Imagine you are a marketer for an online bookstore. You have ample competition, and are eager to convert as many new customers as possible into repeat purchasers. You are emphatic to their shopping preferences, but pretty much in the dark as to the content that will tip the scales and lead them to the checkout. It might be an email offer of 3 books for the price of 2. It might be a multi-channel campaign for 20% off total purchase. Then again, it might be free shipping. You decide to offer the three promotions in equal parts to the customers in your New lifecycle stage, and see which one will make them bite. You run the campaign.
But now, you encounter a whole slew of new conundrums. How often should you check the results to gather meaningful information? And what if the different actions are performing differently for the micro-segments of your New customers? At which point should you kill one of the actions in favor of the other more effective ones? And should you kill it at all, or maybe change the percentage of customers receiving it? And if so – to what? And at which point do you realize Men and Women may manifest slightly different offering preferences? And if so – what are they? And more importantly – what should you do about it?
Time Consuming Experiments
Considering the fact that this is only one of many campaigns you’re running simultaneously, this spells hours of data examination, endless trial and error and a nagging feeling of groping in the dark after answers to questions that are dynamic in their very nature. Somewhere out there is an ideal customer-message-channel mix that will get you an optimal response rate – but there are simply too many possibilities to test out and only so much you can do.
These complex marketing dynamics are one of the main problems faced by modern marketers, and a solution to which we have been working on tirelessly at the Optimove Lab. After testing various scientific approaches we’ve recently reached a breakthrough, and launched Optimove’s new Self- Optimizing Campaign mechanism which directly tackles this challenge.
Optimove’s Self-Optimizing Campaign for A/B Testing
The Self- Optimizing Campaign is a special type of A/B/n campaign (comprised of 2+ competing actions plus a control group for comparison) which learns and improves via its own results over time. Unlike any standard A/B/n campaign, it is designed to automatically adjust and optimize its performance according to the response patterns of the customers receiving it compared with those of the control group.
It does so by personalizing action mixes on a granular level, for the different micro-segments comprising the target group. While classic A/B campaigns will, somewhat rigidly, send action A to 50% of the recipients and action B to the remaining 50% on an ongoing basis, the Self-Optimizing algorithm changes action mixes according to subgroups’ preferences, as gauged by their responsiveness patterns. E.g., if European customers respond better to campaign A and American customers respond better to campaign B, then these groups’ action mixes will automatically adapt separately for each of these groups. The result is that subgroups’ colliding preference structures will not interfere with each other, and each such subgroup will get the ideal action mix for it. At any time throughout the process the marketer may decide to replace a week-performing action, or add a new one to the mix, giving chance for possibly better candidate actions to compete.
Using AI for Optimizing Campaigns
Back to our book-business example, this is how the Self-Optimizing Campaign will pan out for our marketer: After initially launching the campaign (where the marketing actions are sent in equal proportions) the self-optimizing campaign will take the wheel. It will review the data constantly and on that basis may decide, for example, that 3 books for the price of 2, being more effective than 20% off total purchase and free shipping, will be targeted to 75% of the target group customers. 20% off total purchase, being sub-par compared with 3 for the price of 2, will be targeted to 20% of the target group, and free shipping, being the worst performer, will be sent to only 5% of the target group.
As more data is gathered, the campaign will keep on adjusting these figures to better fit customers’ behaviors. In addition, it will simultaneously examine the numerous subgroups comprising New customers. Based on this data it may decide, for example, that the subgroup of VIP customers (defined as New-High-Tier) respond much more strongly to 20% off total purchase, and may fine-tune the distribution of their targeting so that 20% receive 3 for the price of 2, 70% receive 20% off total purchase, and 10% receive free shipping.
In addition, the data may reveal that New customers outside the US react much better to the free shipping promotion compared to the other offers, and so the Self-Optimizing Campaign will adjust the actions for this subgroup accordingly. This process will continue as more data is gathered.
Self-Optimizing Campaign Utilizing AI for Optimization
Optimove’s Self-Optimizing Campaign represents the ultimate synergy between the marketer and her tools. This innovation, using algorithms from the realm of Artificial Intelligence, doesn’t replace the marketer but rather energizes her endeavor. The marketer delineates the process and remains the supervisor throughout, and the automatic nature of the algorithm maximizes the conditions within the scope defined by the marketer, releasing her to concentrate on the bigger picture. On the other hand, neither can the marketer replace the Self-Optimizing Campaign, as the sheer amount of minute details that constitute the big picture is simply impossible to humanely manage.
Optimove’s Self-Optimizing Campaign is a proprietary, highly advanced algorithm enabling marketers to engage their existing customers with the right message at the right time, every time. It is the first real tool to enable true granular marketing, promising more engaged customers, better conversion rates and powerful, agile marketing that really moves the needle.
Reach out to us if you’d like to give it a spin!
The post The Self-Optimizing Campaign: Let AI Optimize your Marketing appeared first on Optimove.