Qualitative methods in UX research: how to start talking with users
Social sciences and UX research: closer than you thought
Let's start with the question. Is it possible to make a really high-quality IT product without user analytics? Yes or no? Who chose yes? If you already know what your users really need, then you can stop reading. Save your time. And money for analytics. True, why do you need it?
But if it is important for you to understand your users, then in this article we will talk about how to do this. The task is not easy. These are not abstract schemes and models to draw: here are living people with their desires, expectations and fears. However, there is good news: social sciences (anthropology, sociology, psychology) have long developed a whole pool of methods for studying man. There is no need to reinvent the wheel: the task is to adapt these methods to your UX tasks without losing data quality or going beyond the budget. This can be done if you follow certain recommendations.
In the article we will consider qualitative methods: observation and interviews. First of all, an interview as a basic method. Quantitative methods, as our experience shows, can become an additional source of information - we are talking about secondary data from open sources. Own quantitative research is expensive and not always necessary.
To understand the motives of users and to study their daily life is possible only with the help of a quality methodology. Of course, you have to go to people: talk (interviews), and even better to see how they live (observation). Fearfully? And without this in any way. This is the only way to get acquainted with the life experience of people in a specific context. Without this knowledge, it is impossible to make a good product. The researcher not only describes the actions (practices), but tries to understand the meaning behind them in order to answer the question: why does a person do this and not otherwise?
Qualitative Methodology in UX Research
Now there will be a little sociology. The main advantage of a quality methodology (or more broadly - a paradigm) is the ability to look at the user with a fresh look. Let's look at this with a simple example.
The young girl Anna cannot turn on her cell phone. Presses a button, but the phone does not respond. Common situation? What would you do? Stop! Answer first, and then read on. Our heroine is looking for the nearest service center on the Internet. How is she looking for him? How to choose from many similar companies? Interesting? Of course. Especially the owner of such a center who wants a new site.
But back to Anna and talk to her. Anna will tell you that her father and elder brother were engaged in repairs in her family. And the equipment was set up, and the "Windows" was reinstalled. Mom called it "men's work." A familiar story? Anna had no experience of close communication with technology, and the idea of delving into the phone herself was completely alien to her. Would you dig around? For example, you can remove the battery and insert it back. I would do so - it works with my phone. Or would you go to the forums? Or maybe they would have acted like Anna?
When studying human behavior, scientists take into account social characteristics: gender, class, cultural differences (and more recently, in addition, age, legal capacity and belonging to a specific community). These characteristics set the framework for our life experience, which forms habits that influence behavior. And we do not always realize this. The story with Anna could have been different if she had been born in another family.
Imagine that Anna did not find a suitable service, and after half an hour she had an important call. Overcoming fear, Anna goes to the forums to look for how to solve the problem. A few simple manipulations, and the phone came to life. Anna is pleased with herself. And now she has a new experience that can form a new habit (behavior pattern). Not necessarily and not immediately, but it can. We, as IT product developers, work with existing habits, but also create new patterns of behavior. Without the understanding of users “here and now” this is impossible. And here a qualitative methodology comes to the rescue.
Large Western IT companies, such as Intel or Microsoft, have long recognized the potential for quality research in UX. The post of corporate anthropologist is already full-time for many. The main methods remain observation and interviews, sociological classics. Researchers come to people home, go to work with them, observe the office routine, living with users all day. And they conduct interviews. Unlike marketers, they are not fixated on the product. They are interested in the daily life of the user. Only by studying it, you can think about which product will help a person solve his problems. Otherwise, your unique selling proposition will be of interest only to you. The results of such studies can prevent the failure of the finished IT product.
From heaven to earth: working in business constraints
Have you been inspired? Ready to get started? It's good to have your own full-time researcher, but as long as you are not Intel, Genevieve Bell is a luxury for you. In general, how much are you willing to spend on user analytics? How many hours do you have for this project? Only honestly. Four? Eight? Sixteen? My experience at Pavlov’s Dog suggests that observation can be forgotten. It’s good if you have time for an interview.
Observation is expensive, but it costs its money. I'll tell you an example. A client came to Pavlov’s Dog who really wanted to understand how customers and staff work with self-service kiosks. Understand not out of sporting interest, but in order to increase sales through kiosks. Here, of course, it is better to see once than hear a hundred times. The observation was organized: the researcher walked around the hall, watched how sellers and buyers behaved, connected himself to work, talked to people. In general, he became a freelance sales consultant for six days. The final findings went far beyond the usability testing of kiosks. It became clear how to adjust business goals, as well as change the placement and color scheme of kiosks in order to increase sales. The customer was satisfied: received valuable results and went to implement.
This situation is rather the exception. There was a large customer, and an understanding of the value of the method was present. In most cases, there is only an interview: cheaper and more understandable for the customer. And this is true not only for the Pavlova Dogs projects. Observation to some extent remains exotic. Not yet ready for this while the Russian customer.
What an observation, when even for an interview they’re not always ready to go. One of the most popular questions from small businesses is: “Can you make a website without analytics?” Of course, they can be understood. But as a result, the analyst has to work under severe restrictions:
- few interviews;
- interviews may not be large;
- material handling should be quick.
It is clear that there are no restrictions. In the social sciences, the researcher is also limited by the grant budget, the requirements of the ethics committee, and university policies. But it is important to understand what such limitations are fraught with. And in the worst case they are fraught with loss of data quality. I must say right away that this risk cannot be completely avoided. But you can try to reduce it. Here are some basic guidelines on how to do this.
Risk to ask too few people
Let's start with the most basic. We really want to understand our users, to make the service convenient for them, but we have barely enough money for only four interviews. Common situation? What problems can arise here? You can reach too few users. Is there enough material collected to draw conclusions on a future product? Less time for an interview - higher risks to make mistakes.
How many interviews do you need? Good question. In the social sciences, they have long been racking their brains over this, but they have not reached consensus. The numbers are called different - from 15 (minimum for high-quality research) to 50 (ethnographic works). As a result of discussions, the concept of information saturation appeared: this is when each subsequent interview does not bring new information important for the study. Then the data collection can be stopped and proceed to processing. The question of when saturation came remains open. Who will decide this? The researcher himself? But what if the following interview brings something important to the project? Dead end.
Practice shows that the decisive role is played not by the number of interviews, but by the correct choice of respondents. There is one more risk - talking to the wrong people. What if there were no active users among them? Or potential users for a future IT product? To avoid this risk, the right selection is necessary.
Our projects use purposive sampling. This is not a random sample that is done in quantitative research. The target sample is determined subjectively, with the help of expert knowledge about the general population: real or potential users. Such knowledge can be obtained from the customer. Perhaps he conducted independent research of his audience. Or keeps statistics. However, practice shows that this is from the category of fiction. Expert knowledge may be absent, incomplete or, in the worst case, incorrect.
In such a situation, you need to use your own expertise. Brainstorming to understand who our users are. When sampling, there is no need to build on demographics. We are more interested in user behavior. For this purpose, this is important.
- User roles. Who will use your product? Head, his secretary, accountant and system administrator.
- Goals. What exactly do users want to do? Make a deal, find background information and order documents.
- Attitude. How is your product or service related? They do not trust, doubt, fear or are ready to acquire.
- Behavior. How do users work with your product? They deliberately delve into the details, want to finish work faster, are not sure and are afraid to make a mistake.
You can develop your own criteria relevant to your project. But the selection criteria given above can help you at the start. Perhaps after the first interview you will have new criteria. For example, during an interview, the respondent said that he has a friend who does everything differently. You should talk to him: it will be useful for your research.
Developed criteria - start your search. This is one of the most problematic parts of research organization. Talking with a person is not difficult, but finding a suitable respondent is much more difficult. You are tempted to question your colleagues: they are sitting nearby and are always ready to help. Do not do this. Unless, of course, the product is not for the same IT people. Try to avoid talking with them - these are no longer “regular” users. It is good when the customer can offer respondents and help with arrangements. Especially if it's top management. If you can’t count on customer help, get ready for a search: social networks, snowball method, etc. And take enough time for it.
When you make a list of potential respondents and agree to conduct an interview, you need to check them for compliance with the selection criteria. A “screener” may come in handy here: a small online questionnaire or mini-interview. Interviews are preferable: listen to how a person expresses his thoughts. Not all people are sociable and willing to share stories, namely you need them. Each interview should be as informative as possible.
Risks of running out of time during interviews
Now about the informative interview. In social sciences, the time for an interview is practically unlimited: the interviewer can talk with the respondent for two, three, and five hours. May come to a person several times. In our case, there are n hours for conversations with users. And you need to talk to more than one. The time allotted for the interview is distributed to several people. On average, one conversation takes from 40 minutes to an hour and a half. We talked to someone more, but there’s not enough time for someone else.
What could go wrong in a limited time environment? You may not have enough time to “warm up” the respondent. A person should get used to you and the interview situation itself, only then he will be ready for a conversation. One more problem: you did not receive the necessary information. The conversation started, they talked "for life" - about service in the Navy, about a dog, children - and the topic of the study was forgotten. As a result, hours were spent and the interview was “empty.” A similar problem: we disassembled one topic in detail, and walked along the other in passing. Unfortunately, in most cases you understand this when processing data. Repeated conversation is not always possible.
There are several problems, but there is only one solution: use a guide and a semi-structured interview. A guide - but simply a questionnaire - contains the main thematic blocks and recommended questions for expanding the topic. Such a tool allows you to control the course of the conversation, but does not drive the interviewer and respondent into a tight framework. Therefore, such an interview is called semi-structured.
A guide is a tool that is created for a specific project. In Pavlov’s Dog he is composed of an analyst (aka an interviewer) and a project manager. The final version is being discussed with the customer: maybe he is interested in learning something else about his users? Interview results often go beyond creating an IT product: they can be used to develop a company’s development strategy. There are some basic guidelines for compiling a guide.
- Any guide contains a block about the respondent. Imagine you are meeting a person. How old is he? What kind of education? What does he do? Such a block goes first to warm up a person. Questions vary depending on the project, but some basic ones may persist in different situations. In our case, a universal block on digital habits: using the Internet, a smartphone, gadgets, etc. Still, we are developing IT products. Link to the block on digital habits: bit.ly/1pOaao7 .
- Remember we don’t have an observation? And we need to learn about the daily life of the user. When compiling a guide, this should be taken into account. It should have questions that make it clear what the user is doing. Each such question has an additional “why.” Motives are also interesting to us.
- Your task is to transfer knowledge about users to the team. In our case, documents are used for this: portraits, life situations, user questions and expectations. Include in the guide questions that will provide the information necessary for the preparation of documents.
- It is important for us to understand the pain of the user. Ask a person about his emotional state: what did he feel when he committed the action? Perhaps he will tell you about it himself. Be careful and ask a clarifying question. For example, "I was afraid to contact a real estate agency." And what exactly were you afraid of?
These are the main guidelines for compiling a guide. But, no matter how good he may be, much depends on the communicative skills of the interviewer and his ability to conduct a conversation. Giving advice here is hard: a lot comes with experience. Here are some basic guidelines.
- Do not arrange an interview from an interview. Your guide is a way to direct the conversation in the right direction. Asked a question - wait. Do not skip immediately to the next question. A good interviewer can listen and ask the right questions at the right time.
- Your task is to hear the story from the user. Not everyone can tell stories. Try to prompt the person with the leading questions to the story. Do not ask questions where the answer is yes or no.
- More why questions. I'm serious. Do not make yourself an expert: you do not know what the other person has in mind.
- Do not limit yourself to product usage questions. Try to find out something interesting about a person’s daily life. Perhaps this will lead you to a new idea. By focusing only on the product, you lose the ability to perceive new information.
Interviews are not as easy as they seem. But it can be mastered. The greater the “run” of the recorder, the better each subsequent interview.
Risks of lack of time in data processing
Speaking of the recorder. After you interviewed, you need to quickly process the data and prepare documents for the team. Documents should be such that all user information is reported.
What could be the problem? I work with a dictaphone recording, but someone can immediately enter information into a notebook or laptop during an interview. Each method has its pros and cons. Recording during an interview distracts you from the conversation. And you may not have time to fix something important: you have to immediately decide what is significant and what is not. Some information may be lost. The main problem of the dictaphone recording: to compose a short recording of the interview, you have to listen to it again. Often this takes no less time than the conversation itself. But all the information that is received from the user is in mp3 format.
If processing time is limited, there is a risk of losing valuable information. Nobody will listen to the dictaphone recording, and a complete decryption is an expensive and long task. Therefore, only you decide what gets on the paper. In order not to miss significant information, you need to clearly understand what you wanted to know in an interview. Think of the project’s meta-goal and create text that works for it.
The interview recording should be structured. The easiest way to do this is to copy the structure of the guide. Another option: make headings and break the text so that it is easy to create the necessary documents from it (portraits, life situations, expectations, etc.). Whatever method you choose, it is important to maintain a structure for all interviews. This will facilitate the subsequent comparison when creating characters and compiling other documents.
The structure is important, but your main task: to convey the pain of the user, to describe his actions and motives. A good recording of an interview in itself is a valuable document that project managers and interface designers can use. You can convey the user's pain and motives using quotes. It is not necessary to completely decrypt the audio recording, but vivid quotes should be kept. Then they will help you revitalize user portraits.
Compiled a document? Viewed it again? Do you recognize the person you were talking to and his story? If so, then this is a good record and you can work on it further. An example of our document can be downloaded here: bit.ly/1pOakM9 .
Afterword. Is it worth it?
Interviews are a qualitative method that is used in the social sciences. But it can also be used to solve UX problems. If you cannot afford observation, an interview becomes the only method of obtaining information about the daily lives of users and their motives.
Working with an interview requires adhering to certain principles that reduce the risks of receiving low-quality data due to limitations: a small number of respondents, limited talk time and fast data processing. The basic principles are as follows:
- Use target selection
- develop a unique guide for the project (an example - a block on digital habits - can be downloaded from the link: bit.ly/1pOaao7 );
- make a high-quality recording of the interview (an example is available at: bit.ly/1pOakM9 ).
The recommendations given above will help reduce the risks. To dive into the topic, you can listen to the audio recording of the report on AnalysDays ( http://bit.ly/1NpbGIT ) and watch the presentation ( http://bit.ly/1So0cSo ).
Unfortunately, there is always a risk of getting low-quality data. And the more restrictions, the higher the risks. The question remains: if there is no way to conduct a lot of interviews, maybe do without them? Why all these problems? What we ourselves do not understand what people need? Not. Even one or two interviews are better than nothing. The more restrictions, the more responsibility on analytics when using interviews. But this does not mean that you need to abandon the method. Just be careful, okay?