IBM’s Watson now powers Lucy, a cognitive computing system built specifically for marketers
Via menus or conversational English queries, a marketer can ask questions of any data from any platform, for research, segmentation or media planning.
Suppose that, someday in the future, a data-besieged marketer could use menu commands or plain English text to ask questions about any data.
That future, according to a company called Equals 3, is here, and it’s called Lucy.
Launched recently, she is the first marketing-focus portal built on services provided by IBM’s now-legendary cognitive supercomputer, Watson.
“There is nothing in the marketplace like Lucy,” Equals 3 Managing Partner Scott Litman told me. “She is a user interface for all my marketing systems.”
The name of his company, which was founded last year, derives from the idea that a person plus a computer can equal something bigger than their sum. Lucy’s name is taken from a grandchild of IBM founder Thomas Watson.
Equals 3 doesn’t have an exclusive arrangement with IBM to utilize Watson for marketing, but Litman pointed out that his company has concocted a unique creature among the approximately 100 commercialized solutions based on the cognitive computing system.
Lucy combines 10 computing services from Watson, including Retrieve and Rank, Natural Language Classifier, Personality Insights, a news aggregation service and Tradeoff Analytics. Litman said that these services have not previously worked together as they do in Lucy, plus Equals 3 created its own interface and layered on other unspecified custom software to focus on marketing.
Lucy can ingest and digest any kind of data, structured or unstructured, from any source that is uploaded or accessible through dozens of Watson APIs. Equals 3 said that Lucy “combs more data in a minute than a marketing team could go [through] in a year.” By contrast, Litman said, her big brother, Watson, is designed primarily for unstructured data.
Lucy is already populated with data from such sources as the Advertising Research Foundation, the American Marketing Association (the entire AMA membership will be able to use Lucy to search the organization’s documents), Facebook, Kantar, Ipsos, Media Ocean, Nielsen, Omniture and Twitter.
A typical enterprise might also add data from customer relationship management systems, marketing automation platforms, PowerPoints, PDFs, outside research reports, news feeds, credit card data, social media feeds — anything, without limit. Lucy can employ either personally identifiable or anonymized customer data, although she mostly uses the former.
A major enterprise like Procter & Gamble, Litman pointed out, has hundreds of brands generating mountains of data from websites, loyalty programs, email, ad agency data, Forrester and Gartner reports and much more.
Lucy digests all of this data, learns how to use it, and then applies it to the request. Designed for the Fortune 1000, Lucy can currently be employed in three areas: market research, customer segmentation and media planning.
For researching, a marketer can query her through free-form text, like asking, “Who buys Tesla cars?”
Litman noted that market research is often conducted by a team of people over weeks or months, in order to launch a new product. By contrast, he claimed, a single marketer working with Lucy can obtain the same results in a day. Here’s a research screen:
For customer segmentation, Lucy can deliver what Equals 3 calls “finely drawn portraits” of ideal customers, which she generates from your data. She defines the attributes and finds matching customers in the target group, filtered by geographic location, age, income or other parameters. She will also offer suggestions on the type of messages that will work best for the specific segment.
For media planning, Lucy creates the ideal media mix model, including embeddable charts and graphics, with specific plans for mobile, web, cable and broadcast TV, print and any other traditional or digital channel. A planning screen:
Current customers for Lucy include the WPP agencies and Havas Media. Litman said that one early, unnamed beta tester was able to reduce external costs — such as agencies and consultants — by 75 percent because of Lucy. He added that Havas has “seen a significant reduction in vendor cost and experienced faster campaign deployment.”
Getting Lucy up to speed is different from traditional systems, as it involves training. It takes a couple of weeks to onboard a business’s marketing and other data, and then she has to be trained to understand which answers are good.
After a month where users indicate whether her responses are useful or not, he said, her answers become “pretty good,” in the 20- to 30-percent confidence range. If a “decent number of users” — say, five to ten — train her over another two months, the confidence level increases to 80 to 90 percent.
Her specialty, he noted, is pulling out just the marketing-related answer from gigaloads of data, such as taking a plain English question and coming back with the right piece of info. Equals 3 is now working on updating Lucy so that she can provide insights, in addition to that needle in the haystack.
Opinions expressed in this article are those of the guest author and not necessarily MarTech Today. Staff authors are listed here.