Qual, quant or social listening: which one answers your question?

Three of our published case studies started with questions that look similar from a distance and needed three different instruments.

• Amazon wanted to know where a purchase in Spain gets decided: in the shop, or on the phone in the shopper’s hand. We followed twelve shoppers in Madrid with mobile diaries, shop-alongs and interviews.
• Kalise wanted to know whether its lead in ice cream and yogurts in the Canary Islands rests on preference or on habit. We surveyed 500 category consumers online.
• A smartphone brand that had doubled its online conversation in Spain in a year wanted to know whether that growth was real interest or campaign noise. It’s a composite case, built from real social listening work with the names removed. We analysed more than 20,000 public mentions from twelve months of conversation and read the forum and Reddit threads in full.

The first is qualitative research, the second quantitative research and the third social listening. Each one answered its question, and each case page ends with a note on what that one lens couldn’t tell the client. That’s the most useful way to think about qualitative vs quantitative market research, with social listening as the third option: by the question each one answers, and by the blind spot each one leaves.

Three questions, three instruments

Qualitative research answers how and why. It works with a small number of people, chosen on purpose, and goes deep through interviews, observation, diaries and going shopping with people. In Madrid, it showed that many of the shoppers we followed needed to see, touch or ask about a product in a shop before buying, and that many then bought online at the best price while others paid in a shop they trusted. What it couldn’t say is how many shoppers in Spain follow each path. How many interviews a study like this needs is a question of its own, and I’ve written about it in how many interviews are enough.

Quantitative research answers how many and how much. It asks the same questions of enough people to measure who prefers what, how strongly and how that differs between groups. For Kalise, it set a baseline the company can repeat every year, and the reasons behind each score were coded so that every number came with the words people used. One finding: “from here” and “knows my tastes” score high and carry warmth, but on their own they don’t answer the questions about innovation and health that younger consumers are starting to ask. What the survey couldn’t do on its own is explain why passive customers stay passive.

Social listening answers what people say when nobody is asking. It reads public conversation on social platforms, forums and review sites, unprompted and in people’s own words. For the smartphone brand, it showed that the biggest peaks came from campaigns with famous faces on X, and that in forums people talked openly about moving beyond the brands they’d always had. What it couldn’t say is how many people in Spain are seriously considering the brand, or why some of them switched.

Why this is a challenge

Each instrument has a blind spot, and it’s easy to forget when the results look tidy.

Surveys record what people say, including what they say they’ll do. Psychologists call the distance between intention and action the intention-behaviour gap. Intentions are among the better predictors of behaviour we have: Paschal Sheeran’s 2002 review of 422 studies, with 82,107 participants, found they explained on average 28% of the variation in what people actually did, a large effect by social-science standards that still leaves most of the variation unexplained. In the studies of exercise, condom use and cancer screening he looked at, a median 47% of the people who intended to act didn’t.¹ A later meta-analysis by Thomas Webb and Sheeran found that a medium-to-large change in intentions led to only a small-to-medium change in behaviour.² Much of that research is about health and social behaviour rather than shopping, but Vicki Morwitz’s review of purchase intentions reaches a similar conclusion for buying: intentions predict sales, imperfectly. Her review also covers how to forecast sales from intention measures, so surveys stay useful as long as you calibrate them.³ If your decision depends on what people will do, don’t rely on what they say they’ll do alone.

Surveys also thin out faster than people expect when you split them. A sample of 500 is solid for the whole group, but cut it into five segments and each one has about 100 people, so the uncertainty around every figure more than doubles. Before you sign off, ask for the margin of error on the groups you plan to act on, not just on the total, and remember that online panels carry error the margin doesn’t show.

Social listening hears the most active people on the platforms you can see. In the United States, Pew Research Center found in 2019 that the most active 10% of adult Twitter users produced 80% of tweets,⁴ and in 2021 that the most active 25% produced 97%.⁵ In Pew’s 2026 figures, 18% of US adults say they use X and 25% Reddit, against 84% for YouTube.⁶ What you can listen to also depends on what the platforms let you see. Twitter, now X, moved its data access to paid tiers in 2023.⁷ Reddit announced a premium access point to its API the same year for heavy users, and named social listening tools among those who rely on its data.⁸ And Meta closed CrowdTangle, a widely used tool for following public posts on Facebook and Instagram, in August 2024, replacing it with a library open only to researchers at qualified academic and non-profit research institutions, so a brand or an agency can’t use it.⁹¹⁰ In the smartphone case, about 91% of the volume came from X, while the clearest talk about switching brands came from forums and Reddit, a much smaller share of the conversation. Public doesn’t mean free to use, either. In the EU, posts by people who can be identified are still personal data under the GDPR, so a report should leave out names and handles, and paraphrase any quote that could be searched back to its author.¹¹ And the search query decides what counts as the conversation, so ask to see it.

Qualitative research doesn’t give you proportions. Twelve shoppers can show you the paths through a purchase and the reasons behind each one, but not which path is most common across the country. That question needs a different instrument. Interviews are self-report too. Asked weeks later, people tidy a purchase up in their memory, which is why the Amazon study used diaries kept at the time and went into the shop with every shopper.

How to choose

Look at the shape of your question before anyone tells you what the study costs.

• If it starts with “why” or “how”, you need conversations and observation, which means qualitative research.
• If it starts with “how many”, “how much” or “which is bigger”, you need a sample large enough to measure, which means quantitative research.
• If it’s about what people say unprompted, in which words and whether that’s changing, you need social listening.

Five questions matched to three research lenses. Why customers leave needs qualitative research, with social listening adding their unprompted words. How many customers would pay more needs quantitative research built around trade-offs, and a real price test if possible. What people say about a brand when nobody asks needs social listening. Whether people will buy a new product needs a survey checked against what people actually do, because intentions alone predict behaviour imperfectly. Which customers to serve first is a segmentation question: interviews find the groups and a survey sizes them, while social listening adds their unprompted words.

Why do customers leave after the first month? How many customers would pay more for faster delivery? What do people say about our brand when nobody asks? Will people buy our new product? Which customers should we serve first?

Human Stories Qualitative research Answers how and why Blind spot No proportions. Twelve shoppers show you the paths, not which one is most common.

Hard Evidence Quantitative research Answers how many and how much Blind spot It records what people say, and for the future that means intentions. Across 422 studies, intentions explained on average 28% of the variation in behaviour. 28% explained

Heartbeat Social listening Answers what people say when nobody asks Blind spot It hears the most active people. In the US in 2019, the most active 10% of adult Twitter users wrote 80% of tweets.

Verdict A why question. Start with conversations and observation. Social listening adds the words people use when they complain unprompted.

Intentions and behaviour: Paschal Sheeran, European Review of Social Psychology, 2002. Twitter users: Pew Research Center, Sizing Up Twitter Users, 2019 (US adults).

Two checks help when more than one answer seems to fit. If the decision depends on future behaviour (will they buy, switch or pay more?), don’t settle for stated intention alone. Look for evidence of what people actually do: your own sales and customer data, a test with real money such as a pre-order, a pilot or an A/B test, or watching people shop. Diaries and social listening help explain behaviour, but they’re still what people report or post. And ask who each method leaves out: people who don’t answer surveys, people who never post and people nobody thought to interview.

When you need more than one

Researchers have argued for combining methods for a long time. Norman Denzin set out different types of triangulation in 1970, including the use of more than one method on the same question.¹² Todd Jick wrote in 1979 that a distinct tradition in social research sees qualitative and quantitative methods as complementary rather than as rival camps.¹³ Robert Kozinets, who developed netnography, the ethnographic study of online communities, advised combining it with other methods when you want its findings to apply beyond the community you studied.¹⁴

In practice, the three instruments can cover much of each other’s blind spots. The interviews explain the numbers, the survey sizes what the interviews found, and social listening checks both against what people say when nobody’s asking. When they agree, you can act with more confidence, as long as they don’t share the same bias, and when they don’t, the disagreement often holds the most useful finding. That’s what our H3™ method is for: it fuses the three into one reading, including where they disagree, so you don’t get three separate reports.

Choose the instrument by the question, and you’ll stop paying for answers to questions you didn’t ask.