【 Interview questions 】
Several functions are planned for the launch of a live broadcast app (including short video, subscription of game equipment for anchors, jump in the live broadcast room, playing with big gods, etc.).
How to determine functional priorities scientifically according to user needs?
【 答 案 】
1. Understand the service scenario
In everyday business, requirements prioritization issues are common due to limited company resources and developer time and effort.
Generally speaking, the function sequencing needs to consider the company, product, user and social effects.
What do you do when you have this kind of analysis prioritization problem?
The corresponding analysis model can be selected according to the combination of different influencing factors.
To take a few examples, the common 2d analysis model (functional importance + urgency)
WSJF model (Boss + Strategy + Performance + Demand score + time to go online)
The interview questions here require prioritization based on user needs, so the Kano model is chosen.
2. Understand analytical methods
(1) Model introduction
The Kano method, developed by Tokyo Institute of Technology professor Noriaki Kano, divides user-oriented needs into five categories: attractive, desirable, necessary, undifferentiated and reverse.
Different types of demand correspond to different stages of product quality:
Therefore, in most cases, the product needs to first meet the necessary needs of users, then expect the needs, and finally exceed the expectations of customers through attractive needs.
So what are the required/expected/attractive/undifferentiated/reverse features?
Essential features:
Features that are expected of the product. Failure to do so will significantly affect the user experience.
Expected functions:
Users expect the app to have this kind of function, which can improve user satisfaction after launch. Otherwise the user will not be satisfied.
When this feature (such as online courses in recruitment apps) is generalized, the expectation function may be relegated to a required function.
Charm features:
Usually something new/emotional.
Even if wechat pat-pat function is not online, it will not affect user satisfaction; After the function went online, it aroused the enthusiasm of users for interaction. Of course, as time goes on and the novelty wears off, the charm feature may degenerate into a desired/must-have feature.
No difference function:
Having or not having it doesn’t change customer satisfaction.
After the popularity of virtual fruit-growing games on e-commerce platforms, some apps in other industries directly copied this function. As the platform attributes and user needs are not considered, most users directly skip this section.
Reverse function:
The reverse will result in less user satisfaction.
Such as maliciously induced open screen ads, through the blurring of the actual “skip” button, to entice users to click on the AD. For the product side, advertising revenue may be increased, but it affects the experience and satisfaction of most users.
(2) Model function
The Kano model can answer interview questions by classifying features as necessary/desirable/attractive/no difference/reverse and determining their priority.
At the same time, the model needs to survey users, and the product department participates in the questionnaire design and research, so it can promote the consensus of the team. The research process helps users have a voice.
(3) Model steps
Corresponding to the role of the model, the use of the model requires the team to jointly set up a survey questionnaire and let users express their needs through the survey process.
Then the questionnaire results were collected and the requirements were classified as necessary/expected/attractive/no difference/reverse by Kano coefficient.
Finally, the analysis results are visualized, and conclusions are drawn, evaluated and adjusted.
[Specific operation]
1. Questionnaire survey
(1) Design the questionnaire
Kano model needs to investigate user satisfaction when each function is available/not available.
For example: do users like/think they should/don’t care/reluctantly accept/don’t like the fact that livestreaming platforms offer the function of playing with friends? How do users feel when livestreaming platforms do not provide short video functions?
The questionnaire questions and choice wording need to be confirmed with the business department and how to explain the questionnaire content to users.
The content includes forward/reverse questions and options. The explanation part of the questionnaire is for the researcher’s own use.
(2) Determine the sample
Determine channels:
When the content needs to be explained in the survey process and the survey sample is not too large, offline questionnaire survey is preferred.
Sample size:
To ensure statistical significance of the questionnaire results, the sample size is required to be ≥30.
In addition, some sample data may not meet the requirements, so it is necessary to reserve fault-tolerance, and ensure that the actual number of users is more than 35/40.
Sample requirements:
Sample requirements need to be communicated to the business unit. Is this survey for all target users of the product? Or core users? Or multi-user?
If for all product users, a random sample survey from product users;
If the core users, according to the requirements of core users, after screening sampling survey;
If multiple types of users are used, ensure the proportion of samples of each type. If the product has 1/2/3 user groups, the samples can be divided into 30 (user group 1 sample) +30 (user group 2 sample) +30 (user group 3 sample).
(3) Filling in the questionnaire
After completing the questionnaire design and confirming the survey sample, invite users to fill in the questionnaire and pay attention to the filling process of users.
2. Result statistics
In order to show the results more fully, we have added some feature items. The statistical result is the user satisfaction data with/without the following six functions.
The final results of the questionnaire were calculated in two-dimensional form:
The “16” in the box below represents: out of 120 samples, 16 users like the feature and are fine with it if it is not provided.
13.3% indicates the ratio of 16 users to 120 users.
Color functional statistics with red for attractive qualities, orange for desired qualities, green for required qualities, blue for undifferentiated qualities, purple for reverse qualities, and gray for doubtful data.
Finally, the total number of charm/expectation/necessary/no difference/reverse qualities of the play god function is summarized.
In the same way, aggregate the quality data for subscription game gear/studio jump/privileged gifts/anchor PK/ short video features.
At this point, the statistical work of the results is complete. To summarize the steps of result statistics:
3. The visualization
Despite the statistics, we have not yet determined the prioritization of features.
A better- Worse factor (an important part of kano’s model) is needed to determine user satisfaction with and without each feature, aided by visual diagrams.
(1) Determine the Better-Worse coefficient
Better coefficient and Worse coefficient represent the satisfaction level of users with or without this feature.
The Better coefficient is closer to 1, indicating that the Better function can improve user satisfaction more obviously. The worse coefficient is closer to -1, indicating that the function is not available and the decrease in user satisfaction is more significant.
Calculate the Better and Worse coefficients of each function:
(2) Statistical map to determine functional zones
Taking the absolute value of worse coefficient as abscissa and better coefficient as ordinate, the function scatter diagram was drawn and the quadrants were divided.
The first quadrant is the desired function, including the privilege gift, play with the gods function.
The second quadrant is the charm function, including live broadcast jump, subscription game equipment.
The third phenomenon is no difference function, including anchor PK.
The fourth quadrant is a required feature and includes short videos.
Conclusion:
Generally speaking, essential functions are preferred over desired functions over attractive functions, and no difference functions can be considered according to the situation of the product.
In the same quadrant, the Better coefficient is closer to 1, the improvement of user satisfaction is more obvious. The closer the worse coefficient gets to -1, the greater the decline in user satisfaction. Therefore, in the charm function, live broadcast jump takes precedence over subscription game equipment; In the desired function, the privileged gift takes precedence over the playmate.
Essential functions are usually the basic requirements of users and should be met first.
Expected functions are the focus of users to compare products of the same type, and should be satisfied as far as possible on the basis of meeting necessary needs;
Charm features are less predictable, but can lead to user loyalty.
In real business, the definition of functionality needs to be constantly updated. With the development of the market and competitive products, the expected function and charm function may evolve into necessary functions, and the product side should explore new user expectations and surprises.
4. Model tuning
(1) Combine other user information to enrich the conclusions
Kano’s model information comes entirely from user questionnaires. In fact, the questionnaire results cannot fully represent the real needs of users, and what users say is not equal to their internal thoughts.
Therefore, when prioritization is determined based on user needs, kano analysis results can be combined with product user portraits to determine the final conclusion.
The user profile contains basic information about the target user, usage scenarios, pain points, and key differentiators. This helps determine the user’s internal desires.)
(2) Kano modeling of daily data
As mentioned earlier, the definition of functionality needs to be constantly updated.
At the same time, the coverage of questionnaires is limited, and a large number of questionnaires need to consume more manpower. Therefore, using python and other software for daily data modeling can help long-term analysis.
In this paper, we mainly introduce the daily use of Kano model, and do not continue to expand the modeling analysis.
【 conclusion 】
Going back to the previous interview questions, how to scientifically determine functionality priorities according to user needs?
In this paper, we determine the attributes and priorities of each function through the four steps of Kano model (questionnaire survey – results statistics – visualization – tuning) : essential > expectation > charm > no difference function (reverse function is the product to avoid).
At the same time, in order to make the conclusion more objective, the conclusion of Kano analysis can be compared and analyzed with the user portrait information to comprehensively determine the user’s internal needs. For long-term consideration, the Kano data model can be built step by step, and then function categories and priorities can be determined directly through the comment data.
Recommendation: Data analysis thinking everyone needs
