Using AI to gather customer insights

Gathering and interpreting customer insights using artificial intelligence offers the market research industry a strategic change in direction

The market research and insight industry will soon undergo a public test of its potency. In two weeks it will predict what it believes will be the outcome of the general election.

Already under fire for failing to predict the outcomes of both the UK referendum and the US presidential election, you’d forgive top polling organisations for being nervous.

The general election, however, is not the biggest challenge facing the market research industry. Insiders say market research agencies are slowly waking up to a much greater threat.

Evolving technology is enabling various other organisations – adtech companies and cloud software vendors, with the likes of Amazon and Google among them – to gather data and interpret customer insights much faster than traditional research methodologies.

Such companies are also among the first adopters of artificial intelligence (AI) to automate both the gathering and interpretation of better customer insights.

Google’s all-powerful search engine, Amazon’s exploitation of the most potent data set in retail, Facebook’s algorithms that determine what “news” should appear in individuals’ timelines: big technology companies now represent a new high watermark in the gathering of information on how we live, what we care about, how we communicate with one another and how we buy.

That watermark continues to get higher with every new innovation. The internet of things – smart fridges, smoke alarms, entertainment systems, heating systems and even windows all with the ability to talk to one another – came one step closer with the emergence of voice-activated assistants. Apple’s Siri, Amazon’s Echo Alexa, Google’s Home and Microsoft’s Cortana are changing the way we choose to search for information and research the internet.

As we use such technologies we, knowingly or not, share more valuable data with the technology owners and companies using the assistants to provide a service. At the same time the companies that own and partner with this technology gather evermore intelligence about us.

Examples of AI in practice

Virgin Holidays recently launched the prototype of a new app for Amazon’s Alexa that allows families to sit together around their speaker at home and ask the app, named Alex, for help finding the perfect Virgin Holidays getaway. When asked questions, the app will understand the question and search three million different data points to generate the best result.

While providing a service, Virgin Holidays is testing whether families are comfortable searching for their holiday by speaking to a technology. Crucially it is also using both the questions asked and the AI prototype to inform where the company should take technology next

Similarly, much smaller, fast-growth technology companies are unearthing far deeper insights at speeds previously unavailable to marketers.

User experience analytics and optimisation business ContentSquare recently added an AI layer in the form of a bot named Arti. The company captures website user engagement data for brands such as L’Occitane to help them ensure every “block” of content on their sites is as potent, engaging and profitable as possible.

With the capture of every move of the mouse, every click and every purchase made by individual website visitors, ContentSquare’s clients can benefit from a mind-boggling volume of data. It would take humans several days to crunch these numbers to uncover the killer insights that could drive immediate competitive advantage.

That’s where Arti comes in; the bot can read and process the data and trigger alerts in real-time that tell e-commerce and market teams exactly what they should do next to benefit from market trends and changes.

There is growing recognition that technology needs to be at the centre of unearthing insight

Similarly, New York-based software company Dynamic Yield, which helps the likes of Under Armour, Sephora and Urban Outfitters personalise their customer experiences and services, is using an advanced machine-learning engine to build actionable customer segments. This enables marketers to talk to each of their customers in a tailored and relevant way and, they hope, form stronger relationships that generate more revenue over time.

Founder and chief technology officer Omri Mendellevich says Dynamic Yield is working towards being the leader in AI-based personalisation. For now the technology is used to empower the marketer’s decision-making process. “We understand how specific segments within your customer base interact with specific variations of the web experience,” says Mr Mendellevich.

“You can try different designs of a purchase stage on your website. Since we have a lot of audiences in the system, we can see through our machine-learning how each segment interacts and engages with the different content. We learn with every user response. This means you actually get traffic directed to the right segment with the right variation in real time.”

Behavioural data, information that reflects real actions on the part of your customers, trumps consumer panels, where customers tell you how they think they might behave in future, every time. Right now, few research agencies possess that level of sophistication or strategic understanding.

Hype tech or game changer

Industry implications

Shamus Rae, head of innovation and investments at KPMG, oversaw the acquisition of research agency Nunwood in 2015, a strategic response to the disruption the market faced even then. The combination of the two has reaped dividends in driving a service that has kept up with demand. Mr Rae though is pragmatic about the broader industry.

“Research is facing competition from new players that are digital to their core. Most traditional industry players are a long way behind where they should be and risk taking the wrong approach to rectify it,” he says.

“There is growing recognition that technology needs to be at the centre of unearthing insight, but the response we see is largely tactical. They buy tools and bots in an effort to ‘seem’ digital. If you asked them for their strategic five-year vision, whether they understand the value of data and how to exploit it, I think they’d struggle.”

KPMG Nunwood is not standing still. It plans to spend $100 million a year for the next four years building a series of AI-led systems. “We’re going all in because we know our clients want this level of sophistication,” says Mr Rae.

Pete Markey, who recently left his role as marketing boss at insurer Aviva to become marketing director at high street bank TSB, sees the shift in power away from research agencies as a strategic problem for the industry to solve.

“Even five years ago your market research function would be the oracle of customer knowledge,” he says. “You would commission research, put some people in a room behind some glass, feed them sandwiches and listen to what they had to say.

“There’s still a value there, but the problem agencies have is that so many functions within an organisation now do research themselves. Look at the customer experience and marketing teams, and what they can learn every day with the use of programmatic.

“You can see and react instantly to what customers are responding to, what they like and what they don’t. The retailers are brilliant at this – seeing engages with certain products on their site and learning where they can direct customers to next.

“In the current market, if you want to get under the skin of what people think, you need to go deeper than the traditional research group. Tools like AI help you optimise and improve performance continuously. As a marketer, that’s the dream gig.”

Also found in smart machines AI