> ## Content Index > Fetch the complete content index at: https://pulsecorp.co/blog/llms.txt > Use this file to discover other available public pages before exploring further. # Data collection methods: how to pick the right one - URL: https://pulsecorp.co/blog/data-collection-methods/ - Published: 2026-10-26T12:00:00.000Z - Updated: 2026-10-26T12:00:00.000Z - Description: Data collection methods compared: what surveys, interviews, observation and records each tell you, and three checks to pick the right one for you. - Author: Khyati Ojha You work at a food delivery company, and your manager has a question: how many customers are ordering less, and why? You have a customer list and a week. So where do you start? How you gather the answers is called your data collection method. Each method answers a different kind of question. If you want to estimate how many customers are ordering less, use a survey or the company's order records. If you want to understand why they order less, talk to customers in interviews. So how do you pick one? Start with three checks: what your answer must say, who you can reach, and how much time you have. ## The short version - Start from the question, because each data collection method answers one kind. - Order records show what customers did, and surveys count how many say something. - Interviews explain why, because you can ask a second question. - Who answers a survey matters as much as how many answer. - Watching people use the app shows what they do, which can differ from what they say. - If you don't yet know the reasons, ask first and count second. ## What each data collection method can tell you Your team has many ways to [collect data about customers](https://pulsecorp.co/glossary/data-collection/), and most teams pick one by habit. The habit is usually a survey. But each data collection method answers one kind of question, so the right one depends on what your answer must settle. At the food delivery company, your manager might need a number, a reason or a picture of what customers do in the app. Each data collection method gives the team a different answer, and each one leaves something out. | Method | What it tells you | Food delivery example | What it can't tell you | | ----------------------------------------------- | ----------------------------------------------- | ----------------------------------------------------- | ---------------------------------------------- | | Survey | How many customers say or choose something | How many customers say delivery is too slow | Reasons nobody on the team thought to list | | Order and app records | What customers did, and when | How often each customer orders, and when they stopped | Why they stopped | | Interview | Why people act the way they do | Why a regular customer now orders less | How common that reason is across all customers | | Focus group | How a group reacts to an idea | How customers react to a new delivery fee | What each person would do on their own | | Diary | How a habit plays out over several days | How ordering fits a normal week of meals | How many customers share that habit | | Observation or usability session | What people do in the app | Where people get stuck at checkout | What they think or feel, unless you ask | | Existing records (support tickets, app reviews) | What customers already said without being asked | What complaints about late food say | What customers who never complain think | Some of these methods count and some explain. Surveys and order records give you quantitative data, which means numbers you can add up and compare. Interviews, diaries and focus groups give you qualitative data, which means words and actions that show reasons. Most real questions need both kinds, because a number without a reason is hard to act on. The split between [qualitative and quantitative research](https://pulsecorp.co/blog/qualitative-vs-quantitative-research-differences-methods-when-to-use-each/) also changes how you analyze what you collect. The last row is different. Support tickets and app reviews already exist, because customers wrote them for another reason, so they are called secondary data. Everything else in the table is primary data, which you collect yourself for this question. Secondary data is quick to check, so it makes a good first stop. But it holds only what customers chose to say. The split between [primary and secondary research](https://pulsecorp.co/blog/primary-vs-secondary-research-differences-methods-when-to-use-each/) tells you how much new data you need to collect. ## Surveys and records tell you how many Say your manager wants a number, such as how many customers think deliveries arrive late. A survey can estimate that, because it asks every customer the same question with the same answer options. The company's order records count behavior. They show that weekday dinner orders fell, and which customers stopped ordering. But records show what people did, and never why they did it. A survey has its own limit. It can count only the answer options someone on your team wrote. So it can count reasons you already suspect, but it can't find a reason you missed. Who answers a survey matters as much as how many answer. The response rate is the share of people you invite who answer, and it is often low. [Pew Research Center reported in September 2026](https://www.pewresearch.org/methods/2026/09/16/why-many-polls-need-to-weight-on-party-affiliation-or-past-vote/#:~:text=The%20Center%E2%80%99s%20American%20Trends,panel%20and%20wave%2Dto%2Dwave%20nonresponse.) that response rates to its telephone surveys fell from 36% in 1997 to the low single digits in recent years. The same September 2026 Pew report puts the response rate for its American Trends Panel, an online survey panel, at about 3% once recruitment and later survey waves are counted. In plain words, most people who are asked never answer. So the people who do answer may differ from the people who don't. At the food delivery company, customers who stopped ordering may be the least likely to open a survey email. Yet they are the customers you most need to hear from. ## Interviews tell you why customers changed ![Three steps of an interview: a customer says ordering got annoying, you ask what got annoying, and she explains the delivery fee went up and her food arrived cold more than once](https://pulsecorp.co/blog/content/images/2026/10/data-collection-methods-q3g0qm.webp) The order records show the drop, but nobody on your team knows why it happened. An interview can find out, because you can ask a customer what changed and then ask again when the answer is vague. Say a regular customer tells you ordering "got annoying." You ask what got annoying. She says the delivery fee went up, and her food arrived cold more than once. That second question is where the real reason showed up. Interviews explain reasons in depth. But a small group of interviews can't tell you how common each reason is across all the company's customers. Other methods that explain show different sides of the same problem. A focus group puts several customers in one conversation, so you can see how a group reacts to an idea such as a new delivery fee. But quieter people may follow the loudest voice, so you hear the group's view more than each person's. A diary asks customers to note their meals over several days. It shows how ordering fits a normal week, such as the nights people cook because they got home early. Observation means watching what people do, and a usability session is observation inside the app. It shows where people get stuck at checkout, which can differ from what they tell you in an interview. Each of these methods leaves you with notes and transcripts, so plan time for [how to analyze what customers said](https://pulsecorp.co/blog/how-to-analyze-qualitative-data-methods-steps-examples/) before your manager needs the answer. ## How to pick a data collection method ![Decision flow: for a number, use a survey or order records if you know the answers to count, or interviews first and a survey second if not; for a reason, use interviews; for a behavior, watch people use the app.](https://pulsecorp.co/blog/content/images/2026/10/data-collection-methods-6fn2n4.webp) You can pick a data collection method by running three checks in order. Each check rules some methods out, so by the end you have a short list that fits your question, the customers you need and your week. ### 1\. Decide what your answer must say Write down whether your manager needs a number, a reason or a behavior. How many customers order less is a number. Why they order less is a reason. Where they give up at checkout is a behavior, and each one points to a different method. ### 2\. Check who you can reach Current customers are easy to reach, because they open the app. Customers who stopped ordering are harder to reach, so they need a different invitation, such as an email instead of an in-app message. Decide which group your answer depends on before you pick a method. ### 3\. Match the method to your time and the decision Order records are usually the fastest, because the data already exists. A survey needs writing, testing and sending. Interviews need the most setup, because each conversation is booked one at a time. The bigger the decision that rests on the answer, the more time it deserves. Suppose the food delivery team runs all three checks. First, the order records show that weekday dinner orders fell. Next, the team interviews customers who now order less, and hears about late food, a new fee and cooking at home. Last, a short survey asks all customers which of those reasons fits them, so the team can count each one. Each step settles one part of your manager's question and leaves the rest to the next step. | Step | Method | What it can show | What it can't show | | ---- | ---------------------------------------- | ------------------------------------------------------- | ------------------------------------------ | | 1 | Order records | Weekday dinner orders fell, and which customers stopped | Why they stopped | | 2 | Interviews with customers who order less | The reasons, in customers' own words | How common each reason is | | 3 | Short survey of all customers | How many customers share each reason | Reasons nobody mentioned in the interviews | Using more than one method in one study is called [mixed methods research](https://pulsecorp.co/glossary/mixed-methods-research/). Your plan for which methods to use, in which order and with whom, is your data collection methodology. If the decision needs a number, count. If it needs a reason, ask. If you don't yet know the reasons, ask first and count second. ## Where Pulse fits Interviews explain why customers order less, but booking calls with enough customers takes time. It takes longest with customers who stopped ordering, and they are the ones you most need to hear from. Pulse runs [AI-moderated interviews with customers](https://pulsecorp.co/product/ai-interviews/). Customers take them by opening a link on a phone or laptop, so there are no calls to schedule. The moderator asks a second question when an answer is vague. You can upload a list of the company's customers, including those who stopped ordering, or share a link. Pulse groups what people said into themes, and every quote links to the transcript it came from. Surveys can run alongside the interviews, so one study can find the reasons and count them. The order records still show what changed and when. If you want to hear why customers order less, [book a demo](https://pulsecorp.co/demo/). ## Questions readers ask ### 1\. What are the most common data collection methods? The most common data collection methods are surveys, interviews, focus groups, observation, diaries and existing records such as support tickets or app reviews. Many teams also use behavior data, like the order records of a food delivery company. Each method answers a different kind of question, so most studies use more than one. ### 2\. What is the best data collection method? No single data collection method is best for every question. The best one is the method whose answer matches the decision in front of you. A survey is best for estimating how many customers share a view. An interview is best for learning why a customer changed what they do. ### 3\. What are some common tools used for qualitative data collection? Qualitative data collection needs a few simple tools. You need an interview guide with your main questions, a way to record and transcribe each conversation, and somewhere to keep notes and quotes. Diaries need a short daily prompt, and usability sessions need a way to watch someone use the app. ### 4\. What is quantitative data collection? Quantitative data collection gathers answers you can count and compare as numbers. Surveys with fixed answer options, order records and app usage data are the usual sources. At a food delivery company, it tells you how many customers ordered less this month and which neighborhoods saw the biggest drop. ### 5\. How do you collect data for a research project? Start by writing the one question your project must answer. Then pick the data collection method that fits it, decide which customers you need to hear from, and collect the data the same way for everyone. Check the first few answers early, so you can fix a confusing question before it spreads. The steps after that, from analysis to the report, are part of [running a customer research study](https://pulsecorp.co/blog/customer-research-methods-examples-how-to-do-it/).