How many steps are needed to build a user search index system?

In daily life, search is one of the functions that users often use. So when you want to understand the user path in the search scenario, or how to improve the search service based on user needs, how can you monitor and analyze the search scenario? In this article, the author summarizes the relevant content of building a user search indicator system. Let's take a look.

Whether in the PC era or the mobile application era, search is one of the important entrances for users to obtain information. Monitoring and analysis based on search scenarios is very important. Next, I will help you break down the process of building a search indicator system. How can you analyze search scenarios?

In the past PC era, users would enter a search term and then the page would show them search results related to that term. Now, I believe everyone has searched for hotels on Meituan or products on Douyin. This proves that user searches have evolved from simple information acquisition to service and product transaction needs; the search format has also evolved from text to multi-modal searches such as voice and vision.

The information acquisition richness brought by the search scenario may have far exceeded your imagination. But if we focus on the search function itself, according to the idea of ​​​​layering the indicator system, the first step is to consider what are the core indicators we focus on in the search scenario:

2. Determine the core indicators of search scenarios

Traffic indicators: search DAU, search duration, number of searches, search PV, search UV, etc. These are the core indicators that can reflect search traffic. They are also the most intuitive indicators that can reflect the user's use of the search function and are suitable for daily market monitoring and analysis.

The number of searches here is generally counted according to the number of search requests initiated, or it can also be defined from the query perspective. Different businesses may have different definitions. As mentioned above, the search function has gradually upgraded to the information->people stage, so the distribution of different types of resources in the search scenario is also an indicator we monitor.

User indicators: per capita search indicators and new and active search indicators can reflect the usage of search users. Per capita indicators can reflect the penetration of search among users. New and active indicators can directly feedback users' experience with the search function and whether it can meet users' needs for obtaining information. Retention indicators reflect whether the search function gives users a good user experience. When the search function is optimized and upgraded, these indicators can provide timely feedback on the optimization effect.

Strategy indicators: For search systems or search projects, users have clear needs and convert them into a query. Users initiate requests to the search system based on the query. The search engine retrieves information related to the query from a large amount of information and returns it to the user. This process involves many algorithms and strategies such as query intent analysis, search result recall and ranking, text matching, causal inference, etc. When upgrading and optimizing search functions, we need to analyze algorithm strategy indicators in addition to considering user experience indicators.

2. Determine the analytical dimensions of the core search indicators

Depending on the type of content obtained by users, search types can be divided into information searches, transaction searches, and service searches .

On the one hand, search applications aggregate search information and services together, and on the other hand, current search portals tend to display and provide users with vertical information, so many search application scenarios have been added. We have already divided the search scenarios into some measurable and quantifiable indicators above. Now we will also divide these various search application scenarios.

1) Analysis by industry

Analyzing different industry classifications can provide feedback on users' different types of search needs. Search services should build life service spaces for users in addition to meeting their daily needs of food, clothing, housing and transportation, as well as medicine, entertainment, and education. They should meet users' full-content, full-category search needs in all scenarios and seek commercial monetization opportunities based on these industries.

2) Analysis by search type

The significance of analyzing different search types such as active search, search history, recommended search, similar search, etc. is that in addition to the differences in the search methods perceived by users, they also represent different search strategies and algorithms, which are of great help and value for the optimization of the search system.

3) Analysis by channel

Now the search application scenarios are no longer limited to PC Internet and mobile Internet. New hardware applications such as smart speakers and car systems have also opened up wider channels for us to reach users and provided a variety of marketing methods and delivery strategies.

4) Analysis by region

Analyzing regions can help us understand the proportion of search users in first- and second-tier cities, as well as the penetration of lower-tier markets. By providing different search services for different regions, we can analyze the differences brought by the regions and conduct refined operations.

5) Analysis by age

Analyzing the use of search functions by age groups can help us understand the interests and differences of users of different generations in search services. For example, we can launch applications with clearer pages and functions for the elderly, and distribute more diverse search information and services to young people to explore consumer needs.

6) Analysis by time period

The time period dimension can better reflect user usage habits, is easily affected by hot consultations, and can enrich our user portraits.

7) Analysis by login status

The login status can be used to analyze the scale of registered users, the effect of login conversion, the quality of user registration and login, and the differences in the behavior paths of users who are not logged in and those who are logged in.

8) Analyze by retention, new and old users

Active users in a certain period of time include old users and new users, so when analyzing active users, you need to consider the ratio of new to old users, such as the ratio of (daily new users) / DAU, to see the proportion of new users in the total active users. New users are attracted by attracting new users, while old users need to be retained. These two dimensions should be analyzed in combination.

3. Dismantle and supplement according to business processes

This link requires the most granular decomposition of the search path, and the consideration should be combined with the delivery channel and marketing link. The basic path is that users obtain information through search to assist their own decision-making, and then obtain the services brought by the search; or they are stimulated by the interest of information flow content and trigger the search.

If the search scenarios are complex or overlapping, you can focus on the main application scenarios where users use search, first sort out the usage links of key scenarios, and set indicators for monitoring at each link. In this way, we have built the core search indicators above, and then we will break down the user path and associate the segmented business indicators with the core indicators.

From the perspective of user needs, although users have more sources of information, search function is still an important solution to user cognitive needs. The core competitiveness of search now should think about how to output high-quality content and accurate search results, and provide diversified services in different scenarios in combination with search needs. In the face of new content, new scenarios, and new forms, search functions and services are also constantly upgrading, so building a search index system is very necessary for daily analysis and decision-making. These are all user indicators, and there are many technical indicators such as search speed and search stability.

The content listed in this article is not the complete and only answer to analyzing search indicators. There are still many contents that have not been shared one by one. Friends who are interested and have ideas are welcome to communicate below~