Overview of Consumer Research Methods

Compare the main consumer research methods, from surveys and interviews to analytics, and see which fits the goal, sample size and budget.

Melissa Tovin

A travel booking app team whose repeat bookings are falling needs to learn why, and the method it picks shapes what it learns. Consumer research methods fall into three groups: asking consumers, watching what they do, and drawing on data that already exists. The team could survey lapsed travelers, watch them search for a trip, or pull booking logs, with each route returning a different kind of answer.

Consumer research methods at a glance

Research teams use a short list of methods for most studies. Each one below gets a single line here and a fuller treatment further down.

Methods teams use to ask consumers:

Surveys: Structured questionnaires that measure how common an answer is across many people.
In-depth interviews: One-on-one conversations led by a human moderator who follows each participant’s story.
AI-moderated interviews: Voice conversations in which an AI moderator follows the researcher’s guide and asks follow-up questions, many at once.
Focus groups: Moderated group discussions that capture shared reactions and category language.

Methods teams use to watch consumers:

Observation and ethnography: Seeing behavior in its natural context, in person or online.
Diary studies: Participants log experiences over days or weeks.
Usability testing: Watching people attempt tasks on a product or prototype, including a site.

Methods teams use to examine existing and passive data:

Product and web analytics: Teams use logs to track behavior inside the product.
A/B and multivariate testing: Teams compare design variants in live tests.
Social listening and review mining: Teams examine public posts and reviews for unprompted opinion.
CRM, support and sales data: The company already holds these records and recordings.
Secondary and competitive research: Teams examine published reports and public data, alongside competitor reviews, for context.

Customer research focuses on a company’s own customers and prospects, while market research covers the wider market.

Two ways to classify any method

Every method sits on two axes. One is where the data comes from, and the other is the kind of question it answers.

Primary vs secondary research

Primary research is a method in which the team collects new data itself for the question at hand. Secondary research is a method in which the team uses data that already exists and was gathered for some other purpose. The travel app’s lapsed-consumer interviews are primary; an industry report on travel spending is secondary.

Any data, primary or secondary, can also be sorted by who owns it:

First-party data: Data the company collects from its own consumers’ interactions, such as bookings and app sessions.
Second-party data: Another company’s first-party data, shared under an agreement.
Third-party data: Data bought or sourced from an organization with no direct relationship to the consumer.

Qualitative vs quantitative research

Qualitative research is an approach in which researchers talk with or observe people to learn why they act as they do. Quantitative research is an approach in which researchers count how many people do something and how much of it they do. A drop in repeat bookings is a quantitative fact. The traveler who says the second trip “felt like starting over” points to the reason. Strella’s comparison of quantitative vs qualitative research sets the two side by side, and its overview of the types of qualitative research breaks the qualitative side down further.

Asking consumers

In these methods the team prompts a response. Surveys aside, they are the core tools of qualitative market research, and each one trades depth against scale differently.

Surveys

A survey is a research method in which a team sends the same structured questions to many people and each person answers them, so the team can see how common each answer is. Surveys include online and in-product questionnaires for CSAT and NPS tracking, alongside studies sent to panels. They are best for sizing a known issue and tracking it over time. Their limits come from the format: fixed answer options only capture the reasons the team thought to list, and open-text answers tend to be short and shallow. The travel app would use one late in a study, after interviews have surfaced the reasons, to measure how common each reason is among lapsed travelers.

In-depth interviews

In-depth interviews are human-moderated one-on-one conversations that suit motivations and decision journeys, and human moderation is most valuable where cultural nuance or emotional weight is high, because a skilled moderator’s rapport and judgment shape the answer. The limit is time, because researchers must schedule and moderate every interview, then synthesize it by hand. The travel app might run a handful with its most frequent former travelers to map how they decide where to book.

AI-moderated interviews

In AI-moderated interviews, real people hold live, voice-to-voice conversations with an AI moderator that follows the researcher’s guide and asks follow-up questions based on each answer. Many interviews run in parallel.

The method sits between surveys and in-depth interviews, bringing interview depth to work at a scale and speed closer to a survey. Forrester’s 2026 report “Meet Your New Research Partner: AI Moderators” credits it with qualitative insight at scale across languages and time zones, though not as a replacement for in-depth, human-led research.

It is best for concept, messaging, pricing, churn and cross-market questions that need dozens or hundreds of conversations. The travel app would use AI moderation to hear from lapsed travelers in several markets within the same week.
The AI moderator follows the guide well, but it does not bring a skilled interviewer’s judgment about which thread matters most, so the guide has to be strong. Teams decide between AI and human moderation study by study, or combine them in a hybrid approach.

Focus groups

A focus group is a moderated group discussion in which a researcher brings together five to eight people per session, usually across several sessions, and the participants react to stimulus such as ads or packaging. Groups also help researchers learn the words a category uses. The limit is the room itself: dominant voices set the tone and quieter participants conform. Researchers can mistake that silence for agreement. The travel app might use groups to hear how travelers describe a good deal before writing survey answer options.

Watching consumers

People’s reports and their behavior often diverge, a pattern researchers call the say-do gap. A traveler may say price drove the switch to a competitor when the booking log shows they abandoned the app at a login screen. Watching methods capture the behavior directly.

Observation, ethnography and diary studies

Observation and ethnography mean seeing behavior in context, whether in homes and stores or through digital ethnography of the communities people use online. Mystery shopping belongs here; evaluators act as consumers to check whether the intended experience happens. Diary studies capture behavior over time, with participants logging entries across days or weeks, and small discovery diaries often run with roughly 5 to 12 people. The travel app could ask lapsed travelers to log every trip-planning moment for two weeks and see which app or site they opened first.

Usability testing

Usability testing is a method in which researchers watch participants attempt tasks on a product or prototype, including a site, to see how well they complete them. Moderated sessions let a researcher ask why someone hesitated. Unmoderated sessions run faster but capture the click without the reason. Rounds are small, often about five participants per round, or three to four per group when testing several distinct groups, with several rounds as the design changes. The travel app would test its rebooking flow to see whether returning travelers can find saved trips and loyalty points.

Existing and passive data

This group needs no new recruiting. It shows behavior at scale but rarely explains it.

Product and web analytics

Analytics record what consumers do; the logs carry no reasons. For the travel app, the booking funnel can show exactly where returning travelers drop off by device and market, with further breakdowns by trip type.

A/B and multivariate testing

A/B testing is a live experiment in which a team shows different variants to different groups of consumers and compares their behavior. Multivariate testing is a live experiment in which a team shows combinations of several elements at once and compares how each combination performs. Nielsen Norman Group’s 2024 article “A/B Testing 101” notes that tests often need thousands of consumers before the team can trust the result and that they show how behavior changes without giving any insight into why. Multivariate tests need still more traffic.

Social listening and review mining

App-store reviews and social posts offer unprompted opinion at low cost. They skew toward the vocal minority, because people with extreme experiences are the most likely to post. The travel app’s one-star reviews will overstate anger and understate quiet drift.

CRM, support tickets and sales call recordings

A company’s support records are often its most underused source, whether researchers examine tickets and CRM notes or call recordings. Research repositories can take these files and analyze them alongside new interviews. The travel app’s refund complaints from the past year may already name the problem.

Secondary and competitive research

Teams use syndicated reports and public statistics to understand how travel demand is moving, and competitor reviews to learn what travelers praise or criticize in rival apps. The travel app could read rival apps’ reviews to see whether competitors show full prices earlier.

How the methods compare

The table sets the main methods side by side using relative time and effort:

Method	Type	Typical sample size	Relative time	Relative effort	Kind of answer	Best-fit question
Survey	Quant, primary	Hundreds to thousands	Days	Low to medium	How many, how much	How common is each reason?
In-depth interview	Qual, primary	5 to 12 per segment	Weeks	High	Why, in depth	How do travelers decide?
AI-moderated interview	Qual, primary	Dozens to hundreds	Days	Medium	Why, at scale	Why did lapsed travelers leave, across markets?
Focus group	Qual, primary	5 to 8 per group, several groups	Weeks	High	Shared reactions	How do travelers react to this offer?
Observation and diary	Qual, primary	5 to 12	Weeks	High	Behavior in context	What happens before a booking?
Usability test	Qual, primary	About 5 per round	Days	Medium	Task friction	Can returning users rebook?
Analytics	Quant, secondary (first-party)	Full user base	Hours to days	Low	What happened, where	Where did repeat bookings drop?
A/B test	Quant, primary	Thousands	Weeks	Medium	Which variant wins	Does the fix lift rebooking?
Social listening	Qual, secondary	Varies	Days	Low	Unprompted opinion	What do angry users say?
CRM and support data	Qual and quant, secondary (first-party)	Existing records	Days	Low to medium	Recorded complaints	What did travelers report?
Secondary research	Quant, secondary	Varies	Days	Low	Market context	Is travel demand shifting?

How to choose the right method for a decision

The decision comes first, then the method. The travel app’s questions map to methods like this:

Why a metric moved: The team uses analytics to find where repeat bookings fell, then interviews travelers in that group to learn why.
A new concept or message: Interviews show how travelers read a new loyalty offer, then a survey picks between the two strongest versions.
Pricing: Interviews reveal what travelers value in a booking fee or membership, then a survey or conjoint study sets the price point.
Usability problems: Usability testing on the rebooking flow finds where people get stuck.
Market sizing: A survey plus secondary data estimates how many travelers book trips more than once a year.
Entering a new market: Interviews in the local language uncover how travelers there book, then a survey sizes the opportunity.

Sample size and relative effort narrow the choice further. A team with a small budget can start with analytics and support data, both low effort, and add a handful of interviews. When the question needs conversations at the scale of dozens to hundreds, AI-moderated interviews fit. Projecting a finding to the whole consumer base takes a survey of hundreds, and because an A/B test needs thousands of consumers, a low-traffic product should rely on usability testing instead.

If the team does not yet know the reasons, it should start with conversations. If it knows the reasons and needs to rank them, it should run a survey.

Combining methods: a worked example

The travel app team runs its study in four steps, using each finding to frame the next question.

The team uses analytics to identify where repeat bookings fell: among travelers who booked once in the past year and searched again but left without booking.
Researchers use AI-moderated interviews with those lapsed consumers to learn why. Several mention fees appearing only at checkout, and others say their loyalty points expired without warning.
The team writes a survey in the travelers’ own words, such as “the price jumped at the last step”, to measure how common each reason is.
The team runs an A/B test, showing full prices on the search page for half of returning users and checking whether rebooking rises.

When researchers have a number with no reason attached, they should follow it with a conversation. When they have a reason with no scale, they should follow it with a count.

Running a consumer research study step by step

Most studies follow the same sequence, whatever the method:

Define the decision and objectives. Write the choice the study informs, such as which retention fix the travel app ships next quarter.
Pick the method. Use the mapping above.
Write the guide or questionnaire. Keep questions neutral and open, using the consumer’s language.
Recruit. Recruit from the company’s own consumers or panels, using expert networks for B2B roles, and screen for fraud. Pew Research Center’s 2026 study of bogus respondents, based on an opt-in survey of 11,114 US adults, flagged 18% of cases as bogus through trap questions alone.
Run the study. Monitor early sessions and fix confusing questions quickly.
Synthesize. Group findings into themes, each linked to the source quote and recording.
Share. Put the findings in front of the people making the decision.

Sample size depends on the question. About 5 to 8 interviews per segment covers a narrow question, and around 12 per segment works when comparing segments. A survey whose results must be projected to the consumer base needs hundreds of responses.

Turning research into decisions

Teams act on synthesis outputs they can use in a meeting: themes linked to verbatim quotes, personas, journey maps, jobs to be done, and highlight reels. For the travel app, a journey map that marks where fees surprised travelers, paired with a 60-second reel of travelers describing that moment, gives a product lead something to fix.

Avoiding bias and staying compliant

Bias usually enters through the questions or format, as well as the sample. Common problems include:

Leading questions and survey fatigue: “How frustrating were our fees?” assumes frustration. “Tell me about the last time you booked” does not. Long questionnaires lose people and flatten answers, so keep surveys short and ask only what the decision needs.
The say-do gap: Pair stated reasons with behavioral data before acting on them.

Compliance comes down to a few plain rules under GDPR and CCPA, alongside research industry codes such as the Insights Association’s:

Consent and notice: Tell participants at the start what is collected, why, how long it is kept and who sees it, and tell participants before recording.
Data minimization: Collect only what the research purpose needs.
Participant rights: Let people withdraw, and respond to access and deletion requests within the legal deadlines (one month under GDPR, 45 days under CCPA).
Separation from marketing: Keep research participation apart from sales and marketing follow-up.

Tools for each method

Each method has its own category of tool:

Survey tools: Questionnaire builders with logic and dashboards, plus access to panels.
Interview and AI-moderated research platforms: Tools for live human-moderated and AI-moderated interviews, including hybrid approaches, with recruiting and synthesis.
Usability testing tools: Prototype and task testing with heatmaps and click paths.
Analytics: Product and web analytics for funnels and cohorts.
Social listening: Monitoring for social posts and app reviews.
Research repositories: Searchable stores for past studies and calls, including transcripts.

Strella is a qualitative research platform that supports AI-moderated and human-moderated interviews, including hybrid approaches, with recruitment and synthesis in the same place. Its AI-moderated interviews are voice-to-voice conversations with real people, and teams choose a level of service from self-serve to fully managed.

Bring your research question to us and we will talk it through. See Strella in action.

Book a Demo

FAQ

What are the main consumer research methods?

The main consumer research methods are surveys, in-depth interviews, AI-moderated interviews, focus groups, observation and diary studies, usability testing, analytics, A/B testing, social listening, CRM and support data, and secondary research. Teams can ask consumers or watch their behavior. They can also use data that already exists.

What is the difference between primary and secondary research?

Primary research is data the team collects itself for its own question; secondary research is data that already exists and was gathered for another purpose. For the travel app, interviews the team runs with lapsed consumers are primary research, while a published travel-industry report is secondary research.

When should a team use interviews vs surveys?

A team should use interviews when it needs to learn why consumers act as they do, and surveys when it needs to measure how many share a known reason. Most decisions call for interviews first and a survey second.

What is the say-do gap?

The say-do gap is the divergence between what people report and what they actually do. Watching methods capture the behavior directly, and teams reduce the gap’s effect by checking stated reasons against behavioral data such as analytics or observation before acting.

How many consumers should a team interview?

For a narrow question, plan on roughly 5 to 8 interviews in each segment; comparing segments calls for about 12 apiece. Findings meant to hold for the whole consumer base need a survey of hundreds.

What is the difference between customer research and market research?

The difference shows in the question being asked: a team checking why its own consumers churn is doing customer research, while a team sizing a category it has not entered is doing market research.