Automating insight generation is not only saving market research professionals time and money, it offers the opportunity to find patterns in data at a scale never before possible
According to Josh Sutton, chief executive at Agorai: “Artificial intelligence (AI) solutions are producing insights in seconds that used to take teams of people days or even weeks to produce. Early adopters are seeing financial benefits already.”
The main metrics for success are a reduction in insight time, followed by decreases in labour demands and overall expense. Mr Sutton adds: “McKinsey recently produced a report which states that over the next decade, early adopters of AI will dramatically outperform followers and laggards.”
2019 is the year that AI was tipped to take over from automation in market research, but seven months in, has AI lived up to its potential?
“We are now at a point in time that is reminiscent of the mid-1990s, when the early winners of the internet were those who identified opportunities and experimented to address business problems,” says Chris Duffey, strategic development manager at Adobe and author of Superhuman Innovation.
Market research industry
“AI is making people better at their jobs in three ways: automation of tasks that can help a person perform their job more effectively; the ability to generate insights from large amounts of data; and ability to enable interactions with technology via natural-language conversations,” says Mr Sutton.
Automation, insight generation and natural-language processing enable businesses to survey the market continually, rather than dedicating specific time, labour and money to the process.
AI solutions are producing insights in seconds that used to take teams of people days or even weeks to produce. Early adopters are seeing financial benefits already
An advantage of using AI in marketing research is the insight time involved in creating an overview of consumer needs is cut down enormously. Mr Sutton adds: “When AI is used to understand the comprehensive picture painted by all the data available via social media, people’s movements, behaviours that can be observed and the corresponding actions taken, we are seeing firms develop insights which give them a better picture than they have ever been able to generate.”
Analysing data in real time
Using AI also means data can be digested in real time. David Benigson, chief executive and co-founder of Signal AI, says this has a powerful impact. “The Washington Post alone publishes over 1,000 articles or news stories a day. If a human being read nothing else, they would still struggle to get through just that one newspaper. Signal’s AI reads 2.7 million sources a day and over five million documents are analysed in less than two seconds. Our clients find that two-hour tasks now take ten minutes,” he says.
Those are some very appealing numbers for companies looking to decrease insight time. But AI learns from human beings, so are there possible pitfalls for a system working at a speed so far beyond human intelligence?
“Superintelligence is a notion that is often represented in movies as an anthropomorphised robotic AI system with an exponential increase of intelligence over humans,” says Mr Duffey. “However, unlike science fiction, when we talk about modern AI, we’re in fact talking about narrow AI, which means AI that’s designed to perform a task.”
It seems that while AI in marketing research is currently able to unearth, read and analyse data far more quickly than human beings, it currently lacks the creativity to deal with unforeseen obstacles in its path and change course accordingly.
“The ability to extract real insights from written comments is a great example of how AI in marketing research can improve insight time,” says Mr Sutton. ”Firms that used to rely on a staff of analysts to review consumer feedback can now see key themes people care about in real time. Over 80 per cent of the data produced every day is unstructured – written feedback, photos and so on – and this data is not being used by AI solutions to provide real insights in a way that wasn’t possible without AI.”
Harnessing this unused data to create a continuous big picture of customer insights can alert marketing teams to patterns and trends sooner than traditional methods, helping firms stay on top of consumer needs.
AI is potentially a very useful tool, but it still requires extensive guidance to thrive. This trade-off is already worth it, says Mr Benigson. “At Signal, for example, clients can train the AI to their bespoke specifications in just a few hours, in effect digitising their knowledge for the long term. If you’re saving an hour a day, you’re already in credit after just a week,” he says. That’s a vast improvement on insight time using traditional methods.
The future of AI for market research
When looking to the future of artificial intelligence in marketing research, it’s important to understand how AI and human beings can work in harmony, each augmenting the other’s strengths.
“AI excels at storing and remembering huge amounts of data and making very complex calculations based on those data sets,” says Mr Duffey. “People are extraordinarily skilled at social interactions and complex tasks, critical thinking and creativity. Keep this
What could this combination of human and AI skills achieve? “Trained AI can also go beyond what a human can do by finding the connections in the data at a scale that would be practically impossible for a human,” adds Mr Benigson. “In effect, it can help us understand not just the known unknowns, but the unknown unknowns, the things we didn’t know we didn’t know.”
With the promise of such boundless possibility, the advantages of using AI in marketing research to reduce insight time are clear and very exciting.