Meet Optibot, the First Marketing Optimization Bot
Optibot bridges the gap between data science and the art of marketing by serving marketers with one-click actionable insights for the continuous optimization of marketing activities.Posted in Company News, Customer Retention, Personalization on 6 October 2016 by:
Optibot, Optimove's latest marketing innovation, is a landslide for customer marketing. The first ever marketing optimization bot uses Artificial Intelligence and machine learning to automate the analysis and interpretation of customer marketing campaigns behind the scenes and recommend actionable marketing insights to help marketers increase campaign performance. The new technology is being rolled out to all Optimove clients, including 1-800-FLOWERS, Zynga, BetAmerica, eBags, Adore Me and LuckyVitamin.
Optibot continuously scans the brand's customer model and campaign calendar, and suggests one-click actionable items, based on insights hiding in the data, that can be used to optimize marketing activities. For example, Optibot can provide recommendations for marketing actions such as:
- Drop under-performing actions: When one campaign action is clearly losing to others, Optibot will recommend dropping the under-performing action and readjusting the existing actions.
- Switch to self-optimizing campaign: Once enough data has been collected on a certain campaign, Optibot will recommend turning on its self-optimizing feature to automatically decide the correct customer mix for each campaign action.
- Focus target group: In cases where only a sub-segment reacts well to a campaign, Optibot will recommend splitting the campaign and creating another action for the under-performing segments.
- Treat neglected customers: Optibot will point to the target groups or specific segments in the marketing plan that are not getting the best possible marketing treatment.
- Uplift measurement: Optibot will verify that all campaigns are scientifically measured for uplift by recommending adding a control group in order to measure the exact campaign impact.
Optibot has been years in the making. Five years ago, when we launched our platform for the first time, we had to commit to a name. After much deliberation we choose "Optimove". We felt that it was true to our vision: our product was built to identify the optimal marketing move and our basic algorithmic concept asked what was the best action, for every customer, every time.
And that is exactly what Optimove has been doing since, by converging data science and the art of marketing. The Optimove model is a scientific predictive data modeling platform, micro-segmenting customers into homogeneous clusters. The Optimove campaign calendar is where the marketer's art comes in as she targets each micro-segment with the most appropriate messaging.
However, there has always been one over-arching problem that has been keeping us up at night. In fact, we have been thinking about it for five years. That problem was how to make the power of the Optimove model more accessible to our users. How to more effectively bridge the conceptual gap between the Optimove model and the campaign calendar – the science and the art. In a utopian world, the model would run all campaigns on its own! So, could we come up with a solution that will provide complete automation, with no price of generalization?
Bridging Two Worlds
Data science is a world of black and white, of rigid truths. A customer is either in one segment, or he isn't. The art of marketing is the exact opposite: it has to do with talent, with taste, with psychology – with many concepts which are impossible to measure. These are two different worlds.
In the world of data science, we are looking for predictors. For example, how much our customer spends can predict his future value. Our aim is to use different data points to find homogeneity, clusters of customers that have the same future value or that respond similarly to different metrics. What guides us here is the quest to unite the data in a precise and insightful manner.
In the world of marketing art, on the other hand, we are essentially looking for signals in the data that can help us start a conversation. For instance – I might discover that a thousand of my players have extra coins in their account. I can use this as an opening line for a campaign. But on its own, this information has no predictive quality. Or, I may have some customers who prefer t-shirts and others who prefer pants. This tells me nothing about their future value. But it does allow me to open up a conversation.
When we contended with ways of bridging this gap, we were toying with an idea which I have come to call "The Fallacy". The fallacy is that we can create an algorithm for complete automation, with no price of generalization.
This diagram is a visual representation of The Fallacy. I have my customer micro-segments on the left, my marketing actions on the right, and all I have to do is create a proper match between the micro-segments and marketing actions, so that we would know, through testing, what is the best action for every micro-segment. And since we're working with micro-segments, we won't pay any price for personalization!
Sadly, however, this concept can't work – and it's not that we haven't tried. But quite early on we understood that it's simply inapplicable. This has to do with the fact that while data science can be configured into an algorithm, the art of marketing simply can't. Data about micro-segments can never be used to open up a conversation. In order to communicate, you need context. And context is still very much an artifact of the human brain. It can't be extracted from a model.
Science Working for Art
Marketing still relies very much on the marketer. And our approach for the next step in the evolution of marketing optimization places the marketer at the center – her creativity, context and fresh ideas are the crux of the matter. What our newest innovation does is make the science and automation work for the art. We make the science of marketing infinitely more accessible, so that the marketer can harness its power to the aid of her art.
This is the philosophical basis for Optibot, Optimove's newest innovation and the first ever marketing optimization bot.
We built Optibot to help marketers maximize the impact of their efforts faster and more easily. Optibot takes the guesswork out of analytics, targeting and personalization, automatically applying data-driven campaign optimization techniques, leaving marketers with more time to focus on the creative side of their craft. We're convinced that it will take your customer communications into a whole new level.
Read more Optibot coverage at MarketingLand