A Amoy search responsible Ghost foot Seven: Personalized search affect who?

Personalized search, a look very beautiful, practice but very bumpy biased phrase. Here you can see the small and beautiful, you can see the big data, you can see the mass and the long tail collision, can also see the technology to give business power.

Wen/World Network business reporter Fantingting

"The results of the cat search has been added to the personalized factors, different buyers ID or different cookies, because of previous purchases or browsing behavior is not the same, see the search results may not be the same." Personalized algorithm in search results and marketing resources, such as the full application of the goal is to let consumers see and his most relevant products show, and finally realize thousand thousand face of the day cat. "This day Cat CEO carefree son in the afternoon of August 2 Weibo, explained the recent Taobao users" search the same keyword but show different results "reason.

This microblogging is like a butterfly effect, its effect can radiate to the whole Taobao big and small sellers. To this "World Network Business" reporter interviewed a Amoy search responsible Ghost Foot Seven, personalized search done a detailed analysis.

At present, personalized search is still in the testing phase, the platform side randomly selected about 5% of the traffic as a test sample, through personalized search after the click Rate, conversion rates and other indicators to further optimize the search Taobao. A preliminary statistical result is expected within six to 3-4 months.

　　What is a personalized search

Personalized search, as the name suggests, is in the search process into personalized elements.

Ghost Foot Seven from two aspects to explain the principle of personalized search: On the one hand, from the user's point of view, everyone's preferences are not the same, reflected in the online shopping behavior, is that each person's preference is not the same. This kind of preference can be reflected in many aspects, such as price preference, category preference, geographical preference, use preference, quality preference and so on. In particular, different purchasing ability will lead to user's preference for different customer unit price or brand. Taobao just happens to have the ability to do some digging on these preferences, and then according to different dimensions of the user to do further subdivision.

On the other hand, from a search perspective, the platform has two things to do: first of all, mining the user's needs, and then clear their consumption point, including a variety of consumer preferences for the mixture of a division, followed by the classification of goods, such as a commodity its high-grade or low-end, brand positioning in what age layer, etc. , the platform can classify goods according to various dimensions through data collection. After these two things are done, the platform will be able to make a connection with the goods when the customer enters Taobao shopping. If the consumer is we commonly known as "cock silk" customers, usually like to buy discounted goods, the price trend is low, then search to give him the results of the page will be biased to lower price products. If you observe a consumer's age in 30+, the ability to consume stronger, and pay attention to the pursuit of the brand, then the platform in its search related pages, will meet his consumption habits of products as far as possible to the front row.

We can use an example to concatenate the relationship between the two: if the system detects that a particular consumer is more concerned about the new product, Taobao will show him more of the current popular products accordingly. So how do you define new products? products can generally be divided into two categories, the standard class and non-standard class. The update of the standard will not be very fast, such as 3C products, basically have a clear time to market, the definition of new products is relatively easy, but not the standard category only the brand will have a new one, such as some clothing brands. So for non-standard products to crawl work is relatively difficult, whether from the technical means or sellers side, it is difficult to be absolutely accurate, but this will be a gradual improvement of the process.

At present, a Amoy excavation efficiency can reach 80%, that is to say that the platform side of the new products, more than 80% is indeed a new product. This targeted search results show that the conversion rate is higher than normal.

In this example, in fact, there are two aspects of personalized search, that is, the division of Users and the division of Goods, as well as the corresponding relationship between the two.

To achieve this effective relevance, on the one hand, the need to cooperate with the category, on the other hand, the business on the shelves, the goods have a more accurate description. And the search party itself will do in-depth excavation.

And for sellers, personalized search helps to improve the value of traffic. The original consumer into the Taobao page, search the same keyword results show page is the same. In this case, on the one hand, consumers can not quickly find their real needs of products, on the other hand, the display of business is a waste of traffic.

And if replaced by personalized search, the formation of thousands of thousands of people search results show, then customers can see more with their own expectations of the products, shops also have more opportunities to display, that is, improve the value of the flow.

This for shop style more clear shop, may not spend a lot of cost to do promotion, can attract the high quality of the target customers, whether it is for businesses or buyers, have saved time costs.

So personalized search, the most important purpose is based on the different people in the past purchase preferences, to provide customers with similar goods, with more accurate positioning, a higher degree of matching search results, to improve product clicks and store conversion rate.

　　

　　The difficulty of personalized search

But Ghost Foot Seven also admits that personalized search, although it sounds beautiful, but in the actual operation of the process will have a lot of problems.

First, there are many difficulties in making a customer distinction. The identification of the user intent can be a very tangled process. First of all, the user itself according to age, gender, geographical, income and other dimensions of the differences, and currently Taobao there are many more than two people share an account of the situation, such as male and female friends, such as parents and children. This situation causes their buying behavior to sometimes appear to be "contradictory" in the system. If the system recognizes that you are a woman, but you buy a male jacket, the system is "dumbfounded". What a personalized search can do is, when you're a woman by default, you search for "jeans," and the results show a lot of women's jeans, unless you're typing "jeans man," and the system knows you're buying a product for men, and it will show you what you want.

So for the platform side, it needs to consider a lot of factors, if the search results are too rigid, but will cause users disgust. Because not all buyers have a clear user behavior, most of the purchase behavior may be more random. If you can identify 20% to 30% of the user intent, in the ghost foot of the seven looks can be considered a preliminary goal of the agreement.

Secondly, the classification of commodities is also a vast project. How do you define a product? Is it cheap or high-grade? is a dress suitable for office hours or on weekends? For a product positioning, but also consider its use of the scene. At present, for these different dimensions of positioning, will take the business and platform side of the way to complete the cooperation between the two sides.

So, personalized search is an interactive process, in the process of integration of personalized elements, Taobao also want more buyers to participate in. It is the best way to be willing to help the platform to optimize the rules of the search if there is a high fidelity customer willing to describe their spending habits. The higher the user's participation, the greater the degree of personalization is achieved.

Ghost Foot Seven to the "World Network Business" reporter played an analogy, we play Micro Bo, everyone saw the page is not the same, your page is displayed on the content of your custom, see the information you want to see. Personalized search is also the hope that through the interaction between buyers and sellers and the peace of the Taiwan side, and ultimately achieve a high efficiency of information to obtain results, so that consumers have a better shopping experience.

The current testing phase, for businesses, the feeling may not be big. Ghost Foot Seven think, at least to the end of this year, 60% of the flow will not be a big change, for the remaining flow, the store can fully display their personalized. But this personalization factor is to consider the product itself and the quality of the service, to the current resources and customer preferences better combined.

Personalized search can really play a role only when the customer has a clear consumer behavior. At this time, the seller's product positioning more clearly, then the effect is certainly the better.

Although it is difficult for businesses to improve their traffic according to the specific rules of personalized search, the store can attract consumers who have higher buying intentions according to their own positioning and the characteristics of their products. For example, a consumer tends to more obvious users, he always fixed the choice of a few shops of clothing, these clothing styles, prices are relatively close, then in his search again, the platform will be the same style of products to show him. So personalized search for the style of a clear shop, will be a good channel to expand traffic.

At present, the number of online shopping in China has more than 200 million, for the changes in the search rules, Ghost Foot Seven said it will take a cautious attitude. But the trend of personalized search will not change, without affecting the user experience of the premise, to achieve the results of the platform side, which for the entire search team, not only this year's a project, but also a long-term need to adhere to the goal.

　　

Weakening the impact of high praise rate on search

In addition to personalized search, praise rate of the solution is also the recent Taobao Forum on the hot topic. There are also countless relationships between ratings and search.

From the seller's point of view, simply summed up the past praise rate, there are three aspects: first, when the buyer browsing shopping, the rate of praise as an intuitive data, to a large extent, will affect the conversion rate of the shop; the second is the seller if you want to Taobao some promotional activities, such as gold coins, through train, and so on, Praise rate is an essential indicator, the impact on sellers is relatively large; the third is the search, in the original search ranking, praise rate as an important reference index, for Buyers keyword Search Rendering page has a certain impact.

But like the ghost's foot Seven says, rules also need to keep pace with the times, if the existing rules are not in line with the current environment, into a "favorable to drill" black box, and nourish a number of dark industries, so if the whole of Taobao ecological health development, these loopholes in the rules to upgrade is inevitable.

The reason why the rate of praise to the operation, it is because a large number of professional evaluation Division has seriously damaged the interests of small and medium-sized sellers. Taobao side is doing, on the one hand, protect small and medium sellers, on the other hand, is to eliminate the whole Taobao environment.

Any rule is to make the market more perfect, this is a valuable rule, if this rule can become an accomplice to extortion, it is bound to be eliminated. And if a large number of people are subject to the rules, then this rule is certainly biased.

In fact, as early as June, in Taobao treasure search and shop search, "praise rate" this indicator has no longer affect the search results. June 27, the praise rate is no longer as a gold seller to come into the index.

Ghost Foot Seven told the World Network business, so the biggest goal is to let businesses pay more attention to the transaction process itself, rather than just to obtain an evaluation results.

Ghost Foot Seven with a Buddhist motto to describe this change: "Because of the effort, the fruit on the go." "That is, if the business can do, as far as the results, do not blindly insist." When we do something, it's important to be mindful. In exchange for Taobao come up with the same, sellers to serve buyers, but if you encounter deliberately difficult consumers, leading to business services have deviated from the normal track, there is no need to get a favorable comment and compromise.

Of course, sellers and praise between the untied, not that the platform no longer consider the rate of praise, but to make the rules more perfect, better to serve the buyers and sellers. Perfect means, the platform will consider synthetically, if the merchant sells hundreds of products, most of them are in the middle difference evaluation, then may the product or the service actually has the question. But if the seller sold a dozen or so, but also by a relatively concentrated difference, the platform will consider whether malicious bad reviews. Taobao has also opened a green channel, businesses in the face of suspicious transactions, can be reported.

In addition, the search page will have other functions, such as good shop recommendations, mainly recommend some repeat customers more, repeat the purchase rate is relatively high, yellow diamond buyers particularly like the shop. And the buyer's preferences also have a great impact on the search results, such as some consumers like "by sales order."

The platform in the weakening of the praise rate at the same time, I hope that businesses pay more attention to their own upgrades, since the praise rate is no longer a search for the ranking of a reference value, then what is the important reference to the order? The answer to the platform is to pay more attention to the overall effect, such as the product refund rate, delivery speed, details page description, the response speed. When the platform is sorted, all the integrated factors are taken into account and the products are recommended to the user. Also hope that when you go to the details page, the merchant can have a better way to retain the customer.

There is also a problem of relevance, such as the user search "casual pants men white", if the merchant's trousers are indeed white, but not in the attributes, then this search behavior, the merchant's products may not appear in the search results. So businesses should try to do as much as possible keyword optimization, to specific to the category name, brand name, attributes, style and so on. This is the keyword optimization.

　　

Conclusion: Overall, personalization for the search results page will certainly have a certain impact, but it is not you see and others different pages, it must be the impact of personalized factors. The whole search is sorted, there are many factors involved, the search optimization is ongoing. But same is, the business must first do their own business, clear their positioning, to do their own market deep, so no matter how the rules of the search changes, will be beneficiaries.