The Tao Of Unilever, Part 2: Amazing AI Applications

Last week’s column focused on how Unilever exemplifies the power of closely integrating marketing into a CPG company’s overall mission and strategy.

This time, a look at the groundbreaking engine that’s increasingly driving not only the company’s marketing, but new-product development, talent recruitment and training, and business operations and logistics: a “cloud-based data lake strategy” that enables continual, real-time artificial intelligence-based analysis and usable customer insights (CI).  

Advanced digital capabilities are “absolutely vital” to addressing the challenges and seizing the opportunities inherent in today’s  “hyper-fragmented” world, where consumers “want products that are tailored to them, and they often want these right now,” Unilever CFO Graeme Pitkethly summed up at this year’s Consumer Analyst Group of New York (CAGNY) Conference.

“As expectations, platforms and channels change, we see a huge opportunity to better understand and connect with our consumers and to digitize across our entire value chain” by leveraging the massive amounts of data being generated by consumers, retail customers and other data sources, he said.

“While consumers are generating a lot of the data through fragmented shopping journeys, we are generating far more data internally, which we need to harness to deliver unique insights and speed up our ways of working,” Pitkethly explained.

The company’s data lake ingests data from more than 150 internal systems and external sources, comprising some 5,000 individual data objects — and grows by about 1.5 terabytes per day.

Direct-to-consumer transactions, consumer device IDs, social listening and influencer data; retailer POS, shelf-price and promotional data; operations data from producers, daily sales, and even ice cream vending machines; and macro factors including market size, GDP, demographics and weather, are just a few of the variables being captured.

This information is fed, through the cloud, to 27 (and counting) “People Data Centers” around the world staffed with data analysts and scientists who generate and share advanced analytics and insights tailored to internal users by market and function.

As noted last week, the company’s new marketing model is based on the larger corporate mission of developing brands with purpose; creating compelling, tailored content; and delivering that content through digital channels and data-informed targeting.

“We are investing in people capability and digital marketing because we recognize that marketing is no longer just creating big campaigns,” Pitkethly stressed. “It’s now about having the right people with the right capabilities around the globe and around the table to manage campaigns in real time. Gone are the days of just having brand managers assisted by advertising agencies. We know that people are working in real time around one table making dynamic changes as companies develop.”

The core goal: “By getting the right insights from our consumers, we can be sure that when we talk to them, they will listen and they will respond to our message.”

A Digital Marketing Army

The teams in the data hub include staff dedicated to creating distinct audiences by segmenting data by geography, interests and demographics; tailored content creation specialists; data governance specialists (Unilever says it applies the EU’s stringent General Data Protection Regulation standards throughout the world); and measurement and optimization specialists focused on maximizing ROI by driving views, clicks and buys.

And that’s only a portion of the CI army. Unilever now has nearly 1,300 digital marketers, 85% of whom have completed its “digital world” training. The force is being built both by “upskilling” existing staff in digital marketing and recruiting new talent.

The company has also accelerated innovation by partnering with, acquiring and/or helping incubate cutting-edge digital startups around the globe. One example: the AI-driven Popular Chips marketing platform, which has reportedly enabled Unilever to streamline the process of finding and vetting segment-appropriate influencers.

Reading Humans By Interpreting Music, Movies

One of the more mind-blowing aspects of this corporate data/analytics juggernaut is laid out in “AI for Marketing and Product Innovation,” a book co-authored by Unilever’s global insights chief, Stan Sthanunathan.

While it’s not hard to understand how machines could beat humans at sorting through reams of social posts to find key words and themes, the book explains that AI and machine learning are also better than humans at spotting and analyzing metaphors reflecting unconscious, shared beliefs and attitudes about life that are embedded in “unstructured” data – including music and movies.

"We started thinking that AI would be used for structured data and the internal mindset was ‘Let’s use this for predictive modeling’," Sthanunathan recently told Campaign Live. “But the big takeaway was: How can it be used for unstructured data? That’s where the richness in data and the big ‘a ha’ moment came for us."

Unilever developed a tool that goes "straight to the source of influence: songs, TV and film," he said. “The movies we watch, the books we read, all influence our thinking. If you can analyze every single piece of dialogue in a movie or song, we may be able to unearth something meaningful. Consuming all that content is not a mind-numbing experience for a computer. We started mining unstructured data to get metaphors. We can use those same pieces of information to identify emerging trends to see."

Those insights have been used to inform branding strategy and management companywide and generate insights that have resulted in home runs such as launching breakfast-inspired flavors of Ben & Jerry’s ice cream.  

Insights from unstructured or structured data have also inspired a Sunsilk partnership with a Latin American influencer (a line of products for curly hair); and an Axe deal with DJ Martin Garrix on a YouTube video that’s generated more than 40 million views to date.

“Both of these not only resonated with Unilever’s target audience for the specific product and region, but also created a deep impact in their minds about how they perceive the brand,” observed TechWireAsia, adding that the company has declared a goal of creating a billion one-to-one consumer relationships through the insights yielded from its data hub.

Taking Recruitment to the Next-Next Level

While it’s not possible to touch on all of the functions being transformed by data insights and applications within Unilever, the company itself has stated that it has already automated more than 700 processes — resulting in substantial and growing time and cost efficiencies.

But on the subject of mind-blowing, it’s tough to resist at least mentioning how AI is being used in recruitment and training of all kinds of executives, including marketers.

The company worked with AI firm Pymetrics to build an online platform that can be used from the candidate's own home, according to Bernard Marr, writing in Forbes. Candidates are asked to play games to assess their aptitude, logic, reasoning and appetite for risk. "Machine learning algorithms then assess their suitability for whatever role they have applied for, by matching their profiles against those of previously successful employees." The second phase asks the candidate to answer questions, on video, for about 30 minutes. An algorithm assesses language and body language to determine the fit with Unilever. 

The overall process has cut about 70,000 hours from interviewing and analyzing candidates for Unilever, which recruits more than 30,000 people per year and processes about 1.8 million job applications. 

"We look for people with a sense of purpose — systemic thinking, resilience, business acumen,” Unilever HR chief Leena Nair told Marr. “Based on that profile, the games and the video interview are all programmed to look for cues in their behavior that will help us understand who will fit in at Unilever. “Every screenshot gives us many data points about the person, so we work with a number of partners and use a lot of proprietary technology with those partners, and then we select 3,500 or so people to go through to our discovery center.”

After spending a day with real leaders and recruiters, Unilever ultimately picks about 800 people who will be offered a job, Marr reports.

And that’s not the end: Those hired are trained and guided in the corporate culture in part by a natural language-processing bot “designed to understand what employees need to know and fetch information for them when it is asked.”

Amazing — and impressive — for sure. But at risk of the dreaded Luddite accusation, also a bit on the scary side, for those of us who've become concerned about the long-term consequences of continually reducing the need for human interaction.

Next story loading loading..