Modern software engineering handout 6 user research

[Part of modern software engineering handouts]

The process of software development is the process of transmitting, transforming, implementing, twisting or losing "what users need most" in the following chain.

What users need most>

Expressed by the user>

What the software team can understand (boss/PM) + the team's business goals>

Specifically expressed by software team members (PM write spec)>

Under various constraints, execute the specific expressions (dev writes code)>

Verified (Test)>

Inform target users through various channels (release/Promotion)>

Users can finally use it, but they are not satisfied>

A well-known swing chart in the software industry has a similar situation:

(This is what the user wants. Has the user expressed it? Does the software team understand it? Is it executed? Is the user satisfied? )

(Maybe the company is good at three-tier architecture, so it also needs three-tier swing)

(PM wrote the spec)

(The developer writes the function based on spec)

(The tester finally agrees to release the software)

Is it the same as what the user wants? Different. Are Users Satisfied? Not satisfied. What do users want? Let's investigate and start a new loop...

 

 

When developing software, we always want to know what the user is thinking, what their preferences for various features are, and grasp the information, so that we can satisfy the user's needs step by step. You can rely on intuition, the boss's command, the Internet to upload a variety of information, rely on copying other software, rely on other unreliable means... Of course, we can also rely on some proven methods. The following are several methods of User Study:

1)Focus Group)

Find a group of target user representatives to discuss what the user wants and evaluate the software. The Focus Group is a common survey method and has some weaknesses:

• When a group of people are together, they usually give their opinions out of the favor of others to avoid inconsistent opinions or conflicts.
• Those who are not familiar with things (such as disruptive innovation) cannot express valuable ideas-before the advent of a car, let's look for a bunch of drivers to think about "future transportation ", they may not contribute valuable ideas.
• The discussion crowd is easily influenced by the host intentionally or unintentionally.
• Researchers often choose from different opinions which best suit their own ideas, and then claim that this is the consensus of everyone.

2) In-depth interview (in-depth interview)

Through detailed interviews, you can gain a broad and in-depth understanding of your background, psychology, and needs. This is usually a one-on-one interview. This method is good, it is time-consuming.

3)Card Sorting)

List the desired software features and classify them. When I was at Microsoft's Asia Research Institute, we had also done "card sorting"-people from different backgrounds gathered together and imagined what features the new software had and how it could solve their own pain points, or there is something interesting about it. Write these features on the small card. A host sorts and discusses different cards to further clarify the relationship between various desires. In a sense, these cards are the opinions of the quantified focus group. These cards are categorized, sorted, and defined to help us better define the information architecture of a software, user workflow, software menu structure, website browsing path, hierarchical relationship of various content, etc.

4) User Survey)

Give the user a pre-defined question and ask the user to answer it. We have met many people on the street. Sometimes, when you are browsing a website, a pop-up window breaks your thinking and asks you to answer a few questions. When answering such questions, are our users absent-minded and confused?

User surveys seem easy. In fact, there are many common problems:

A) The problem definition is inaccurate, for example:Which search engine do you use?Users may provide multiple reasonable answers: recently used; favorite but not always used (for example, the favorite search engine cannot be accessed for some reason ); used for a certain field (for example, querying images or English words). The use of e-search engines may vary in the last week, January, or one year.

Questions with inaccurate definitions may confuse users. We may be able to collect many answers, but still cannot understand the users' ideas accurately.

B) vague adjectives and adverbs are used to describe time, quantity, frequency, and price: recent, sometimes, often, occasionally, seldom, many, quite a lot, very expensive, very cheap. These words have different meanings for different users and in different contexts.

C) Ask the user to spend extra effort to answer the question: how many mobile phone applications do your family download each year on average?

D) The problem has A guiding tendency: users generally believe that search engine A contains A lot of copyrighted materials and refuses to recognize errors. Search Engine B wins user trust, will you select A or B?

E) the issue involves user privacy and the situation of the user's company.

You can use the following methods to solve the user questionnaire:

A) open-ended questions: for example, what is the expectation of your opponent's calendar management software :________________

This problem can make users speak freely, but it is difficult to sort and quantify.

B) Two-choice questions: you only need to answer Yes/No. This type of problem facilitates statistics and analysis. However, users do not have the opportunity to further clarify the reasons, so it is difficult to reflect the differences between opinions and degrees, and the situation is not in-depth enough.

This type also has a variant, that is, only one of the two options can be selected for comparison.

C) multiple-choice questions, which you may encounter multiple times in your usual exams.

D) proper choice: the main considerations for choosing the mobile phone back word software are (1, 2, 3 ,... ): _ Vocabulary; _ ability to record progress; _ ability to customize word lists; _ ability to synchronize with PC; _ ability to support specialized Word lists such as Level 1; _ ability to support pronunciation.

5) User Diary Study)

Users are required to record their daily work or activities related to the software they use for future analysis. You can write a text description like a diary or enter a table every day (for example, tracking your daily diet ). As I personally understand, the user survey will be extended in time. This requires high user self-discipline. In addition, how to protect user privacy is also a problem.

6) Study of ethnic logs/human species (Ethnographic Study)

This seemingly academic method can be interpreted as-working with the target user. For example, instead of sitting in the office and imagining how to design a mobile phone for the elderly, it is better to spend a few days with the elderly and get data and experience from their lives. This is an example.

User surveys in anthropology may sound profound, but they may not-maybe you have been living in the target group, but you are not sensitive enough to these needs. In the movie the social network, Mark asked him, do you know a girl has a boyfriend? Mark pondered for a while and ran straight back to the dormitory. He realized the "do you have friends" function on thefacebook.com.

Some college students who have ideas give directions in the ivory tower to boost the text. In the real world, you can see your real needs. The following is your epiphany:

Http://www.cnblogs.com/meng-meng/archive/2011/11/14/2248589.html

I usually use computer science, and my website is geek or hacker. I almost never use qq, I never post hundreds of times, I never play games, I don't need 360 or any anti-virus software, and I don't need hao123 as my homepage. I have nothing to do with google reader. I turn to twitter and facebook on the wall. I often visit hacker news and quora. I am happy to try various new cool sites abroad, I have never paid for software or services.

I didn't know much about the massive number of Chinese users. The real users were not what I imagined.

In the past, I didn't understand why 360 of the installation volume is so large. Now I understand: 1. a massive number of users do not know how to manage and use computers. 360 of that silly one-click solution is what they need. 2. they don't want to spend money, but they don't want to find any "cracked version", "Serial Number", or "registration machine"

I didn't understand why hao123's "mentally retarded" website has such a big impact. Now I understand that my father can easily go to Sina to view news through it, but if you ask him to enter the URL directly, he will certainly enter "xinlang.com"

In the past, I didn't understand why so many people were willing to pay for the virtual image on qq. Now I understand, my cousin, as long as they had to go online, qq was definitely suspended, and girls love beautiful and vanity, they don't care about spending some money to dress themselves up.

I have seen some computers of non-computer students, mostly messy, with files everywhere. The software is installed on the C drive by default, and a bunch of rogue software is bundled during the installation process, all kinds of software are started upon startup, and it does not start after 3 minutes ......

 

This is a description of digital life in China's third-tier cities in 2013 (by Pan yufei). Isn't it a kind of Anthropology?

7) Research on software availability (Usability Study)

Study the difficulties of users when using the software, and how to improve the software to make the software easier to use. The common method is to ask the User to come to Microsoft to have a special User Study Studio and often recruit the target User for testing, I have also visited users on the site to use the new version of Outlook (we are behind the one-way glass window ). More team members can watch the video of these user surveys afterwards. Investigators usually ask the subject to complete some tasks, such:

In Excel, how can you change rows and columns in a table?

After the Chinese New Year, If you want many customers to send congratulation emails with similar content, but their names and addresses are different, how do you use Word/Outlook to complete this task?

In the Excel software, you need to move back and forth when reading some large tables, but the title bar of the table will not be seen. How can you lock the title bar?

I was very impressed by the fact that the user was looking for a function in our long menu, and we worried about him behind the glass window... Our interface is too far away from "being simple and being simple.

8) Eye Movement Tracking (Eye Tracking)

Your software shows a lot of information and has a lot of interactive control. How can users easily find the information you want them to see and find the functions they want to use? What is the common rule for users to view a large amount of content on a webpage? Some research foundF mode.

Users usually browse the topic title, and then look down on the left, and then browse the subtitle in parallel. If you want users to know the important content, where should you put it?

9) Paper model survey (Paper Prototype)

If you want to do a good job of the software, it will take too much time and the modification cost will be high. Can users get feedback quickly? In this case, we may take some paper models for users to use and get feedback. I have never done this method, but it sounds good. This is also an example of user Participatory Design. In addition, the model doesn't have to be made of paper, and it can be done with small wood blocks.-Jeff Hawkins, founder of Palm Pilot, uses a small wooden board to make it the same as the design. He put it in his coat pocket, from time to write and draw... the last release of Palm Pilot and its series of products created a new industry like PDA.

 

10) A/B Testing (A/B Testing)

If you already have some users using your product and want to improve the user interface, but you do not know how many users will like the new interface, what should you do?

For example, your website has two columns of layout, but you would like to try the three columns of layout method, just like the question map.

For example, if you want to use a pop-up window to prompt users to respond to an important message, do you want to put the pop-up window in the lower right corner or in the center of the screen?

 

At this time, you are looking for new users to do one-on-one in-depth research, or go to the street to send a questionnaire? Why can't your existing users tell you which design is better? Consider A/B Test.

A/B testing looks simple:

1) determine the two different UIS for the test, as well as the measurement standards, data collection procedures, test run time, and number of people

2) implement A/B testing technically (typically running tests on 5%-10% of users)

3) Collect data, analyze data, and draw conclusions

 

Wired Magazine wrote an article about A/B testing-taking the Obama campaign as an example, it is worth noting.

Many Internet companies are using A/B testing, and researchers are also using this method for research. When you visit their websites, it is possible that they have been tested, take a look at Amazon and Microsoft examples (thesis ).

 

A/B testing also has weaknesses:

A) the operation cost increases over time, increasing the complexity of website O & M and testing the website data collection and data mining capabilities.

B) You cannot see the user's emotions. You only see the interaction behavior, but the interaction behavior is not all feedback from the user. (For example, if you see more advertisements in the center of the screen, but more users may curse advertisements in the center of the screen)

C) What is the relationship between different factors? For example, Which of the following is the relationship between "three-column layout revision" and "User stay time on the website?

-Unrelated. The currently collected data is only random.

-Related, but not the deciding factor

-Causal relationship

D. experiment with the current website user. In case of great dislike, the user will be lost.

 

Attitude: behavior qualitative: Quantitative

Represents the separation of these methods in (attitude: behavior qualitative: Quantitative:

 

There are many types of users

We have talked about this multi-user survey. Many people assume that the software is purchased, but the software is used, but not necessarily. See the following example:

A) You want to write a software for middle school students to learn English. Who are you looking for to do user research?

Middle school student-End User
Parents-they are the payers. They do not use software every day. Some people do not speak English, but they also have demands.
School teachers-they are very influential people. They may set up a rule, and our class will use XX software!

B) You want to write an enterprise management software. Who are you looking for to conduct user research?

 

What if I overdo it? 

We have so many tools and the Internet has brought us so many users and data. This is a good thing and has side effects.

Companies that can access user data in the world and perform analysis and improvement based on the data are probably the leader in Google. This data-centric approach has taken the lead, and there is also a tragedy:

Douglas Bowman used to be Google's head of visual design. One day in 2009, he couldn't stand it anymore:

Yes, it's true that a team at Google couldn't decide between two blues, so they're testing 41 shades between each blue to see which one performs better. I had a recent debate over whether a border shoshould be 3, 4, or 5 pixels wide, and was asked to prove my case. I can't operate in an environment like that. I 've grown tired of debating such minuscule design decisions...

When your company asks you to use data to prove which one of the 41 blue colors is better, or dispute over a sidebar whose width is 3, 4, or 5, what do you do if you want to prove it with data?