How many quotes should I code per interview?

Expect 15-30 codeable quotes per 30-minute interview. Not every sentence is worth coding. Focus on statements where the user describes a problem, behavior, need, cost, or value judgment. Skip greetings, small talk, and generic statements like "It would be nice if everything worked better."

What if my themes keep changing as I code more interviews?

That is normal for the first 5-7 interviews. Your code list evolves as you learn the domain. After 7-10 interviews, codes stabilize. If they do not stabilize, your interview questions may be too broad. Tighten your focus area and re-interview a few participants.

How do I avoid confirmation bias during analysis?

Three safeguards. First, code quotes before grouping them. Grouping first encourages you to fit quotes into categories you expect. Second, always log contradicting evidence alongside supporting evidence. Third, have a second person independently code a subset (20%) of quotes and compare your codes. Disagreements reveal blind spots. The [Product Discovery Handbook](/discovery-guide) covers additional debiasing techniques.

When is qualitative analysis enough, and when do I need quantitative validation?

Qualitative analysis is enough for understanding the shape of a problem: who has it, what they feel, what they do about it. It is not enough for measuring the size of a problem across your entire user base. After qualitative analysis, run a targeted survey (using the [Product Analytics Handbook](/analytics-guide) as a guide) to validate that your patterns hold at scale before committing major resources. ---

Customer Interview Analysis Template

What This Template Is For

Running customer interviews is only half the job. The other half is turning hours of conversation into clear, actionable insights. Most teams skip this step. They write a few bullet points after each call, file them in a Google Doc, and forget about them. Three months later, nobody can remember what they learned.

This template provides a systematic process for analyzing customer interview data. It covers quote coding (tagging quotes by theme), affinity mapping (grouping coded quotes into patterns), and a decision matrix that translates qualitative patterns into prioritized product actions.

It works best when you have completed 10-20 interviews and need to extract patterns from the accumulated data. For guidance on conducting the interviews themselves, use the Customer Discovery Interview Template and the Product Discovery Handbook.

The RICE Calculator is useful after analysis when you need to score and rank the opportunities you have identified. For understanding key discovery terms, the product-market fit glossary entry provides essential context.

How to Use This Template

Gather your raw data. Collect all interview transcripts, notes, and recordings in one place. You need the original material, not just summaries.
Code individual quotes (Part 1). Go through each interview and tag every meaningful quote with one or more theme codes. This is the most time-consuming step but also the most valuable.
Build the affinity map (Part 2). Group coded quotes by theme. Look for clusters where 5+ quotes from different users point to the same pain, behavior, or need.
Extract patterns (Part 3). For each cluster, write a pattern statement. Quantify how many users mentioned it, how severe it is, and what evidence supports it.
Run the decision matrix (Part 4). Score each pattern on impact, frequency, and feasibility. This produces a ranked list of opportunities.
Present findings (Part 5). Use the research summary template to share results with stakeholders. Lead with the top 3 patterns, not a data dump.

The Template

Part 1: Quote Coding Sheet

Go through each interview transcript. For every meaningful statement, log it here with theme codes.

How to code quotes:

Read each statement and ask: "What is this person telling me about their problem, behavior, or need?"
Assign one or more theme codes from the code list below
Create new codes as needed. Your code list will grow during the first 3-5 interviews, then stabilize

Starter code list (customize for your domain):

Code	Theme	Description
PAIN	Pain point	User describes frustration, difficulty, or failure
BEHAV	Behavior	User describes what they actually do (not what they wish they did)
NEED	Unmet need	User describes something they want but cannot get from current solutions
WORK	Workaround	User describes a hack or manual process they use to compensate
SWITCH	Switching behavior	User describes trying, adopting, or abandoning a tool
VALUE	Value statement	User describes what matters most to them in a solution
COST	Cost/impact	User quantifies the cost of the problem (time, money, outcomes)
CONTEXT	Context	Background about the user's role, team, or environment

Quote coding log:

Quote ID	Interview	Verbatim Quote	Codes	Notes
Q-001	[Int #1, Name]	"[Exact words]"	PAIN, COST	[Your interpretation or question]
Q-002	[Int #1, Name]	"[Exact words]"	BEHAV, WORK	[Your interpretation or question]
Q-003	[Int #2, Name]	"[Exact words]"	NEED, VALUE	[Your interpretation or question]

Coding rules:

☐ Use exact quotes, not paraphrases. You will need the original words for stakeholder presentations
☐ One quote per row, even if the user said multiple things in one sentence
☐ Apply codes based on what the user said, not what you think they meant
☐ Mark surprising or contradictory quotes with a star for later attention

Part 2: Affinity Map

Group coded quotes into clusters. Each cluster represents a theme that appeared across multiple interviews.

Cluster template:

Cluster name: [Descriptive label for this pattern]

Theme code(s): [PAIN, NEED, etc.]

Quote count: [How many coded quotes belong here]

Unique users: [How many different interviewees contributed quotes to this cluster]

Representative quotes (top 3-5):

"[Quote]" - [User name/ID, role, company type]
"[Quote]" - [User name/ID, role, company type]
"[Quote]" - [User name/ID, role, company type]

Summary statement: [One sentence that captures the core insight from this cluster]

Confidence level:

☐ High: 8+ users, consistent language, strong emotional signals
☐ Medium: 4-7 users, similar themes but varied language
☐ Low: 2-3 users, possibly coincidental

Part 3: Pattern Extraction

For each high-confidence cluster, extract a formal pattern statement.

Pattern template:

Field	Content
Pattern ID	[P-001]
Pattern statement	[One clear sentence: "Users in [segment] experience [problem] when [context], causing [impact]"]
Evidence strength	[High / Medium / Low]
User mentions	[X] of [Y] interviews
Behavioral evidence	[What users DO about this, not just what they SAY]
Emotional intensity	[How strongly users feel: mild annoyance, significant frustration, rage, resignation]
Current workaround	[What users do today to cope]
Cost of the problem	[Time, money, or outcomes lost. Use user-reported numbers.]
Contradicting evidence	[Any quotes or behaviors that challenge this pattern]
Open questions	[What you still do not know and need to investigate]

Pattern registry:

ID	Pattern Statement	Evidence	Mentions	Intensity	Cost
P-001	[Statement]	High/Med/Low	X/Y	High/Med/Low	[$X/time]
P-002	[Statement]	High/Med/Low	X/Y	High/Med/Low	[$X/time]
P-003	[Statement]	High/Med/Low	X/Y	High/Med/Low	[$X/time]

Part 4: Decision Matrix

Score each pattern to create a prioritized list of opportunities.

Scoring criteria (1-5 each):

Dimension	1 (Low)	3 (Medium)	5 (High)
Frequency	2-3 users mention it	5-7 users mention it	10+ users mention it
Intensity	Mild annoyance, low emotional energy	Moderate frustration, active workarounds	Strong emotion, significant time/money cost
Feasibility	Requires new technology or major infrastructure	Moderate effort, 1-2 sprints	Can ship in days, extends existing functionality
Strategic fit	Tangential to product vision	Supports product direction	Core to differentiation and roadmap
Revenue signal	No willingness to pay mentioned	Some users would pay more for this	Users explicitly ask to buy a solution

Decision matrix:

Pattern	Frequency	Intensity	Feasibility	Strategic Fit	Revenue Signal	Total	Rank
P-001	[1-5]	[1-5]	[1-5]	[1-5]	[1-5]	[sum]	[rank]
P-002	[1-5]	[1-5]	[1-5]	[1-5]	[1-5]	[sum]	[rank]
P-003	[1-5]	[1-5]	[1-5]	[1-5]	[1-5]	[sum]	[rank]

Action classification:

Total Score	Action	Timeline
20-25	Build now. High confidence, high impact.	This quarter
14-19	Plan and validate further. Run a prototype test or targeted survey.	Next quarter
8-13	Monitor. Log the insight, revisit after more interviews.	Backlog
5-7	Deprioritize. Low evidence, low impact, or poor strategic fit.	Archive

Part 5: Research Summary (For Stakeholders)

Customer Interview Research Summary

Date: [YYYY-MM-DD]

Researcher: [Name]

Interviews analyzed: [count]

User segment: [Description of who was interviewed]

Methodology: [e.g., "20 semi-structured interviews, 30-45 minutes each, with engineering managers at B2B SaaS companies (50-200 employees). Interviews focused on project planning workflows and tool usage."]

Top 3 findings:

[Pattern statement] ([X] of [Y] users, [evidence strength] confidence)

- Key quote: "[Representative quote]"

- Implication: [What this means for the product]

[Pattern statement] ([X] of [Y] users, [evidence strength] confidence)

- Key quote: "[Representative quote]"

- Implication: [What this means for the product]

[Pattern statement] ([X] of [Y] users, [evidence strength] confidence)

- Key quote: "[Representative quote]"

- Implication: [What this means for the product]

Surprising insights:

[Something the team did not expect]
[A use case or behavior that challenges assumptions]

Recommended next steps:

[Specific action, owner, timeline]
[Specific action, owner, timeline]
[Specific action, owner, timeline]

Open questions for further research:

[Question 1]
[Question 2]

Filled Example: CloudDeploy (CI/CD Platform)

Quote Coding (Sample)

Quote ID	Interview	Quote	Codes
Q-014	Int #3, Amir (DevOps Lead)	"I spend 2 hours every Friday cleaning up failed deploys that nobody claimed."	PAIN, COST
Q-015	Int #3, Amir	"We built a Slack bot that pings the person who pushed the bad commit. Took us a week to set up."	WORK, BEHAV
Q-031	Int #7, Kenji (SRE)	"Rollbacks should be one click. Ours take 45 minutes because we have to manually verify each service."	NEED, COST
Q-044	Int #11, Lisa (VP Eng)	"I would pay 2x our current CI/CD bill for reliable one-click rollbacks."	VALUE, COST

Affinity Map (Cluster: Deploy Failure Ownership)

Cluster name: Nobody claims failed deploys

Quote count: 14 quotes from 9 users

Representative quotes:

"I spend 2 hours every Friday cleaning up failed deploys that nobody claimed." - Amir, DevOps Lead
"When a deploy fails at 5pm, nobody wants to own it. It sits broken until morning." - Carla, Eng Manager
"The hardest part of my job is figuring out who broke the build, not fixing it." - Sam, SRE

Summary: Teams waste 2-5 hours/week tracing failed deployments to the responsible developer because current tools do not auto-assign ownership.

Decision Matrix

Pattern	Freq	Intensity	Feasibility	Strategy	Revenue	Total	Rank
Auto-assign deploy failures	5	5	4	5	4	23	1
One-click rollback	4	5	2	5	5	21	2
Deploy queue visibility	3	3	5	3	2	16	3

Key Takeaways

Code quotes individually before grouping. Premature grouping creates confirmation bias
Use exact quotes, not summaries. Stakeholders respond to real user words, not researcher paraphrases
Quantify patterns. "Users mentioned this" is weaker than "14 of 20 users independently described this pain"
Always log contradicting evidence. If 15 users confirm a pattern and 3 contradict it, the contradictions often reveal important edge cases
Score patterns before acting. The decision matrix prevents you from chasing the most emotional quote instead of the most impactful opportunity

About This Template

Created by: Tim Adair

Last Updated: 3/5/2026

Version: 1.0.0

License: Free for personal and commercial use

Customer Interview Analysis Template