1. Who are the users and customers?

(1) Survey: The cheapest way to know who users are, what they want, what they are doing, what they buy, where they shop and what they have is to survey them.

(2) People/market segmentation: transforming the survey results into meaningful clustering. What features do specific user groups want, and what do they care most about when making purchase decisions? Don't just consider gender, income and age, but also take tasks and field experience as key distinguishing indicators.

(3) Competitive analysis: Few products or websites have been done before. Know your market, find out similar companies in the market and look at similar industries. What * * * features are the same? What can please customers? Use industry benchmarks, such as net recommendation score to measure word of mouth and system usability scale to measure usability.

(4) contextual inquiry: users don't always explain clearly what they need or want. By observing how users solve problems and achieve goals in the workplace or at home, we can find unmet needs and understand the tasks they perform.

(5) Stakeholder interviews: A lot of information already exists in different departments of the company. Structured interviews can be used to ask about customer service, QA, development, marketing and sales to find out what needs to be established, modified and eliminated.

(6) Quality function deployment: organize the ideas from internal stakeholders and the data from users and customers into a matrix to understand which functions can meet most internal and external needs.

2. What do users want to do?

(1) Task analysis: decompose the tasks that users want to complete to understand how the application should make the tasks more efficient and effective.

(2) Top Tasks analysis: Your application can't do everything for everyone all the time. Most people use applications (software or websites) only to accomplish a few tasks. Investigate your users and find out which key tasks can meet most of their needs most of the time. Colleagues should ensure that your application can accomplish these tasks well.

Design and development

How does the interface look?

(1) wireframing: In the early days, you can draw the main elements of the interface with pen and paper, Visio or PowerPoint. This is enough to understand functions, processes and find opportunities for improvement. This will allow you to get the design before the stakeholders.

(2) Prototyping: Improve the fidelity of the design and test it as soon as possible and often.

Testing and evaluation

1. How to organize?

(1) Card Classification: How to solve your functions, screens and abstract concepts? How do you organize them? Don't guess, let users classify these items and name each category.

(2) Tree test: Only using abstract classification, let users try to locate items on navigation to test the wireframe and prototype of navigation.

(3) First click test: If users go the wrong way, they are more likely to get lost and fail in the task. Understand where users will start.

(4) Keystroke-level modeling: Without testing a user, you can also know whether the task completion time or the proposed improvement has increased or decreased the completion time. The KLM method uses some core HCI rules to estimate how long it takes a skilled user to complete a task.

(5) Heuristic evaluation: Find problems before bringing them to users. Heuristic evaluation can find about 30% of the problems that users will encounter. Ideally, you have at least two independent evaluators to guide HCI's law and knowledge in this field. Solve these obvious problems before wasting users' precious time.

2. What problems will users encounter?

(1) Moderated In Person Testing: an ideal way to test mobile devices, or when it is difficult to provide prototypes remotely, users can be tested in laboratories, conference rooms and even aisles to find out which tasks have problems and need to be corrected.

(2) Modest remote testing: With cheap and popular services such as GoTo Meeting or WebEx, you can recruit people from all over the world to participate in the task, and even record their facial expressions with a camera. Don't just ask them what they think of the design, let them participate in the task, investigate the difficulty of the task and collect quantitative data.

(3) Model-free remote usability testing: If your design and tasks are clear, you can test your prototype online, and users can participate in tasks remotely without face-to-face. You can even use hotspots to test images. Using user zoom, Usertesting.com, Loop 1 1 and other services to realize structured tasks and ask specific questions, you can get data results from more than 10 to hundreds of users in one day. Test like this, and then test again.

Development and online

(1) usability benchmark research: Let a representative group of users participate in the task, and you can know the usability of a website or software. Collect quantitative data and use confidence intervals to obtain reliable benchmarks. Standardized questionnaires can also be used after testing or research. These can be implemented in a laboratory environment or remotely.

(2) Unmodeled remote usability testing: You can use an online website to let users participate in the tasks you determined in the critical task analysis and configuration design stage. You can record clicks or even record the whole process to see what troubles users will encounter when you are away.

(3) Benchmarking research: How difficult is it for users to complete a task by using the competitive products you identified in the demand stage? Recruit users, and use factors such as success rate, time and task difficulty to examine the advantages and disadvantages of the website. Sometimes, the best comparison is the best website that provides similar services in different industries. If you are selling your mobile service, consider comparing the checkout experience of DirecTV or Zappos.

(4)A/B test: Don't guess, test. Design and improvement didn't end after you released the product. Test forms, buttons, copies, pictures and prices. Don't be afraid to try wildcards.

(5) Multivariate testing: test one variable at a time to fine-tune the website, but if you want to test a lot, it will take a long time, and you don't know how the two elements interact. For example, when you combine lower prices with different products, unexpected things may happen. You can conduct multivariate tests on an online website, or use attitude data instead of actual purchases to simulate this experience in a research and development environment.

(6) Survey: Will your users recommend your website or products? Do they trust it and find it attractive? Compare your score with industry benchmarks and use standardized questions. Ask users for suggestions for further improvement and link open comments with quantitative data.