An Information Service Design Project for Developers

*Considered information safety, the details of the product and research information are hidden

Project Background

The demand from the client team

“There are many complain emails from customer saying the ‘product‘ is difficult to interpret, and the overall experience of using the ‘product‘ is too complex. Customer often had to ask the sales engineers manually set up the ‘product‘ and environment, which is time-consuming and frustrating process for both”

The root of the problem

The company is promoting a revolutionary software solution to connect software and hardware product lines in AI industries. The product will benefit developers of key customers as well as technology partners. In order to build an ecosystem of product solutions, the company is willing to teach programmers to use exclusive programming languages and new software and encourage them to contribute to the ecosystem for business growth.

However, understanding the product and using the new software requires a profound technical background in Deep Learning, Neural Networks and fundamental knowledge of how software is implemented into hardware. An easy-to-interpret platform service with UX strategy is required in the current design situation.

How can we incorporate the business strategy and technical product information into a user-friendly platform service and reduce customer complains?

The Deliverables

• inspect the critical UX issues from the current commercial website and learn tutorial lessons. The team needs to outline a measurable UX baseline (#of pages, clicks, transitions. etc)

• the design team is required to deliver a professional, thorough user research report

• the design team will offer design proposals to the corresponding product owner and their in-house UX designers

• Supervise the front-end development outcome that matches the proposal and provide feedback. (Long-term)

Industry Research & Business Strategy

Before heading into the design phase and began resonating with the users, we had a glance at the industry as a whole.

Business

• how the technology will affects the practical use scenario, the possibilities of the product application

• why the business strategy is shaped this way, which portion of the strategy are we looking at

• what are the current use cases, that can be used to narrow down the target audience

• what is the trend of the industry, and how can we relate the product to the trend through information design

• …

Technical (the most stressful part for me)

• what is the general process of NN DL development, and what is the relationship between the software and hardware

• what process and knowledge are required to use our product: ML DL NN, how does the hardware work with the software

• knowing the skill sets to become a developer in the industry

Competitive research

Pros & Cons

Comparing and examining the detailed technical advantages and differences:

• the programming language difference, the cost of learning, the overall working and installation steps

• the hardware differences, which hardware runs, faster, in what scenario, and what happens when it faster

• the comparison of the overall customer experience, user feedbacks, ratings

Figure out:

• the related service that had a major impact, in this case, is the official website’s promotion page and the tutorials.

• what technology advantages do the competitors not have that we need to promote

• what are the cons that require UX or other services to interfere and offset

Target Audience

We narrowed it down to a limited range of users at this stage.

We ranked the audience groups in the most profitable order, considering the following:

• the hardware demand (most profitable for the company), who will use the hardware for sure

• who uses the software as a supplementary

• rated industry, what industry are they from? what information emphasis on them

• technical scenario, what development process will or have the potential to use the new product (in this case, the scenario has a linear relationship)

Profitability to the company is a long-term goal, in the current situation, the short-term goal is finding the groups that have the potential to be profitable and have a higher intention of making a contribution to the ecosystem.

User Research

User Interviews

According to the business strategy, we have selected the secondary user group to interview and the most relevant programming scenarios for the product. Because the selected technical scenario is the only process producing the needed component of the ecosystem.

We focused the research on the following purposes:

1. Resources and feelings: Developer's learning process, resources channels in different work/learning stages, mood, and confusion.

2. Standardization and consistency in the workflow: Consistency of use of services, tools, languages, plugins, wiki documentation, etc.

3. Calculation process before/while coding: The process of calculation and analysis, what are the information or environment they needed.

There are two subgroups in the secondary audience:

• Grad school student developer: develop for self, entry-level programming in DL NN

• In-house developer: develop for business, academic programming in DL/NN

• Student developer: 6 people

  • Gender ratio: 5M 1F

  • Years in grad school: 1-3

  • The number of projects developed in the ‘product‘ :

    • 3 people with one project experience

    • 3 people with more than one experience, up to 3

  • Method：

    • User presentation (understanding the work routine of the targeted user)

    • In-depth online interview (Observing user flow and pain points)

• In-house developer: 6 people

  • Gender ratio: 4M 2F

  • Background with DL: 2

  • Background with hardware: 2

  • Other backgrounds: IoT 2

  • The number of projects developed in the ‘product‘: generally more than 3, up to 5

  • Method:

    • In-person interview (Asking questions and finding pain points & observing the physical environment)

    • Focus group (with similar questions of one-to-one interview)

Interview deliverable 1: Personas

Interview deliverable 2: Workflow & Key activities

Identify the workflow, key activities and the estimated time frame in each phase

User Journey map

Key Findings & Insights

Pain points card sorting

Card sorting and prioritizing the pain points in services

Spotting the service

Inspect Into Information Architecture

We went through the service, then broke down website service into 3 major scenarios:

1. Kowing

2. Learning (website tutorials & internal documentation)

3. Installation

1. Knowing

   • The main channel for users to obtain core product information during the product promotion period, which can influence users' impression of the product and stimulate the demand for further understanding.

   • The main issue is that there are fewer product value descriptions and incentives, and the information structure and language narrative are confusing.

2. Learning

   • Learning and conversion scenario-based helps users form a systematic, sustainable learning experience. It is an important stage to motivate users to continue learning and exploring the product.

   • The main issues arise in establishing user hierarchies, providing diverse learning paths, strengthening user incentives, and increasing missing information and help channels. At the same time, when entering the use phase, ensure that there are enough code examples and logical and fluent guidance.

3. Installing

   • Installation and deployment is the first step when the user decides to start using the product. The installation environment of the product needs to be built according to the user's own equipment, especially for different operating systems or hardware had very complex installation approaches.

   • The main problem is that users often download the wrong package version because the guidelines are unclear and missing information.

4. Using

   • Inside the product

Experience Baseline

Based on the existing website user flows, I listed all the entrance and destination pages and counted the measurable experience baseline for the “before & after“ purpose

Design Solutions

The design process under each scenario.

Scenario 1: Knowing

Design Goal

Strengthen the delivery of practical functions, increase the amount of information from the user's perspective to attract new users, and strengthen the content with the ecosystem, attract old users from related sofware/hardware products.

Design Solution

Scenario 2: Learning

Emphasize & Define

Problems discovered from user research

• Developers are not sufficiently motivated to complete the project, they just wish to get tasks done.

• Course resources are relatively old, the latest tutorial is not available in time, resulting in the coding stage user not knowing how to proceed.

• There is a lot of terminology in the introduction of knowledge points, which is obscure for newcomers.

• There are only a few code samples and limited use cases in each curriculum, and it is difficult for programmers to reference.

Problems discovered from information structure

• Course index has no navigation:

  • The course lacks of user-differentiated design

  • new users need to know the specific course name to find it along with other courses

• Service title and learning objectives are not clearly described:

  • users need to click into the course video page several times to view the learning content, which is time-consuming.

  • course catalogue’s information hierarchy is not unified, it is difficult for users to understand the course outline.

• Incomplete coverage of practical lessons:

  • The curriculum only covers fundamental methodologies. The cases are too basic and simple.

  • The current course belongs to the confirmation inquiry type in the learning inquiry model, which means the instructor demonstrates the whole process students follow.

Design Goal

In order to further match the pain points and needs of different user groups. We need to establish user stratification standards to meet the differentiated needs of different groups from information.

Highlight the learning advantages of the platform, provide a more substantial growth path, help users choose the most suitable learning plan and supporting resources, and guide the content courses learned before and ongoing course learning without barriers, and establish a suitable user high-quality learning experience.

Ideate & Design

The Users

Back to the target audience table, with selected technical programming scenarios, we extracted and expanded the information from the previous table. We list the professional skills and knowledge they need to achieve their goals.

This table will help the curriculum and documentation information architects to collect information more accurately with the business goal.

Because we are designing a learning process, after knowing what each user group may be interested in, we need to transform the business-targeted audience into the role of students to understand what their learning demands are.

We permute the users through two models:

• Initiative (learning motivation): Self-determination theory in psychology

• Professional background (learning needs): Bloom’s taxonomy in pedagogy

We categorized the “students” into：

• Active learners: who need service upgrades to optimize the learning experience. They require an easy-to-understand learning process (the experience with incentives is optional).

• Passive task completers: They need to improve learning motivation through external motivation, and they wish to learn how to implement the code fast. (the experience with incentives is better).

• Professional field deployers: They need to learn the specific practical methods to code in specific professional fields.

• Professional leaders: They don't need to use our learning services, we want them to interact with other types of students.

We prioritized the most problematic students(passive task completers) in the design solution.

The Solution

In the Emphasize & Define stage, we summarized 4 leading issues, then with the information above we provided corresponding solutions for each problem:

We have involved few strategies for page’s content design:

We distributed the strategies in designated pages with new features as the final design solution:

The final design proposal of the “Learning scenario” results in the flowing experience optimization :

• reduced the total page transition from 45 to 6

• reduced the mouse click from 101 to (about) 40

• unique page reduced from 39 to 8

Scenario 3: Installing

“ I thought the “product” installation was the most difficult part… well, it was not difficult, but it was very troublesome... You will encounter various problems, the environment can eventually be set up, but not so efficiently. To be honest with you, when I know the steps of setting up the environment I already wish to quit. ”

By an anonymous user

Emphasize & Define

Problems discovered from user research and technical forum:

According to the technical forum's newest 100 user queries from March to April, 15% of problems are installation failures.

• The package version is not updated or an error occurred after the update.

• XXXXX(an architecture) does not match the software package and causes bugs.

• The installation in XXXXX(tool) always fails and reports an error.

• Installation failure caused by many XXXX(product) hardware environment problems

• The installation fails due to the mismatch between the computer operating system and the software version

• Unable to download using XXXXX(installation tool)

• Unable to confirm the installation path of XXX(the product)

• Other bugs with unknown reasons

Problems discovered from information structure

• The download and installation information is isolated from the developer documentation, and it is difficult to find the entry.

• Document information is scattered on different pages that are not friendly to new users, and the navigation is unclear.

• The information path is nested, causing a time-consuming user installation process. For hardware-related installation information, you need to find documents on other products’ websites.

• Many pages only have a few links for the document directory, most of the pages are visually blank.

• There is too much information on the hardware-related installation instruction pages with inadequate narratives.

Design Goal

The goal is to improve user installation efficiency, reduce document browsing time costs, and improve user experience by:

Filter the installation package through different user's equipment or software environment, provide accurate and fast product download process, one-stop installation design.

Disassemble the software/hardware environment setup process, provide modular path assembly guidance instructions through a reasonable narrative method.

Ideate & Design

One of the most important steps before designing is to sort out the relationship between all product installation-related information. (I spent one month sorting it out) The installation package can be filtered with multiple factors, but on the current website there is no navigator to connect them.

The Solution

We have divided the current installation process into 5 decision points:

1. Decide to install and proceed to the next step

   • to provide users with a noticeable visual guide to the entrance of the installation path.

2. Users acquire software packages on demand

   • Customized resource aggregation, filtering out installation packages based on the user's device software/hardware environment.

3. Deploy the physical environment

   • Exclusive service provided for hardware customers, with the product information can filter out the packages for hardware.

4. Install software packages for hardware

   • Quickly filter out the software packages that match the software and hardware environment configured to the hardware product.

5. Install environment dependency to run the product.

   • Provide guidance and quick links to install different dependencies for environment preset

The final design proposal of the “Installing scenario” results in the flowing experience optimization :

• reduced the total page transition from 88 to 20

• reduced the mouse click from 87 to 32

• unique page reduced from 69 to 21