The Question Robots Can’t Answer (No Matter How Hard They Try)
75% of organizations are focused on using data to create better customer experiences, but only 42% of these organizations actually use the data to implement change in their customer experiences (Forrester Consulting, 2022).
Great! AI programs give you access to all the data you could want. Now what? This is the question many organizations are asking themselves. What good are billions of data points, if they don’t lead to insights an organization can use to improve CX? The key to success in today’s world is finding your needles of actionable insights in the haystack of data. Finding the needles is quickly becoming a necessary part of any good market researcher’s job. Market research and AI can be a powerful team, if the researcher knows how to work with AI.
AI IN MARKET RESEARCH
It’s fast, but it’s still not human. AI can pull billions of data points quickly, organize them for you, and even help spot trends and make predictions – all in less time than it would take a human. However, while AI can spot trends and make predictions, the researcher still needs to make the connection to the real world. Why are data points grouping together? How can a piece of information be used to improve CX? What is the plan to implement this change in our CX experience? Making the connection between the data and the actionable insight is still something AI cannot do.
It’s neutral. Top researchers allow the data to tell them the story, rather than try to make the data tell the story the researchers want. It is never the intention of the researcher to bias the results, but sometimes, despite all our best efforts, bias can sneak into the data, particularly when you have multiple people interpreting the same dataset.
AI has low bias. AI programs do not care about your feelings – or even your bottom line. AI will tell you what the data is saying with low bias, reducing human error.
AI has a deep understanding of the nuance of language. AI programs can process language with much greater detail than humans. AI can spot trends and make predictions based on qualitative responses. These programs can interpret emotion, word choice, tone, etc. These detailed analyses of qualitative responses allow for a researcher to better explain WHY a customer feels/thinks/acts the way they do.
AI can analyze data in ways that are ridiculously hard to do by hand. A major benefit of AI’s ability to interpret qualitative responses is that numbers can be assigned to qualitative data. This allows for traditionally quantitative analyses to be used on qualitative data – t-tests, ANOVAs, correlations, etc. Now we can see if there are statistically significant differences in qualitative data – which prior to AI was difficult to do.
REAL WORLD EXAMPLE
In a recent client project, PATH was creating a profile of a company’s unhappy customers to find out what was going on. These individuals gave the company a 6 or below on a 10-point scale and are easy targets for competitors.
Using AI, we found a difference in the word choices and emotional tone of the writing of individuals who gave a 1 or 2 rating compared to those who gave a 5 or 6 rating. Individuals who gave a 1 or 2 rating were extremely upset and used negative emotional words to describe their experiences. Individuals who gave a 5 or 6 rating were a bit unhappy about a specific aspect of their experiences, but were generally quite positive about the company.
From this insight, PATH determined the best strategic opportunity for the company was to address the exact areas the 5s and 6s wanted to improve. We also created a long-term plan to address the company’s relationships with the 1s and 2s.
Market researchers and AI can successfully work together to deliver an actionable plan to clients. The AI program grouped responses based on emotional tone and word choice. The market researchers took the insights and created a clear, actionable, plan for the client to easily implement.
WE STILL NEED CUSTOMERS IN OUR CUSTOMER EXPERIENCE RESEARCH
AI does not – and likely never will – eliminate the need for actually speaking to real customers. Yes, AI can pull all customer data to answer who, what, when, and where – that’s a lot of information on your customers retrieved quickly. However, missing from the AI data is the all-important why.
From AI you can find out that Julie places parts orders for her reefer truck through your online customer portal about once a year. What you don’t know is though your company also offers repair services, why doesn’t Julie come to you for her repairs? AI can likely find out where Julie does go for her repairs, but you still won’t know why she doesn’t come to you. The only way to get this answer is to ask Julie.
CONCLUSION
To paraphrase Tony Stark, finding those actionable insights is like “finding a needle in the world’s largest haystack…fortunately, I brought a magnet” – in the case of market research, the “magnet” is the researchers, the human touch necessary to move from data to action.