Help Information Architecture | User Category System Construction

transitioned from complex models to concise thinking 2022.02

Apart from building a user-friendly help path (discussed in another project summary), continuous optimization of problem-solving capabilities is crucial. The solution system must meet three main requirements: findable, understandable, and solvable. The first step is prioritizing findability.

This article delves into the reconstruction of the problem classification system from the user's perspective, driving collaborative efforts between business and product teams for optimization.

01. Problem Analysis & Goal Definition

1. What is a category? What purpose should it serve?

From the perspctive of cognitive science, our understanding of something begins with categorization, akin to storing information in existing mental drawers. Rational categorization assists in hierarchical maintenance of content.

For instance, tables and chairs at home fall under furniture, roadside grass and flowers are plants, and humans form families, enterprises, progressing to cities, countries, Earth, and the universe. The hierarchy spans from parent classes to subclasses, from macro to micro, akin to an infinite nesting doll.

In the context of internet content, there are two main categories: user-facing frontend categories for searching and backend categories for maintenance and organizational purposes​​.

 2. The challenges and pain points of categories are twofold

From the user's perspective: the platform needs to understand how users instinctively find reasonable classifications that align with their cognitive habits.

On the backend: establishing a mapping correlation between backend and frontend categories is based on organizational roles, and the optimization goal is to create a seamless connection between them​​.

02. Extracting the Main Thread from the Disarray

Issue Inventory:

1. Try to isolate and look at the larger main line, so that the information elements

The help product does not exist independently. It is a mapping of the user's frontline main link.

This point of view is easy to obtain, but sometimes when we work, we only work on this part and neglect the overall overview. The help product is not independent, it is absolutely affiliated with the front-end process/product/platform being helped. It is easy to understand from this point of view. For users who want to help merchants open a store on Taobao and Tmall, the help platform is the way for them to seek help when they encounter problems during their business operations.

For example, Alipay’s consumer assistance is organized according to life scenarios (travel, financial management, payment, etc.), while Meituan’s consumer assistance framework is distributed according to large local service types (beauty salons, in-store food, takeout services, etc.).

Based on the main link classification of front-end operations as the benchmark, we will refine the large nodes that help adjust and serve as the basis of the first-level category.

2. Model of information element organization

Now that the link elements are there, what perspective should be used to link them, which is more consistent with the logic of help? Here, three sets of information model frameworks for merchants are summarized.

3. Sort out the main lines as a whole and obtain the preset baseline classification.

Use the [User Service Journey] method to visualize the content when multiple parties collaborate to sort it out, making it easier to understand and reach a consensus.

As a result, we can obtain a merchant category classification system that is (relatively) consistent with the user's perspective.

In fact, it is enough to stop here. If you are not sure and want users' research input, you can continue.

03. 用户视角的补充输入

03. Supplementary input from user perspective

1. From core questions to formulating an outline plan for user research:

In the process of obtaining basic and reasonable user categories, we have several core questions that require in-depth analysis with the help of user survey input:

The cross classification is reasonable in several categories: Is Taobao Live broadcast content or marketing? Is the refund rights protection an order or a rights protection?

The description and expression of the copywriting are sometimes prone to ambiguity: market/industry? Marketing/promotion/traffic? Order/transaction/after-sales?

Therefore, we conducted research in the form of [Card Sorting (Hybrid)] + [Interview].

Card sorting related information:

https://www.nngroup.com/articles/card-sorting-definition/

https://www.usability.gov/how-to-and-tools/methods/card-sorting.html

Based on the proportion of Bao Tmall users served by Taobao merchants, we selected 16 interview users (including new and old merchants and role division distribution), and followed the following steps to conduct the card sorting test:

Let users match the second-level category to the first-level category. If there is no suitable category, do not classify it, or create a category name that they think is more suitable.

Record the reactions during the process, and ask targeted questions after the division is completed. Pay special attention to asking about hesitation, areas that are different from expectations, areas that you do not understand/cannot understand, and your understanding of some copywriting expressions.

2. Research process & results, listen to users’ voices

A week-long user interview was conducted, each user lasted about 30 minutes, with a total of 16 merchant users.

The data was sorted and analyzed:

3. Conclusion:

3.1 A reasonable first-level top-level classification can be inferred through qualitative analysis. The unreasonable first-level classification will be exposed during card sorting and interviews, and can be adjusted and verified to refine the overall granularity and expression of the copywriting.

For example, marketing is renamed: marketing/promotion, and the industry market is changed to: supply market; the new warehouse distribution logistics is independent of orders/after-sales.

3.2 Whether the mounting of the secondary classification is reasonable and whether the copy is easy to understand, we can use a quantitative method to calculate the above three indicators to monitor the health situation and adjust the strategy.

For example, Taobao Live's classified mounting content and marketing are reasonable. Dual mounting, uniqueness of data analysis & background maintenance method are solved by defining the main level and establishing multiple mappings, which can significantly improve the matching degree and be more in line with users. Cognition.

​

04. Review

This project is very special to me. It allowed me to go through the process of "gnawing" problems out of chaos. Like a beam of light, through step-by-step hypotheses and scientific verification, I continue to get closer to a more reasonable "origin". . You can summarize it with the image below~

Thanks to this experience, many program expansions derived from it continued.

The closer you get to the user, the better you understand the experience.

🤔A little thought: When doing projects, I often make analogies to phenomena in life.

Written at the back: If self-service help is compared to shopping in a mall, in addition to designing the escalators, direct elevators, and shopping paths by category on each floor, you also need to manage products: in what dimensions are the products classified into different floors, and which products should appear? Where. With the difference in overall positioning, there are differences in the supporting paths + product designs for buyer-oriented, market-oriented, and lifestyle-oriented products. This kind of general thinking is consistent with making Internet help products.