At Selligent Marketing Cloud, we are always striving to find new solutions for our clients’ most pressing marketing challenges. That often means internal changes in our process, new research, and new investment in talent and technology. Our goal is to help B2C relationship marketers maximize every moment – and nowhere is that more critical than when it comes to Customer Experience.
Challenges Inherent in Traditional Approach to CX
The traditional approach marketers use for improving CX is to rely on careful analysis of data and what I would call "per-channel trial-and-failure strategies." For example, a marketer would set up an A/B test with different variants of an email message or website page to guide their choice of content. The message with the best response (in terms of view rate or conversion, for example) would then be used in the journey. While a standard A/B test is typically used to test a limited number of designs or versions of a message, a more advanced multi-variate testing (or "MVT testing") would allow you to test a large number of content variations in a single message, which is a further enhancement in the automation of decision making.
While these tests make the choice of the message content more data-driven, the marketer still must be heavily involved throughout the testing process to control the results and ensure statistical significance. Even more constraining is the fact that the goal of these tests is to select one single version of the message which will be then communicated to the *entire* audience. This enhanced version of the old "spray-and-pray" approach cannot be successful anymore, as it still encompasses a lot of the marketer's own experience, and therefore bias (e.g. the content to be sent in either version of the A/B or MVT test is still defined by the marketer in the first place).
Moreover, with journeys becoming more and more complex across an ever-increasing number of marketing channels, the process of running tests, analyzing results, and making decisions at each step of a journey is becoming too cumbersome to be handled manually. This is becoming an even bigger problem for marketers as consumers are getting more and more sophisticated and expect to be treated as individuals rather than put into segments, as is the case in standard marketing strategies. In a study we did on 7,000 consumers, 74% of all consumers surveyed told us that they are willing to share their personal data in order to get a personalized experience. That's a strong signal for marketers to go beyond RFM segmentation and start exploring innovative ways they can improve their customer's experience.
A Better Way to Improve CX
There's good news! Large amounts of data, coupled with machine-learning (ML)-based algorithms, are a solution to this challenge. This combination can help the marketer to automatically "learn" from historical data and make automated decisions for future events. These "predictions" can be used in different ways across customer journeys. When starting to develop Selligent Cortex (the AI-layer built into Selligent Marketing Cloud), we decomposed our platform into three pillars, where the use of cross-channel interaction data and machine learning can empower marketers to move closer to true 1:1 personalized experiences. These pillars are:
- Real-time content personalization of email/web/mobile channels: finding content in real-time which is likely to be relevant to a specific user based on his or her profile and behavioral data
- Journey optimization: predicting the best time and channel to use to send a specific message to an individual consumer
- Behavior-based audience segmentation: predicting which individuals should be part of an audience based on their behavior, as well as the characteristics of the message to be sent
These groups of features allow B2C relationship marketers to automate the personalization of their consumer journeys at scale. Data from a survey we did on hundreds of our own clients' CMOs suggests that this is a high-priority goal for marketing organizations: two-thirds of marketers say that their top marketing automation goal for 2019 is to speak to their customers in a more relevant way, ahead of driving more sales for the company.
Our Approach to Solving the Personalization Puzzle
The bad news is: building personalization features which use data and AI is hard, and I don't believe it’s something any organization is capable of doing on its own. At Selligent, we invested a lot in the last two years in hiring the right talent in data science, data engineering, machine learning, and UX to build these ML-based capabilities. These features are quite different from everything else we are used to building, as they require continuous research and development effort with our clients, prospects, and partners.
To get these features right from a performance and usability perspective, there are lots of processes we had to put in place. For example, one important prerequisite was the availability of data which could be safely used throughout the development process and performance testing of a feature, and this has to be done taking into account important privacy laws (e.g. GDPR), which we must comply to and enforce. Another important requirement for building ML-based features which bring a concrete improvement to personalization (and therefore direct ROI for our clients) is that the data needs to be available in real-time.
From an engineering perspective, ingesting, storing, and having access to billions of events (e.g. click in an email/SMS/mobile push, page view on a website, form submission containing personal data, price update in the product catalog, etc.) in real-time is a real data architecture challenge. To tackle this challenge, we built a "real-time data layer," which is natively integrated inside Selligent Marketing Cloud. It works hand-in-hand with and complements each of our clients' custom data models. Today, all our ML-based features in Selligent Cortex use the real-time data layer to train models and treat cross-channel interaction data in real-time.
These are just a few examples of the changes we had to make to our internal processes and overall software architecture to help our clients move from standard marketing automation to AI-driven CX. All of these features are available out of the box in Selligent Cortex and can be used by our clients with just a few clicks in the Selligent Marketing Cloud UI, without having to worry about the data pipeline logic or training machine learning models.
I often talk to marketing leaders who wonder how they can start using AI/ML on their cross-channel data to improve their marketing, and whether they need to hire data scientists. I always tell them that they can already do a lot by letting their existing marketing operations people start using AI/ML features which are built into their marketing cloud. For example, if one uses Selligent Marketing Cloud across different channels, cross-channel data is already available in real-time. Using real-time content recommendations on websites, mobile push notifications, and emails (a feature called Smart Content) or activating Send-Time Optimization on a campaign can be done in minutes. There is no need for hiring more people or starting lengthy development projects with no guarantee of a result with a positive impact. Of course, some clients can have very specific needs, in which case they can use Selligent Cortex APIs to build custom reporting or integrate predictive features into custom channels.
If you want more detail on how to start an AI/ML project in your organization, or if you have any questions about how we foster innovation at Selligent, contact us at innovation[at]selligent.com. We would be happy to help you.