Deloitte reported that 42% of questioned executives think AI will be “of critical importance” within two years. We researched the marketing space to discover how and where AI might be driving business value for companies in marketing. Some companies are finding results using AI-powered personalized marketing strategies.
This report in particular focuses on companies that claim to offer a specific subset of machine learning for segmenting customers and building marketing campaigns, one called predictive analytics. We’ve covered predictive analytics for marketing broadly before in a previous report.
What Marketers and Business Leaders Should Know
We’ve found that many of the artificial intelligence vendors selling martech to businesses are often not entirely truthful about whether or not their software actually involves machine learning, or at the very least they are not transparent about how their software works. We discuss this further in our report on AI for marketing agencies.
More specifically, there are several companies offering software they claim leverage AI for customer segmentation purposes and for building out marketing campaigns. Of those discussed in this report, two are established firms, IBM and SAP, and two are startups founded after 2005, AgilOne and Optimove.
As a rule of thumb, younger companies are generally the most likely to offer legitimate artificial intelligence solutions because they’re more likely to either have been founded by people with data science talent or have brought on data science talent as some of their earliest employees. That said, we are skeptical of one of these startups.
Optimove’s Chief Data Scientist holds a Master’s of Science in Information Technology, which he earned in 2012. He worked at Optimove as a “Researcher and Algorithm Developer” from the time he graduated to 2014, when he became the company’s Chief Data Scientist, according to his LinkedIn profile. Although it’s possible his schooling and experience at Optimove involved data science, his background is not indicative of the kind we would expect of someone with the title of Chief Data Scientist. Generally, business leaders can quickly figure out if a company’s AI offering is legitimately machine learning or not by looking at if any of the chief executives or AI leaders at the company have a robust background in AI.
At Emerj, we like to see executives with PhDs in computer science or some hard science, such as physics, or, even better, a PhD in machine learning. It’s a high standard, but many of the companies we’ve covered meet it, and these are the companies we are the least skeptical about when it comes to their claims.
In addition to the background of their Chief Data Scientist, Optimove lists many data scientists, “data integration specialists,” and “marketing data scientists” on LinkedIn. This would at first seem like a powerful trust signal, but upon further investigation, most of the people the company lists as data scientists are recent graduates of undergrad programs in various fields, the most prominent of which is industrial engineering, not computer science.
They also employ some “data integration specialists” that are still doing their undergrad. When it comes to data science talent, we look for people with a Master’s in computer science or experience working at marquee AI companies like Google, Facebook, Amazon, or NVIDIA at the very least.
This is not to say that Optimove will not garner results for its clients; they very well might be able to help businesses market to new segments of their customer base. All we are saying is that it seems unlikely that the way they generate those results is based on machine learning.
In contrast, AgilOne employs quite a few data scientists with Master’s degrees or specializations in computer science and/or data science, and some have experience working with data at companies such as Hewlett Packard. In addition, the company’s Chief Product Officer is former Vice President of Product Management at Oracle. For these reasons, we believe that AgilOne has a decent likelihood of offering a solution for customer segmentation that actually uses AI. The company is also backed by Sequoia.
Venture capitalist firms in Silicon Valley often have more knowledge about the possibilities of machine learning than firms in other parts of the world, and venture capitalist firms do their homework on the companies they choose to fund. For this reason, we consider it a trust signal that Sequoia decided to back AgilOne. It lends more credibility to their claims of doing AI.
To Optimove’s credit, however, they do list case studies where AgilOne does not.
As for the larger firms, we could not find the people leading the teams at IBM and SAP responsible for their campaign building solutions, but we believe their claims of doing AI are credible, especially IBM considering their history with the technology.
That said, we don’t believe that business leaders should take the size or age of a firm as an indicator of their legitimacy when it comes to offering AI solutions to businesses or using AI internally. IBM and SAP are firms with evidence to back up their AI claims, but many of even the largest enterprises don’t have any.
Many established firms pivot to either branding themselves an AI company or open up an AI department to latch onto the recent AI hype. It’s not easy to actually do this; data science talent is hard to procure, as many recent graduates with data science expertise are likely to look for work at the tech giants that are predicated on AI: Google, Facebook, Amazon, Netflix, etc.
What this means is that larger firms often struggle to retain data science talent. Enterprises in insurance and banking will often hire data science talent just to say that they’re doing AI, to stay on the cutting edge; enterprise IT firms, on the other hand, will often try and rebrand themselves as exclusively focusing on AI to appear as if they’ve always been ahead of the curve. We want business leaders to keep in mind that just because a company—even one with a long history of offering tech solutions—is established and well-known doesn’t mean that their claims of offering AI solutions are legitimate.
There are very few companies that business leaders should feel comfortable believing about their use of AI, and again, they’re the obvious ones, the tech giants that don’t have a value proposition without machine learning: Google, Amazon, Facebook, and Netflix. IBM and SAP have the history to back up their claims, especially IBM for its Watson product, but unlike the tech giants, their whole business isn’t AI.
Customer Segmentation
AgilOne
AgilOne offers a namesake customer data platform (CDP) which it claims can help enterprise businesses accurately segment their customers for more effective marketing using predictive analytics.
We can infer the machine learning model behind the software was trained on thousands of historical transactions from a variety of customer segments. The data would then be run through the software’s machine learning algorithm. This would have trained the algorithm to discern which data points correlate to more specific subsets of customers. For example, a man could browse an eCommerce clothing store and buy a tie. That transaction would contain demographic data points such as the customer’s gender, age, and perhaps level of income, in addition to transactional data such as the cost of the tie and marketing data such as the email subject line used to generate the open that lead to the conversion.
This would have trained the software to discern which confluence of data points correlate to an entirely new customer segment that the client was unaware of. They could then market to that new segment in future campaigns.
The software would then be able to predict what the customer will buy next from the company based on that data.
Below is a short 4-minute video demonstrating how AgilOne works:
AgilOne does not make any case studies available on their website. They do list Joann and Moosejaw as some of their past clients, however.
Gangadhar Konduri is Chief Product Officer at AgilOne. He holds an MS in Electrical Engineering and Computer Science from MIT. Previously, Konduri served as VP of Product Management at Oracle.
Optimove
Optimove offers Relationship Marketing Hub, which it claims can help retail businesses discover new customer segments to market to within their existing customer base using predictive analytics.
We can infer business owners can integrate the software into the client company’s CRM.
We can infer the machine learning model behind the software was trained on thousands of data points from retail customer profiles. In addition to past sales and the conversion rates of previous marketing campaigns, these data points could also include personal information such as gender or geographical location. This would have trained the algorithm to discern which data points correlate to higher conversions for a particular customer demographic.
The user could then run customer profiles into the Relationship Marketing Hub. The algorithm behind the software would then be able to determine which customers belong to the client’s defined customer segments and evaluate which strategies may work best for those segments. The system then provides recommendations for marketing decisions related to the desired segments.
The software would be able to predict which customers belong to the client’s defined customer segments, and evaluate which strategies may work best for those segments. This may or may not require the user to upload information about their existing customer segments into the software beforehand.
IBM also lists Georgia Aquarium and ING Direct as some of their past clients.
Optimove claims to have helped Paul Stuart segment their customer base, establish multi-channel communications, and centralize their data management. Paul Stuart integrated Optimove’s software into its CRM in order to segment their data by customer lifecycle stages. For Paul Stuart, these stages are “active,” “VIP,” and “lapsed.” They also wanted a central hub on which to customer segments on a variety of channels. According to the case study, Paul Stuart found even more specific customer segments to market to as well as the ability to view specific customer profiles with the solution’s Single Customer View.
A specific customer segment could be a group of produce customers who are predisposed to purchase more apples than other produce customers. The new segment could include customers who fall into more general segments as well.
Optimove also lists Adore Me, Deezer, and Foodpanda as some of their past clients.
Tal Kedar is CTO at Optimove. He holds an MS in Computational Linguistics from Tel Aviv University. Previously, Kedar served as a member of the Forbes Technology Council.
Marketing Campaigns
IBM
IBM offers a software called Watson Campaign Automation, which it claims can help companies of all types create more effective campaigns using predictive analytics.
We can infer the machine learning model behind the software was trained on thousands of data points from purchases from previous campaigns. For example, an eCommerce bookstore that wants to find out which demographic will respond best to their emails driving to a landing page for an eBook will need to feed the software historical data on their past marketing emails. Also, the eCommerce store would need to provide how well past emails for a similar offering did with certain customer demographics at multiple points during the year.
The data would then be run through the software’s machine learning algorithm. This would have trained the algorithm to discern which data points correlate to greater conversions and which demographic is likely to convert.
The software would be able to predict the kinds of emails that will lead to the highest conversion rate for certain demographics.
We could not find a demonstration video showing how Watson Campaign Automation works.
IBM claims to have helped Gavl advertise and grow their live-streaming app for real estate auctions. We can infer that Gavl integrated IBM’s software into its existing data streams such its website or a CRM. According to the case study, Gavl increased their mobile app downloads by 212% by the time of publishing.
IBM also lists Georgia Aquarium and ING Direct as some of their past clients for their Watson Campaign Automation product.
SAP
SAP offers a software called SAP Business Objects Predictive Analytics, which it claims can help B2C businesses make sure they are advertising to their customers at the right time. They claim to accomplish this using predictive analytics.
We can infer the machine learning model behind the software was trained on thousands of data points from the client company’s historical transactions and customer demographic information. This could include their shopping habits, gender, and what types of advertisements engage them the most. The data would then be run through the software’s machine learning algorithm. This would have trained the algorithm to discern which data points correlate to which products the customers will demand next.
The software would be able to predict a customer’s interest in the company’s products, when they are most likely to purchase them, and what marketing methods would work best for those customers. This may or may not require the user to upload information about their past marketing initiatives and customer segments into the software beforehand.
Below is a short 2-minute video demonstrating how SAP Business Objects Predictive Analytics automates the analytical process:
SAP claims to have helped Skylark speed up demand prediction. Skylark used SAP’s software in conjunction with its proprietary customer data to create a software that improved ROI and customer satisfaction. According to the case study, Skylark saw an 80% decrease in analysis time since publishing. However, they case study was not clear on what exactly this analysis entailed.
SAP also lists ARI, mBank, and Colgate Palmolive as some of their past clients.
Bjorn Goerke is CTO at SAP. He holds a Master’s degree in Computer Science from the University of Karlsruhe. Goerke has spent the last 18 years of his professional career at SAP.
Header Image Credit: Marketing Donut
