A focus group validation SOP gathers 5 to 8 customers in one conversation to react to a problem, a few concepts and the words you use to describe them, and a full round of three or four groups takes about two weeks. Groups are fast and they show how people talk each other into or out of a purchase. They also produce false consensus when one loud voice takes over. This SOP keeps the speed and blocks the groupthink with one simple rule: everyone writes before anyone talks.

What is the focus group validation SOP and when should you use it?

A focus group is a moderated discussion of 60 to 90 minutes with a small group from one customer profile. Richard Krueger and Mary Anne Casey's Focus Groups: A Practical Guide for Applied Research recommends running three or four groups with each type of participant, because a single group tells you about that room only.

Use this SOP:

When the purchase is social, such as family travel, kids' products, wedding services or team tools chosen by a group.
When you have two or three concepts and want to hear people compare them in their own words.
When you need customer vocabulary fast, before writing a landing page or ads.

Skip focus groups for sensitive topics and for measuring individual behaviour. People do not admit money worries or embarrassing habits in front of strangers. Use one-on-one customer interviews for that. If customers will build the solution with you, the customer co-creation SOP is the better format.

What do you need before you start?

One customer profile with screening questions, so the room is homogeneous on the things that matter.
Two or three concept cards: one paragraph and one image each, with a price if you want price reactions.
A moderator and a note-taker. You can moderate yourself if you stay neutral, but the note-taker must be someone else.
A recruiting source such as User Interviews, Respondent, your network or a community, and an incentive budget.
A private scoring form in Google Forms that participants fill in on their phones.

How do you run it, step by step?

1. Write the three questions the groups must answer

Limit yourself to three: "Do they recognise the problem?", "Which concept do they prefer and why?" and "What words do they use?". Everything in the discussion guide serves one of the three.

2. Recruit 7 to 10 per group to seat 5 to 8

Screen for your profile and exclude people who work in market research, advertising or your industry. Expect a few no-shows, so over-recruit by 2. Never mix customer types in one room: managers and the people they manage give different answers when sitting together.

3. Open with the problem, without your product

Spend the first 20 minutes on their current experience: "Tell us about the last time you planned [task]." Count how many people raise your problem without being prompted. This is your cleanest signal, because no concept has been shown yet.

4. Collect private scores before discussion

Show each concept card. Give everyone two minutes to score it from 1 to 5 on the form and write one sentence about it. Only then open the discussion. Private scores show what people thought before the group pulled them toward a shared view.

5. Moderate for disagreement

Ask "Who sees this differently?" after every confident statement. Call on quiet people by name. When one person dominates, thank them and turn to the room: "Let's hear from someone who hasn't spoken yet." The Asch conformity experiments showed how strongly a group pulls individuals toward the majority answer, and your job is to pull them back.

6. Capture the score shift

At the end, ask for a second private score on the leading concept. Compare it with the first. A big jump after discussion means the room sold itself on the idea, and you treat the first score as the real one.

7. Compare across groups

After three or four sessions, put private scores, unprompted problem mentions and favourite concepts side by side. A concept that wins in one group and loses in two others has not won.

Template you can copy

FOCUS GROUP DISCUSSION GUIDE (90 min, 5-8 people)

Welcome (5 min)
"There are no right answers. We want to hear different views.
We record for our notes only. I didn't design these ideas, so be honest."

Warm-up (10 min)
1. Introduce yourself and the last time you [task area].

Current experience (20 min)  [no product shown]
2. Walk us through the last time you [task]. What was hardest?
3. What have you tried to make it easier? How did that go?
   Note-taker: count unprompted mentions of [problem].

Concepts (35 min)  [repeat per concept]
4. Show card. Silent private score 1-5 + one sentence (2 min).
5. What stands out? What is unclear? Who would use this?
6. Who sees it differently?

Comparison (10 min)
7. If you could only keep one, which and why?
8. Second private score for the leading concept.

Close (10 min)
9. What would you call this, in your own words?
10. Can we contact you for a follow-up? (form)

SCORING SHEET
Group # | Participant | Unprompted problem Y/N | Concept A score 1 |
Concept B score 1 | Concept C score 1 | Favourite | Final score | Best quote

Worked example: a shared-chores app for families

This is an illustrative case with invented numbers. A founder tests three concepts for parents with children aged 8 to 14: a chore chart with rewards, a family calendar with chores built in, and a pocket-money app tied to chores. She runs four online groups of 6 parents each.

In the opening discussion, 15 of 24 parents raise chore arguments without prompting. Private first scores: the pocket-money concept averages 4.1, with 17 of 24 (71%) giving it a 4 or 5. The calendar concept averages 3.0 and the chore chart 2.6. In group three, a confident father talks the room into loving the calendar, and its final score jumps by 1.2 points there. In the other three groups it stays flat, so she discounts the jump.

The pocket-money concept wins in all four rooms. Her next step is a concept test with 200 parents to see if the preference holds at scale.

What should you measure, and when do you stop?

Metric	Pass signal	Warning signal
Private concept score	60%+ give the lead concept 4 or 5 before discussion	Under 40%
Score shift after discussion	Under 0.5 points average shift	Over 1 point: the room sold itself
Unprompted problem mentions	50%+ raise the problem before any concept is shown	Nobody mentions it until you do
Concept preference split	Same favourite in 3 of 4 groups	A different winner in each group

Stop after three groups if they agree, and run a fourth if they split. If four groups still disagree, your profile mixes two segments. Split them and test again.

Which mistakes should you avoid?

Letting the founder pitch. If you present the concepts with passion, people score your enthusiasm. Read the cards in a flat voice.
Skipping private scores. Open discussion without them gives you the opinion of the loudest person.
Treating "I'd buy that" as proof. Group approval is cheap. Use it to choose a concept, then test buying behaviour with a pre-order campaign or a landing page.
Running one group. One room is one conversation. Plan for at least three.
Mixing customer types. Bosses and staff, or buyers and users, censor each other.

How can AI speed this up?

Record the sessions, transcribe them with Otter.ai or a similar tool and ask an AI assistant to list every reason people gave for liking or rejecting each concept, with speaker labels and counts. Ask it to flag moments where several people changed their view after one person spoke. Check those moments against the recording, since transcripts lose tone and AI misattributes speakers. An AI tool also helps write neutral concept cards: the free product description generator gives you a plain first draft to strip of hype.

Where does this SOP fit in your validation plan?

Focus groups come after interviews and before quantitative concept tests.

Find the problem with customer interviews.
Compare concepts and collect language in focus groups.
Test the winning concept at scale with concept testing and its wording with marketing message testing.
Once you have users, keep a standing group through a customer advisory board.

Find related methods in the Customer Discovery category. To turn the winning concept into one clear sentence for the next round, try the free elevator pitch generator.

FAQ

How many people should be in a focus group?

Five to eight. Fewer than five gives too little variety, and more than eight leaves quiet people with no airtime.

How many focus groups do I need?

Three or four per customer profile. Stop at three if the results agree, and add a fourth if they split.

Are online focus groups as good as in-person ones?

For most consumer and B2B topics, yes, and they are cheaper and easier to recruit for. Keep online groups at 5 or 6 people and 60 to 75 minutes, because video fatigue sets in sooner.

How much should I pay focus group participants?

Pay what an hour of their time is worth. For general consumers that is a modest gift card; for doctors, lawyers or senior managers, expect to pay several times more.

Focus Group Validation SOP