Predominant qualitative data collection methods are high-touch, time-consuming and consequently expensive. But now, AI-enabled chatbots are presenting opportunities to collect qualitative data at scale.
There are two ends to the qualitative spectrum essentially. On the one side, you have your deep exploratory research, where the guide is really open and the moderator must be alert to all sorts of paths that may emerge as the discussion progresses. On the other are more structured discussions where we have a pretty good idea of what our questions are from the outset.
Concept and advertising qualitative generally fall into the latter. The moderator shows some stim, and asks what participants think the main message was, how relevant it was to them, what they think of the message coming from the brand, and how the message might impact their propensity to use the brand. While of course probing for reasons behind their thinking and feeling along the way. Frankly, pretty standard stuff once you have a few years of experience under your belt.
Qualitative at a Crossroads
Clients know this too, and they do a lot of this type of research, so this is the area where a moderator gets most of those tricky requests. You know, the ones prefaced with: “How fast can you do this because we need to optimize this idea to be ready for the quant test next week”, and “We don’t have much of a budget for this because, well, we didn’t budget for it” (you can stop reading here if this has never happened to you).
Qualitative research is by its very nature expensive, and this is directly related to the time involved. Hours on the phone, on a webcam, in a facility, in homes, in a car or on a plane, to get the feedback from consumers that leads to breakthrough insights. And that’s even before analysis.
And what are the alternatives, really? Give a big discount? Do a very limited number of groups or interviews? Post some questions on an online bulletin board?
Traditional qualitative data collection is very high touch and time-consuming, dominated by in-person focus groups and IDIs (GRIT report 2018 Q3-4). Quantitative data collection, on the other hand, is highly automated with online and mobile surveys dominating, resulting in cost and time savings for clients over human-delivered surveys.
But part of the skill of a moderator is in creating rapport with participants so they’ll feel comfortable, confident, and challenged to share more than they would in an open-ended text box. Moderators use focus, engagement and empathy to connect with participants, build trust and encourage lateral thinking, leading to deeper insights. How can automation possibly replicate this?
Learning from Customer Service
Customer service is way ahead of us here. Many companies are already using AI-enabled chatbots and while they have certainly had some early frustrating stumbles, they are starting to figure it out. It is clear that at this time, customer service chatbots are not well suited for complicated issues, but they excel at answering routine questions. Used well, customer service chatbots have been shown to increase customer satisfaction (TELUS International) as a chatbot provides immediate response, any time of day, and never gets tired.
When people think about AI specialists, they generally think about computer programmers, mathematicians, and engineers. But an important aspect to programming, if the goal is to engage with humans, is language skills to make the interactions fluid, and the ability to anticipate how a conversation might go in advance. It’s been argued convincingly that an important component of AI in the future will be the inclusion of the thinking of liberal arts majors (see “The Future Computed,” Microsoft president Brad Smith and EVP of AI and research Harry Shum).
Moderators Have What It Takes
The skills we have as researchers, particularly those on the qualitative side, equip us very well to play a key role in the AI revolution for research. Qualitative researchers take all of a client’s objectives and structure their guide so that the discussion progresses in a logical way to the participant. They are highly trained in not only asking the questions but in keeping the participant focused, engaged, and feeling listened to.
These elements can all be programmed into a research experience that replicates a one on one conversation with a human. AI can’t replicate (yet) the deep, exploratory skill that a qualitative researcher employs when it’s difficult to know what branches the conversation might take.
But we are at the point with technology where AI-enabled chat is a feasible means of qualitative data collection in cases where we essentially want to ask all the participants the same questions. And it is so much more engaging to participants than filling out open-ended text boxes that they are willing to share more data with deeper insight.
Using AI to conduct qualitative interviews is faster, less expensive and viable, and experienced moderators have the skills to make it happen.
Author: Laura Craig, Co-founder, CRIS Research, Owner at Elsient
Article source: Greenbook blog