall health of your e-commerce enterprise, particularly in regards to the rate of returning customers you have in proportion to new customers. If you consistently have a high proportion of only new visits for several months, for many verticals that would signal an acquisition issue.
Our qualitative and quantitative research within travel, electronic and fashion e-commerce verticals shows that customers returning within 0.1 days (2.4 hours) and within the day tend to have the highest CR% compared to the other returning segments. Particularly in Travel, these customers browse – decide on ticket – leave – verify with family/friend/work – return to site – convert. In general, customers who return after 30 days are regarded to be customers who have „forgotten“ and „reengaged“ with the brand.
1st Session Starts are sessions of customers who have never visited the website before; so peaks in 1st Session_start traffic indicates an increase in either spending on or the performance of acquisition marketing channels like Paid, Social, SEO, etc. Returns within 1 day are sessions of customers who re-visit on the same day; so peaks in within 1 day traffic indicates an increase in browsing – leaving – browsing kind of engagement.
Returns of 2 days or more are sessions of customers who are re-visiting after more than 2 days without a session; so peaks in Returns of 2 days or more indicate an increase in returning marketing channels like Email, Referrals.
How do you implement the solution in Exponea?
We will show you how to create a basic segmentation and also a traffic report with a view per day. It is done using an Expression and Event segmentation in Exponea. In order to get meaningful insight at least 2 months of the data are needed for this analysis.
Creating the segmentation
We will need to create an Expression to calculate the time between session_starts, which we will then be used the Event segmentation itself.
1. Create the expression
In the Data manager, create an expression for the event session_start.
Add the timestamp attribute as per the GIF below. This means that Exponea will return the `timestamp` of every `session_start` event we are currently looking at.
Add an arithmetic operator using your keyboard and add a running aggregate last session_start timestamp. Here Exponea will look at the timestamp of the previous `session_start` event so we can subtract those two and get the time difference. If the dynamic variable does not exist, the Expression will return no value. This will signify 1st.session_starts.
Using your keyboard, divide the whole formula by 86 400 and add brackets as per the screenshot below. As timestamps count each second, if we want to report our data by days, we need to divide the result by the number of seconds of one day, which is 86 400.
2. Create the event segmentation _session_starts by(# of days since last sessionstart)
Create a new event segmentation in the data manager. You will need to create 7 segments, each for the group you want to look at, as shown in this screenshot. The GIF below shows how to create the first 3 of those segments.
Creating a report with a daily overview
Now you can create a report based on the segmentation which will show you what groups of customers are coming to your website, unique per day (based on the last session_start). This can then be saved in Dashboards.
This Traffic Report allows you to see the number of all session_starts with each returnee value. As you can see in the table above traffic for the 12th to 19th November, the spike in traffic on the 16th of November is primarily attributed to a spike in 1st session_starts. This makes it clear that there was something working extra on the acquisition marketing on the day, such as paid or social. However, we also see increases in the retention segments, as 30+ day returns have doubled in the period on the 16th of Nov. In general, the very high proportion (60-70%) of daily traffic could signal.
The whole setup of the report is highlighted in the screenshot and the table below:
Set for which period do you want to report (1 week recommended)
Format: Year month day
Select the event segmentation you created in the previous steps. It will appear in the list of events, under Event segmentations
Count event session_start
In this section we would like to show you anonymised examples from three different e-commerce verticals – fashion, household, electronics. In each we intend to interpret the data to inform questions regarding:
1. What does this traffic signal about our traffic health?
2. Are there any actionable patterns happening in the data?
3. What personalisation opportunities do these represent?
For this fashion brand we can observe 3 high-level insights:
Traffic spikes over the weekends: peaks of returnees on the 8th and 28-29th of September, suggest channels like Email, Referrals, Retargeting are potentially having an effect.
Traffic spikes over the weekends: peaks of 1st session_starts on the 8th and the 28-29th of September suggests channels like paid traffic, search, organic, social are having an effect.
During workdays, almost a third of all the session_starts return within 1 day; this indicates an increase in browsing – leaving – browsing kind of engagement;
A low proportion of customers returning to the website after 7+ days to those visiting the website for the first time or returning within 24 hours; may indicate an acquisition issue for the company.
For this household brand we can observe 3 high-level insights:
A changing pattern starting the 13th of September suggests new tracking in new markets was established. (note: This project has intentionally decided to track separate websites into a single project to increase the quality of recommendation results through an increased sample size)
Almost half of all the session_starts are done by new (1st session_start) customers.
Returning customers are nearly a third of all session starts. Not great. Not terrible.
Customers returning within 1 day are comparably lower than the other two businesses, suggesting a much lower appeal in browsing for customers. (note: this business has a much smaller selection of products in comparison to the other 2, by the factor of 20.)
For this electronics brand we can observe 3 high-level insights:
Workday/weekend pattern: session_starts consistently peak during the workdays. During the working week, a third of all session_starts return after more than a day; with Mondays seeing a peak of returnees after 2 to 7 days; suggesting customers return to visit the website at work after the weekend;.
Throughout the whole week, we see that approx. a third of all the session_starts are done by new (1st session_start) customers.
Another third returns within 1 day; meaning they are session_starts of customers who recently returned and are actively browsing.