Collage with human hands moving blocks

7 Keys to Success for Using AI in Market Research

Michael Patterson, Radius Insights, Radius Global Market Research 2021/10/michael-patterson-bio.jpg

by Michael Patterson, PhD

Chief Research Officer

Over the past couple of years, we have been testing AI tools and solutions to determine which ones help us enhance the value we provide to our clients. We’ve found some very helpful use cases for AI and Generative AI, particularly in working with large data sets, but, as my colleague Shari Aaron wrote recently, a machine is still just a machine, and there needs to be a duality in market research between AI and human insights. This confirms what we have always known in our practice, human knowledge and experience and the ability to understand needs and underlying emotions are the defining ingredients that lead to successful research outcomes in our work.

Specific applications, such as digital shopping experiences, survey design, survey data analysis (and many more functions) require human oversight from research professionals to ensure accuracy and data integrity from the early stages of the project.”

 

As we continue to explore the possibilities and capabilities for AI, we’re seeing promising opportunities for content generation, modeling, data handling, and automation that will speed our work and enrich outcomes. However, specific applications, such as digital shopping experiences, survey design, survey data analysis (and many more functions) require human oversight from research professionals to ensure accuracy and data integrity from the early stages of the project.

In short, we’re finding ways to optimize our use of AI and developing best practices that will help us develop and manage our work to ensure outputs accurately reflect our clients’ research and activation objectives.

 

AI presents risks that require careful human oversight.

We have come across many instances during our experiments with AI where we’ve had to pause to ensure we weren’t violating data privacy guidelines laws. AI delivers results without citations or references, making it difficult to trace where specific information is coming from, and we need to maintain a healthy cynicism about the results and dig deeper if we identify content that we want to use.

We’re also hyper aware of security issues, particularly when it comes to protecting our clients’ data, research, and confidential brand information. We continue to develop guidelines for data use and issues such as survey integrity and fraud.

Biased algorithms continue to present problems, and we all need to be aware of the ethical implications associated with using AI unchecked. We work with our staff and client teams to help them maintain awareness about these issues, for example, we host workshops to help them hone empathy skills, and caution them against an overreliance on advanced technology.

 

Collage with hands pointing to different sections of a pie chart

We are using AI applications in qualitative and quantitative research to help us efficiently and effectively address core business questions and hypotheses.”

 

How we’re using AI and developing best practices.

We are using AI applications in qualitative and quantitative research to help us efficiently and effectively address core business questions and hypotheses. We’re also harvesting web data using custom scraping code to obtain large volumes of unstructured data which are then analyzed via AI.

Following are some of the use cases and best practices we’ve found most helpful.

 

1. Desk Research

We use tools like ChatGPT or Gemini for activities such as developing attribute lists, barriers, and claims for quantitative surveys along with screeners and discussion guides. We also use it to gain a faster understanding of industry trends, market landscape, key brands, and important themes.

Best Practices

  • Begin AI research using content or data that has been vetted by the research team or subject matter experts. This will provide stronger results, and establish guardrails to ensure the AI outputs are accurate and reliable. Preparing the content or data up-front also helps with the next critical step: ensuring human experts validate the AI-generated insights.

 

2. Respondent Quality

AI tools enhance data quality by identifying and removing poor respondents in real-time.

Best Practices

  • Implement ongoing monitoring to detect problematic respondents.
  • Regularly update AI algorithms to adapt to new data anomalies.

Outcomes and Opportunities

  • Faster data cleaning to eliminate suspicious completes, reduce misspellings and noise and dedupe data.
  • Increase overall data quality by identifying false answers, survey bots, etc.

 

3. Intelligent Probing

AI provides natural language follow-up questions to open-ended responses.

Best Practices

  • Design AI follow-up questions to mimic natural conversation.
  • Cross-verify AI-generated responses to ensure relevance, clarity and meaningfulness.

Outcomes and Opportunities

  • Reduce nonsensical responses for more reliable results.
  • Gain deeper insights from respondents through intelligent follow-up questions.
  • Deeper engagement with respondents and achieving a higher response rate for surveys.

 

4. Qual/Quant Integration

Combining qualitative questions with quantitative surveys greatly enriches insights.

Best Practices

  • Balance the number of qualitative and quantitative questions to avoid respondent fatigue.
  • Use AI to intelligently probe qualitative questions to gain deeper insights.
  • Ask relevant qualitative questions based on specific responses to quantitative questions.

Outcomes and Opportunities

  • More useful insights and greater value for clients.
  • Quick mini-qual sessions.
  • Enhanced respondent engagement and greater willingness to provide insightful responses.

 

5. Analysis of Unstructured Text

AI tools analyze unstructured qualitative data including boards, IDIs, groups, and survey data), providing comprehensive summaries.

Best Practices

  • Use diverse AI models to analyze various data types.
  • Regularly update analysis tools for evolving data trends.

Outcomes and Opportunities

  • Comprehensive summaries and high-level topic identification.
  • Video analysis with facial/emotional features.
  • Sentiment analysis and word clouds.

 

6. AI-Powered Netnography

Harvesting and structuring comments from websites, discussion boards, social media and other electronic data sources provides valuable brand insights.

Best Practices

  • Ensure ethical data scraping to protect consumer privacy.
  • Oversee and validate AI-driven insights with human analysis.

Outcomes and Opportunities

  • Define core business questions and hypotheses.
  • Identify data sources such as consumer voice, branded content, or expert opinions.
  • Larger volumes of data from custom scraping code.
  • Identification of themes, dimensions, and relevant verbatims.
  • More thorough reporting by mixing quantitative and qualitative insights for strategic recommendations. 

 

7. Synthetic Respondents/Digital Personas

Synthetic respondents can supplement studies under specific circumstances.

Best Practices

  • Use synthetic respondents for exploratory phases.
  • Validate synthetic data with vigor and human oversight.

Outcomes and Opportunities

  • Validate qualitative interviews.
  • Generate hypothesized segments before quant.
  • Help clients refine concepts before testing.

 

Future potential of AI in marketing research

Looking ahead, AI’s potential applications in marketing research are promising. Using efficient design and adaptive techniques, along with conversational research tools like ChatGPT, will improve our ability to deliver consumer insights our clients can use to activate new growth opportunities.

Despite these advancements, human scrutiny is essential. Fact-checking AI results and making multiple requests to ensure accuracy will remain a standard element of AI-based research. Establishing best practices for AI usage in market research will help teams maintain data integrity and actionability.

AI should complement, not replace, human insights. At Radius, we balance AI with human intuition. Our approach combines Qual and Quant to leverage AI’s strengths while ensuring nuanced, actionable insights for our clients. This balanced methodology keeps our research robust, reliable, and impactful.

 

Are you looking for a research partner to collaborate with on your approach to AI?

Contact Us