By Tim Hudson
February 4, 2022
At the heart of all AB testing is the desire to better understand the customer and improve their experience on your website. While the standard AB test itself tells us much about what the collective opinions are of those who are sampled by the campaign, there is a wealth of insight to be learned by dividing the results of the A/B test along user segments.
At SiteSpect, we tend to find the visitor population is a diverse group with a range of behaviors and preferences. This population of visitors can be thought of as different sub-populations, or segments. Some standard segments are crucial to analyze in isolation and are therefore a part of every campaign, such as new or returning visitors, or visits which happen from within your organization.
Beyond these two standard types of segmentation, however, there can be much more to learn about the sub-populations who visit your website. One useful way to analyze your traffic is to consider different ways to partition time. For example, web traffic and customer behavior can vary around the date of the month or the day of the week. Simple segmentations could be done to analyze traffic immediately after common paydays, or to contrast weekday traffic against weekend traffic. Customer behavior may also vary based on clock time, such as between AM or PM, near midday, or at the beginning or end of the standard workday. In these cases, segmentation may reveal that different types of visitors have preferred times for their visits, or it may indicate changes in habits and preferences for the same users depending on when they visit your website.
Another powerful dimension for segmenting visitors is in terms of loyalty, or strength of relationship. For e-commerce websites especially, loyalty can be interpreted in multiple ways and suggest dramatically different behaviors between committed customers and casual shoppers. Consider segmenting customers by visit count, number of purchases, average cart value, the time between visits, number of visits per month or week, or whether or not visitors have signed on as members, previously used member rewards, or by member tier. For websites who have a large proportion of loyal customers it may be particularly important to segment between those who are top buyers and those who are less likely to purchase. In some cases an A/B test may have a significant effect (positive or negative) upon the group of high-value customers while having little effect on those who are outside of that high-value group, an insight that may be less pronounced when only segmenting between first-time and repeat traffic.
Here are some other examples of useful attributes to segment on:
- Device type, such as mobile vs. desktop
- Operating system or mobile brand
- Browser type such as Firefox, Safari, or Chrome
- Geography or preferred language
- Before or after a promotional period or other significant event
- B2B or B2C customer status
- How visitors arrive at the site, such as by entry-page URL or email
- Presence of a cookie
In each of the above cases, segmentation can be done by either identifying audiences prior to the A/B test campaign, or post-hoc to segment the analytics results based on the metrics collected in the campaign. In the SiteSpect tool, applying segments can easily be done with simple filters or by including or excluding specific audiences in the analytics report at the time the report is run.
The power of such segmentation as described here can be particularly important in the further design of A/B testing campaigns. In the simplest cases a better understanding of who customers are and how their preferences and behaviors may vary is yielded. In the more advanced cases, entirely different experiences can be developed for the website which respond to triggers such as customer login, geography, or time of day. These changes can range from the simple such as modifying elements of the homepage experience to the complex such as varying calls-to-action.
SiteSpect’s built-in segmentation filters allow for easy segmentation between visits and users, including or excluding specific audiences, and/or within specified date ranges. Advanced segmentation is made possible by including or excluding audiences identified through ad-hoc or post-hoc analysis.
In summary, to improve personalizing the customer journey you need to be utilizing a built-in segmentation filter. Segmentation between visits and users will take your A/B test results to the next level. This overlooked feature is included with the SiteSpect platform and is easy to use. If you are a current customer of SiteSpect and would like to learn more about this feature, please contact your account manager.
If you are using another optimization platform, we welcome the opportunity to audit your current application and see if there are areas where SiteSpect can help with segmentation filters. Click here to request a demo.
To learn more about SiteSpect, visit our website.
Subscribe to our blog: