With a manual A/B activity, you might be losing conversions because you can’t deliver the winning experience to your entire audience until the activity completes. Your traffic distribution remains fixed even after you recognize that some experiences are outperforming others, and the activity must run its entire course before you can act on a winner.
If you want an option to serve the winning experience more often and earlier in the activity while simultaneously removing or reducing the set-up and calculation cost of picking sample sizes, confidence levels, and other statistical concepts, Auto-Allocate is your best option.
Auto-Allocate uses the principle of the multi-armed bandit. If the term is unfamiliar, a one-armed bandit is a colloquial term for a slot machine (think: Las Vegas). Visualize auto-allocation of traffic as having multiple slot machines, in this case, test variations, and at first pulling all handles equally. Over time, one or more machines, or test variations, might pay out more than others. When this situation happens, a gambler would naturally start pulling the handles of the ones that win more often. In traffic allocation terms, Target serves more visitors the experience or experiences that are winning more.
Consider the following illustration of a two-week A/B activity. With Auto-Allocate, as a winning experience emerges, Target diverts more of the traffic to that winner early on in the test.
The upside is clear: more visitors see the variations that perform best. And as a single variation pulls ahead, even more visitors get diverted to that winning experience, while the test was still running. This method is especially helpful if the A/B activity being run is happening during a core business moment such as a holiday, product launch, or world news event.
Auto-Allocate finds the winner faster than a manual A/B split, and also allows you to exploit that winner immediately–capturing upside revenue that would have been lost in a traditional or manual approach. Because Auto-Allocate directs more traffic to the experience with the highest conversion rate, it can increase your revenue while the activity runs and learns.
In the following example, Auto-Allocate gained more revenue during the test by pushing more traffic (40%) to Experience D, which had the highest conversion rate.
When you must rank-order how every experience performed relative to the others, a manual A/B test is most applicable. Auto-Allocate finds and exploits top performers but does not guarantee differentiation among the lower performing experiences. Use manual traffic allocation for complete control of how much of your visitor traffic sees each test variant and for customization of the statistical thresholds that are relevant for your business.
Ready to launch your first Auto-Allocate activity? Learn how here.