Thinking about the way of sorting commodity list

The list of items here does not contain a list of search results in the site, sorting of product search results, and content rendering, please refer to my other article: The search in the Business-to-consumer station-sorting and content rendering.

In traditional retailing, the display of shelves has a huge impact on sales. How to reasonably allocate limited space, it is particularly important to attract customers through promotional items, thereby promoting the sales of the largest profit products, cleaning up unsalable goods and inventory goods. Online retailing Although there is no shelf and space restrictions in theory, but the opportunity to display goods is always unfair and unequal.

Someone summed up the 28 rule of sales of the product, 20% of the goods contributed 80% of the sales, but many sites may be more extreme, 10% of the following products contributed 90% of sales. One of the reasons for this result is a serious imbalance in the display of goods.

In conclusion, the following factors will directly affect the opportunity to display the goods:

The landing page caused uneven flow distribution.

There are usually 3 main landing pages for the Business-to-consumer Web site: Home page, topic page, and Product Details page, and the flow tilt caused by artificial subjective recommendation, SEO and ad delivery leads to too much opportunity for some products to be displayed, while more goods lack the opportunity to display.

An imbalance in the opportunity offered by the smart recommendation system to display goods

Whether it is based on content filtering or collaborative filtering algorithm, the indirect amplification of the product display uneven situation.

In-site search offers uneven display opportunities for goods

The accuracy and recall of search in the station, and the sequencing of search results will also affect the display opportunity of the goods.

Navigation, classification, list of items to give goods to the display of uneven opportunities

This is the focus of this article to explain the direction

As the general users of the Web site through navigation, category browsing to reach the target product page is a huge proportion, we mainly talk about the order of the list of goods.

Most Web sites currently offer the following sorting methods:

Sort by sales

Sort by popularity (by number of views)

Sort by new (by shelf time)

Sort by price

Sort by number of items commented on

These sorts are mostly one-dimensional, and the system defaults to a sort (and, of course, manual sizing), while supporting the user's freedom to choose other sorting methods.

The problem of ordering a single dimension in a product list

1, the single dimension of the sorting method easily caused by the "Matthew Effect" of commodity sales, such as by the order of sales, the beginning to sell good goods increasingly sell well, a bad start to sell goods and new products will be difficult to have the opportunity.

2, if the default according to the order of sales can basically reflect the needs of users, then according to the new order can only be regarded as the Web site's wishful thinking, the old customers will pay more attention to your site's new products.

3, sorted by price and ranked by the number of evaluation also caused the lack of new products or best-selling products opportunities.

4, most users are lazy, only to see the default sort; Although you think you offer a variety of obvious sorting options for users to choose from, most users are just too lazy to click (How much of this percentage can be counted).

We know that giving the user not to accept the goods too much display opportunities, or not to have the potential of new products enough to show opportunities, will bring sales losses. When the commodity SKU is young, you can sort the list of items perfectly by manual sorting. But when the commodity SKU number is numerous, the commodity is on the shelf frequently, the artificial sort appears time-consuming laborious. To address unfair opportunities, the principle of sequencing should be:

1, as far as possible to give users the goods recognized more opportunities to display;

2, to give users do not recognize the goods less display opportunities;

3, from the point of view of the Merchant is: the highest profit of the goods displayed to the user;

4, give the sales potential of new products enough to show opportunities.

To sum up, these things have to be done through the system + manual way. In the following view, the system's default sorting factors should be simple to contain the 5 aspects:

1, the number of products in a number of times the display (here with a)

2, the goods in a certain period of sales (here with B)

3, the profit of the goods (gross margin: Sales price-Price) (here in C)

4, the product from the current shelf time (here with D)

5, the goods within a few days of praise rate (here with e)

1 and 2 calculate the sales of the average X time of the product, such as how we can calculate the sales per thousand times in the last 7 days. But if the order only contains these two factors will have obvious flaws, such as in the same display volume, a price of 100,000 diamonds and a price of 10 Yuan book sales is theoretically nothing compared. Therefore, we need to give due consideration to the issue of the profit amount of commodities. Also consider the shelf-time factor and the ratings factor.

The sorting coefficient of the commodity = B/a*c *1/d*e ("B/a*c" indicates the average sales profit of each opportunity, "1/d" indicates the coefficient value of the current number of days) the higher the coefficient, the higher the rank.

For example, a commodity m, in a number of days to display 1000 times, sales of 5, profit of 150 yuan, from the current shelf time of 5 days, a number of days the rate is 98%, then the product's ranking factor: 5/1000x150x1/5x98%=0.147

For example two, a product n, in a number of days displayed 100 times, sales of 5, the profit of 15 yuan, from the current shelf time of 1 days, a number of days the rate is 95%, then the product of the ranking factor: 5/100x15x1/1x95%=0.713

Finally, a product n should be ranked in front of the product m.

This article put aside a lot of details to consider the problem, but also did not combat, only the idea, welcome criticism!

Article Source: aboutb2c.com