Passive initiative--talking about the recommendation function in the product

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

[Core hints] behind the recommendation should not just show the content to the user, let the user passively accept, if all is useless things, the result will be worse.

While listening to music on the road, I always used to cut the song in the "Random play" mode until I cut to the song I wanted to hear. I do not know similar scenes have "experience", is my personal habits or people have such a feeling, so want to write out by the habit of some of the recommendations on the recommendation of some feelings.

Behavior: Select "Random play" always want to cut the song to "want to hear the song"

Random playback

From my habits of use, "random play" occurs in commuting or just busy a little relaxed, choose "random play" means "listen to the song" This is too big Listen to what song ", belong to behavior drive big too target drive, resemble hungry want to eat but haven't thought good what hotel, want to see a movie but didn't decide to see what movie. Just "Listen to the song" as a relief rather than a good appreciation of music.

Under this premise, if lets the user first go to the song chooses, builds the list again to start to play what, the phase must not be so joyful, just wants to listen casually.

Want to hear the song

In the "Random Play" encountered a song do not like how to do, click on the "Next" button on it, if you do not like how to do, continue to the next song. So why not go straight to the songs you like to listen to and switch in this way? A diagram that you can see:

In choosing the song that you want to listen to, the user's mental model is probably like the picture, in so many songs to actively look for the song (blue point), if the target (blue dots) is very large, then it is easy to find, but there is no particularly strong target (blue), to pick out the songs you want to hear, it is not so easy, It's easy to choose difficulties.

In the "Random play" state, the song sequence is linear, the player will actively push the song to you, do not need to think about the next one is what, just the current song you want to listen, do not want to listen to "cut", and then for you, in the passive "random play" still hold the initiative.

Active in passivity-personalized recommended products

I describe the behavior just described as "passive active", the passive is that information display content and order is not the choreography, but the initiative is to change the content of passive rendering information through some operations. This kind of personalized recommendation in the form of people in peacetime already can be found, these will certainly will be the future trend.

Recommended Music Products

The "Random play" mentioned earlier also has good music products, such as foreign Pandora and the domestic watercress/shrimp radio stations, users do not consider what song library problem, the product will first through the hot or label to confirm your initial taste, and then through your active to the song for more detailed operation, Constantly digging out the appropriate songs for your recommendation algorithm, increase stickiness.

Recommendations in the electrical business

Amazon.com's home page is very famous is the function of personalized recommendation products, in the recent revision of the domestic electrical business also found that the position of the recommendation has been more and more heavy, these are "passive", and " Active "is the user in the previous browsing of goods, attention to things, businesses through the data mining to make effective recommendations, the transformation is very helpful.

The recent fire of Mushroom Street & Beautiful said & Fruit Library belongs to the shopping guide, the initiative has not seen out, but compared to Taobao's huge, has been regarded as a precision recommended.

Recommendations for content classes

Watercress in this piece is not intended to do in the domestic relatively good, reading channels in the watercress guess there are "guess you are interested in the new book","Guess your reading interest","guess you are interested in the Bean row", book/Film on the single page has "like xxxx also like" The following beans recommended ", these recommendations can activate the whole watercress content system, intertwined with each other into a net, and the production of recommended information is the user's active access to produce.

Sina Weibo's "Recommended","right column recommended","micro-relationship", respectively, according to Weibo popular, label & relationship, focus on the analysis of recommendations, but also to activate his entire membership of the relationship between the existence.

There are similar peas guess, the app Store Genius is the same, not one by one example.

Passive & Active & Recommendation

Recommended can reduce the threshold of use, can help the rapid transformation of goods, more able to pull the link between the content, users will need such a function, but behind the recommendations should not just show the content to users, so that users passively accept, if all is useless things, the result will be worse. More to give some users the initiative to control things, so that users do not know the circumstances of the participation in the recommendation, so that the recommendation more effective and reliable.