You've sent out the survey. Even so, you've crunched the numbers. The data says 73% of respondents "somewhat agree" with your new feature concept Which is the point..
Great. Now what?
You still don't know why. But you don't know what they're picturing when they read the question. Here's the thing — you don't know what they're leaving out because the multiple-choice options didn't include it. And you definitely don't know how they'd react if they heard someone else disagree with them And that's really what it comes down to..
That's where the focus group comes in.
Not as a replacement for quantitative research. As the thing that explains it Worth keeping that in mind. That's the whole idea..
What Is a Focus Group for Attitude Research
A focus group is a moderated discussion with 5–10 participants selected because they share characteristics relevant to your research question. You're not testing usability. That said, you're not A/B testing copy. You're exploring attitudes — the beliefs, feelings, associations, and mental models people hold about a topic, brand, product, or experience Small thing, real impact. Surprisingly effective..
The magic isn't in the answers people give you directly. It's in what happens between the answers And that's really what it comes down to..
Someone says "I'd never trust a bank without physical branches." The first person pauses. And "Huh. " Another participant leans forward: "Actually, I've used an online-only bank for three years and their customer service is better than my old branch ever was.I didn't know that was an option Practical, not theoretical..
That moment? You don't get that in a survey.
It's Not a Group Interview
This distinction matters. In a group interview, you ask each person the same questions in turn. In a focus group, you create conditions for interaction. The participants talk to each other. They challenge, build, contradict, and reveal. Your job is to set the stage, then get out of the way.
Why Focus Groups Work for Understanding Attitudes (and When They Don't)
Attitudes are social. But we form them in conversation. We test them against other people's reactions. We refine them when someone pushes back. A focus group replicates that process in a controlled setting — which means you get to watch it happen Not complicated — just consistent..
You see:
- How strongly someone holds a view when challenged
- What language people actually use (not what they write in a text box)
- Where consensus forms and where it fractures
- The emotional undercurrents — hesitation, enthusiasm, defensiveness — that never show up in Likert scales
But focus groups have blind spots.
They're terrible for:
- Predicting behavior (people are notoriously bad at forecasting their own actions)
- Measuring prevalence (you have 8 people, not 800)
- Sensitive topics people won't discuss in a group (finances, health, family conflict)
- Anything requiring deep individual reflection — the group thinks for them
Know the tool. Use it for what it's good at And that's really what it comes down to..
How to Design a Focus Group Study for Attitude Research
Most failed focus groups were doomed before the first participant walked in. The design phase is where you win or lose.
Defining Your Research Questions
Start here. Not with "what do people think about X?" That's a topic, not a question That's the whole idea..
Better: "What barriers prevent first-time buyers from trusting our brand?Which means "
Better: "How do long-term users describe the value they get — and where does that language differ from our marketing? "
Better: "What assumptions do non-users hold about our category that we could address?
Write 3–5 core questions. Everything else — recruitment, discussion guide, analysis plan — flows from these The details matter here. Took long enough..
Recruiting the Right Participants
This is where most projects go sideways. Even so, you don't want "representative users. " You want relevant participants Worth keeping that in mind. Simple as that..
If you're studying why people churn after 30 days, recruit people who churned at day 28–35. Not "current customers." Not "people who signed up once." The specific slice matters.
Screen for:
- Behavioral criteria (did the thing, stopped doing the thing, never tried the thing)
- Articulateness (can they express themselves? a 15-minute phone screen tells you more than any demographic quota)
- Diversity of perspective within your criteria — you want disagreement, not an echo chamber
Offer an incentive that respects their time. markets. $75–150 for 90 minutes is standard in most U.Don't cheap out. S. The quality of your data depends on who shows up It's one of those things that adds up..
Crafting a Discussion Guide That Actually Works
Your guide is not a questionnaire. It's a flexible framework.
Structure it like a funnel:
- " Builds comfort. In practice, no mention of your brand/product yet. Establishes norms. "What comes to mind when you think about [category]?Consider this: let them talk to each other. Broad exploration (20–30 min) — Open-ended prompts about the topic area. Because of that, Wrap-up (5–10 min) — "If you could change one thing about [topic], what would it be? 4. Think about it: " "What frustrates you about [current solutions]? Get reactions. "
- Probe. Targeted discussion (30–40 min) — Now introduce your stimulus. Warm-up (5–10 min) — Easy, low-stakes questions. But "Tell me about the last time you [relevant behavior]. Still, 2. Could be a concept statement, a prototype, a competitor's ad, a user journey map. " "What didn't we talk about that you expected to?
Write probes, not just questions. Now, " "What do others think? Because of that, "Say more about that. " "You mentioned X earlier — how does that connect to what Y just said?
And leave room to follow the energy. The best insights often live in the tangents.
The Moderator's Role: More Than Just Asking Questions
A good moderator is part therapist, part traffic controller, part anthropologist.
They:
- Create psychological safety in the first 3 minutes
- Notice who's quiet and invite them in without putting them on the spot
- Manage dominant participants ("Thanks, Alex — I want to hear from someone who hasn't spoken yet")
- Detect performative answers (people saying what they think *should
This is the bit that actually matters in practice.
said they should) and genuine answers (people revealing what they actually believe) Most people skip this — try not to..
They also know when to go silent. A pause after a powerful statement is often where the real gold surfaces — the person who was holding back, the unexpected admission, the half-formed thought that suddenly crystallizes.
From Conversations to Clarity: Analysis
The hardest part of qualitative research isn't running sessions — it's making sense of what you collected.
Start within 24 hours while everything is fresh. You'll forget 40% of what you heard within a week, and most of it within a month.
Step 1: Transcribe and organize. Every session gets a full transcript (or close to it — even rough notes beat nothing). Tag each participant with a code so you can trace quotes back to their profile.
Step 2: Open coding. Go through the transcript line by line. Highlight anything that surprises you, confirms a hypothesis, or contradicts what you expected. Assign short labels — "fear of commitment," "price as proxy for quality," "distrust of onboarding." Don't worry about categories yet. Just capture the raw material.
Step 3: Thematic grouping. Look for patterns across participants. If three people who churned at day 30 all independently mentioned feeling "trapped" by the setup process, that's a theme. If nobody mentions price despite your team's obsession with it, that's also a theme — sometimes the absence is the signal Simple, but easy to overlook..
Step 4: Synthesize into insights. An insight is not a quote. It's a so what.
- ❌ "People feel overwhelmed by the onboarding flow." (Observation)
- ✅ "New users don't experience onboarding as guidance — they experience it as a test they're failing, and they quit before they've given the product a real chance." (Insight)
The difference is the second one points to a decision — redesign the flow as support, not assessment. The first one just says "make it simpler," which everyone already assumed.
Turning Insights Into Action
Research that doesn't change anything is just an expensive way to validate what you already believed.
Bring findings to stakeholders with a clear ask:
- What decision does this inform?
- What should we start doing, stop doing, or keep doing?
- **What's the risk of ignoring this?
If you can't answer those three questions, your research wasn't actionable — it was just interesting.
The Quiet Power of Qualitative Work
Numbers tell you what is happening. Qualitative research tells you why — and more importantly, it surfaces the things you didn't think to measure in the first place.
It's slower, messier, and harder to present in a slide deck than a dashboard. But when you're building something that people actually need — something that changes behavior — the depth that only comes from real human conversation is irreplaceable.
The projects that fail aren't the ones with bad data. They're the ones where nobody asked the right questions, recruited the right people, or had the courage to act on what they found That alone is useful..
Start with the questions. Now, find the people who hold the answers. Listen like it matters — because it does.
