Product Information Architecture: Internet Product Path Planning

Article Description: several stages of product path planning.

Compared to the previous blog on the publication of their own entrepreneurial projects when the data, and after 3 months, the total number of the kitchen under 3 times times more than the registered user 2.6W, the independent user is also more than 2W per day. Of course, our team has doubled 4 times times compared to the previous 3.5 people.

Before starting a business to write a PPT, a general description of the next kitchen path planning, so far, basically according to the path of the development of ideas:
1 The first stage, mainly relies on "the feeling" to drive the product. (In the knowledge that once answered a question can explain the "sense" of the source) if it feels right, should soon be able to do 3W independent IP per day (previously estimated 6-8 months, and the status quo is basically consistent)
2 The second stage, should rely on "data" to drive the product. We believe that the existing user behavior, is already a relatively representative sample. Validating ideas with data is the next most important thing.

In the evening, my venture partner Subdragon to discuss a question: is the recipe information a vertical area long enough to end?
Subjectively there is a guess, a user needs only 100 recipes, and each user needs these recipes overlap very high, then, the next kitchen is only to let the editor to tidy up 3000 recipes, enough to meet 80% of the public users of the menu needs?

If that is the case, it may be necessary to overturn our previous information architecture.

We decided to use the distribution of the user's collection to verify the long tail, and in order to spark the discussion, we made a bet:
1 Subdragon feel in the collection of recipes, the top 10% of the list of recipes number of collection, accounting for the total collection of 90%;
2 I think this curve is not so steep, at least between the first 20%-30%;

Actually ran down the data:
1 down kitchen near 2.5W recipes
2 Total collections in the 20W (per user average of 7.6 courses)
3 of the recipes that have the collection behavior is 1.8W (far beyond our imagination)

The fact is: the total number of recipes in the top 50% of the list, which accounts for 90% of total 20W (long tail is scary)
Of course, we feel that the actual promotion and operation behavior, let the data generate some noise. But presumably this will not affect the conclusion.

Under the kitchen of the current user base, I am sensitive to understand also more concentrated in the "food lovers", compared to the public users or "high-end" (neutral word, easy to understand):
1 look at the current daily update of the work quality:http://www.xiachufang.com/explore/dish/time/ (average level is beyond the homely range)
2 Take a look at the user active site search word top (the last 15 days), extract 23 Direct Search the name of the recipe (I organized a menu to facilitate your one-click Collection) They are:

braised pork , cola chicken wings , shredded pork with fish flavor , double skin milk , eggplant with fish flavor , mashed potatoes , cooked pork, Kung Pao chicken , sweet and sour pork ribs , egg Fried rice , boiled fish , fried noodles , boiled pork , pickled cabbage fish , eggs , Preserved egg meat porridge , mapo tofu , ground delicacies , pot-wrapped meat , potato cake , braised eggplant , tiger skin green pepper , Pumpkin pie

At present the kitchen product progress only completes I expected 30%, the basic function also needs some time complete and the iteration, then makes the cool some product to come out.
Then, following the original plan, there are two remaining stages:
1 Market Drive phase (1 years later 10W independent access)
2 Commercial Drive phase (2 years later 100W independent access)
I hope we can live to that time, haha, also expect in such a vertical field, can make some innovation, I still firmly believe that China's food culture: first, the demand and foreign have different; second, the market should be bigger.

Long time no blog, I also know that we do not like to see the theory, as far as possible to share some similar combat data, case bar.
(Welcome everyone to reprint, help bring on the chain, thank you slightly)