From the mathematical logic calculation, we can know that the ranking logic of Amazon search engine is not a simple listing. The greater the sales volume, the higher the ranking under a certain keyword search result, nor the higher the degree of “matching” in our subjective logic. The higher the ranking, the more it meets the following basic criteria:
(1) The search keyword is included in the title/advertisement/keyword of isting. No matter how good your product is and how excellent your listing is, if a keyword is not included in the title/advertising keyword of the listing, and the keyword has no highly popular synonyms, then the listing will not exist in the search interface anyway. For example, under common keywords in some categories, the “Bestseller” (best-selling) product with the white logo on an orange background will not be displayed, but the “Amazon Choice” (Amazon Recommended) product with the white logo on a black background will be displayed. )product.
This logic is relatively easy to understand. Suppose a product is particularly suitable for A/B/C scenarios or has A/B/C selling points, and the product title/advertisement/keywords only contain A/B keywords. And there are no synonyms/synonyms of C, so even if C is one of the “valid keywords” of the product, Amazon will not display the product under the search results of C-related keywords. (Of course, in order to prevent this from happening, you can use bundled marketing reviews to achieve forced associated exposure).
(2) The results presented on all pages conform to the order with the largest order amount. We know that order volume = traffic X conversion rate; order amount = traffic X conversion rate customer unit price. Amazon charges a certain percentage of commission based on the order amount, so for Amazon, the arrangement of each keyword search interface is in line with its own principle of maximizing long-term commission collection.
After understanding the above two basic principles, how to use advertising to increase sales? There is only one answer: improve keyword rankings.
That is, the product listing ranks high in the search results of certain high-traffic keywords and has a high click conversion rate and order conversion rate (click conversion is related to pictures, this section mainly talks about the order conversion rate) .
Two technical indicators are involved here: high traffic and high conversion rate.
The following is a Q&A format to answer questions you may have.
Q: Why high traffic?
A: Advertising traffic is a booster for listings, which can quickly increase the overall traffic in a short period of time and verify that the product is under each keyword. performance effect. Suppose the title of our product is “women loose fit cocktail party elegant long maxidress”, then if the advertising keyword is set to “loose fit cocktail party long dress”, a precise long-tail word that extremely overlaps with the title/keyword, even if our product The keyword search ranking can be ranked first through advertising every time, but the search traffic of the keyword itself is too low. It’s difficult to attract high traffic and generate orders. At this time, not only the sales volume is lost, but also the effective promotion time is wasted. The listing with advertising is the same as without advertising, and the meaning of advertising promotion is lost.
Q: Why should we improve the order conversion rate?
A: With sufficient funds, any store has the opportunity to use advertising to push its product listings into this category. The first position in the ranking of major category words. But there are two problems here: 1. Forcibly using advertising methods to promote the product regardless of the actual situation, often due to the mismatch between the product’s own elements/selling points and the specific search fields of the A9 search engine, resulting in a very high advertising investment-output ratio (ACoS) , resulting in the unfavorable situation of “spending money to earn praise”; 2. Amazon has a complete set of bidding ranking rules for the display of advertising listings. If your listing occupies the best display position but does not have matching exposure click-through rate and clicks Conversion rate, Amazon will gradually reduce the exposure of the listing, eventually leading to the embarrassing situation of being advertised but unable to spend money.
Q: Why is the ACS sometimes very high when automatic/manual advertising is turned on?
A: The product attributes do not match the advertising promotion attributes. Each product has various attributes, such as color, size, price, style, reviews, etc. There are many sub-attributes in these parent attributes. For example, clothing category sizes generally include XS, S, M, L, XL, XXL, etc., and standard product prices generally include single product price, package price, customized price, etc.
The reason why a certain product is really hot-selling is because its sales volume is very high, and the prerequisite for very high sales volume is to have an extremely excellent ranking under matching high-traffic keywords, and we are going to do that. Among these various sub-attributes, find the most matching attribute and strengthen its connection through advertising. However, once the keyword combination strengthened by the advertisement is not the biggest attribute and selling point of the product, then the ACoS will become very high due to the extremely low conversion rate.
Q: How to analyze each attribute when the product is put on the shelves? What about search weight?
A: Let me take the title keyword “women loose fit cocktail party elegant long maxi dress” as an example. There is no doubt that its product root attribute is “dress”, and the parent attribute It includes color, size, price, style, evaluation, etc., but since this is a skirt, the parent attribute of style is very important. According to its title/keyword, we can conclude that the style of the parent attribute contains the following sub-attributes: loose fit, cocktail party, elegant, long, maxi.
When it was first put on the shelves, the weight of each sub-attribute and the search results of each sub-attribute.
Why does “loosefit” have a larger weight? This is because its search results under the women category are only 10,000, which is much smaller than the search results for other sub-attributes, so it has a higher weight than “loose fit”. “dress” has a greater exposure probability than other combinations. However, it should be noted that here the title/keyword is directly split and combined with the root attribute “dress”. When matching with the A9 algorithm, it will be combined twice, three times or even multiple times, so we still choose The six sub-attributes just were combined twice, and the weight of each combined word and the search results of each sub-attribute were combined twice.
Through the secondary combination, it can be found that although the sub-attribute of “loose fit” has a higher weight before, it will show a completely different weight distribution when paired with other words. Of course, the current simulation is a comparison of the weight of each attribute after the product has just been put on the shelves. When the product begins to be exposed, the A9 algorithm will match the corresponding search field based on the product’s real attributes, and the role of keywords/titles will gradually weaken. But even so, if the keyword/title used by the product that has just been put on the shelves is a combination with a lot of traffic, the weight of its sub-attributes is likely to be very low, making it difficult for the product to be exposed.
In contrast, some attributes of the product are not included in the title/keyword, so no matter how high the weight of other sub-attributes is, once the hot-selling attribute is the missing word or phrase, The growth of product listings will be very slow.
This is like you are a melon farmer selling watermelons at the beginning. There are two selling points of watermelons: sweet and sweet, but you only promote the selling point of sweet (that is, the keyword lacks a certain selling point or the The weight of sub-attributes is not large), so even if there are sweet watermelons and no seedless watermelons on the market, consumers will not be able to know the characteristics of your product without seeds and make a purchase.
Q: How to find the sub-attributes (selling points) of matching products?
A: There are two situations here.
The first situation is that the product is a creative product, that is, there are no/very few products with this attribute (selling point) on the market before, so its matching sub-attributes need to be forced through the title/keyword when it is put on the shelf. The given and completed association is shown in Figure 7-42 (these belly bags were very popular at the end of 2017).
The sub-attribute word that closely matches the belly bag is “funny”, so if there is a very eye-catching attribute and innovation when putting a creative product on the shelves, you can directly find the matching sub-word through manual methods. Attributes. The second situation is that although the product is different from similar categories, its matching sub-attributes are very vague, that is, the biggest selling point cannot be found. If you encounter such a product, you can use the method of keyword optimization to create a competitive product analysis. Table, and use manual query method to count the frequency of similar products appearing on the search results page of different keywords, and find a batch of keywords with the most occurrences for placement. Of course, sellers can also use the listing operation and the calculation of the A9 algorithm to find the most matching sub-attributes of their products within a period of time (7 to 21 days), and then strengthen the correlation between their attributes and products through advertising (usually manual advertising) to improve keywords Ranking. The shelving operation means that when putting the product on the shelf, you need to ensure that the title/keywords contain the main sub-attributes of the product and cannot be omitted. For details on the subsequent operations, please refer to “Advertising Optimization Method – Probability Matrix Matching Method”.
Advertising optimization method-probability matrix matching method.
Two properties of the Amazon A9 search algorithm:
(1) The search keyword is included in the title/ad/keyword of the listing.
(2) The results presented on all pages conform to the order with the largest order amount.
From this we can understand that Amazon’s A9 algorithm must find the most matching sub-attribute within the scope of the product listing title/keyword in order to maximize sales for the product, thereby giving a good keyword ranking. , then as long as through certain methods, after the product has been on the shelves for a period of time and has a certain sales volume, the most matching sub-attributes of the product can be analyzed through the results of the Amazon A9 search engine algorithm, and then a large number of them can be exposed through advertising. When exposed products match sub-attributes, ACOS will not be too high in the long run because Amazon will not give a product with a low conversion rate a high keyword ranking, so if the long-term ACoS is too high, it means that the current advertising exposure Attribute words match attribute words for non-products. Note that it is normal for short-term ACOS to be high, because it is impossible to cover 100% accurate keywords).
Then we can make the following mathematical deduction.
Suppose a certain listing keyword combination is matrix A:
Among them, a11~ann are all one of the keyword permutations and combinations, that is, n×n combinations.
Take “sexy, casual, dress” as an example (note that the root attribute word “dress” does not need to be in the matrix arrangement):
After excluding duplicates, it can be transformed into It will involve secondary combinations such as “sexy casual dress”, because the combination can be in the form of “sexy casual+dress” or “sexy+casualdress” or “sexy+ casual+ dress”. In the Amazon A9 algorithm, the so-called “big words” + long-tail word” or “adjective + noun” and other forms are formed by the matrix arrangement and combination of this phrase.
Enter each combination in the Amazon search engine and then select the corresponding category to obtain the ranking set R.
When we enter the “sub-attribute word + root attribute word” combination corresponding to “y,” we can see the number of search results for this category corresponding to the Amazon search engine.
The number of search results and search rankings directly determine the exposure rate of our products. We set this value as P. According to the calculation, we can use the traffic formula
The single-day traffic is yi(. 1≤i≤n), the exposure amount in a single day is xi(1≤i≤n), the keyword search volume in a single day is fi(1≤i≤n), and the total number of products that can be searched in a single day is ni( 1≤i≤n).
The calculation of the exposure probability Pu value involves the single-page traffic loss rate. This value can generally use a fixed value ranging from 0.3 to 0.7, or it can be captured through web page data. obtained, but this is just a reference variable and does not need to be too rigorous in the advertising optimization stage.
So what is the logical meaning of the Py value? Suppose that a product searches for the total number of products under a certain keyword search result. Quantity “y” is 10000, ranking 7 is 5, then Py=(1000-5)/1000, which is about 99.95%, then it can be considered that the probability of customers seeing the product is 99.95%, and the probability of successful exposure is 99.95% . At the same time, assuming that the total number of products searched for a product under a certain keyword search result “y” is 10,000 and ranking 7 is 100, then Py=[(10000-100)/10000]×(1-0.6)^3 (assuming The value of the traffic loss rate is 0.6), which is about 6.336%. Then we can think that the probability of a customer seeing the product is 6.336%, and the probability of successful exposure is 6.336%.
Exposure probability matrix. Each element in can be calculated.
The next step is to analyze the matching degree of the sub-attributes.
The selling points of this product are roughly divided into: patchwork, maxi. , floral, pocket.
After deleting the duplicates, search the sub-attributes and their combinations in the search engine in sequence (the following data was crawled in May 2018, and may have been by the time you read this book) There is a large error, so the data in this chapter are only for demonstration. In actual operation, you can manually enter keywords, record the ranking data one by one, and finally calculate it. Optimizing the single advertising probability matrix will take 20 to 40 days. minutes).
Suppose the single-page traffic loss rate l is 0.5 (for categories with a large page loss value, it can be set to 0.7; for categories with a small page loss value, can be set to 0.3), then the exposure probability matrix is:
During the exposure matrix analysis process, the single attribute matching patchwork becomes the strongest matching single sub-attribute with an exposure rate of 99.65%; the combination of sub-attributes matches the patchwork+floral combination It becomes the strongest matching sub-attribute combination in the first row with an exposure rate of 9965%; the sub-attribute combination matches the maxi+floral combination and becomes the strongest matching sub-attribute combination in the second row with an exposure rate of 616%; the sub-attribute combination matches the floralpocket combination with a 12.23% Exposure rate becomes the strongest matching sub-attribute combination in the third row.
There is no sub-attribute combination in the fourth row, so there is no strong matching item.
We can draw the following conclusion:
(1) The strongest matching sub-attribute of the root attribute is patchwork.
(2) The strongest matching combination of the sub-attribute patchwork is “patchwork+floral”.
(3) The strongest matching combination of maxi except patchwork is “maxi+floral” (4) The strongest matching combination of maxi except patchwork is
According to the above conclusion, the following matrix combination can be obtained by deleting the weak matching sub-attributes and their combinations.
In order to facilitate comparison, the search results are shown. Select the common multiple of the existing elements in the matrix, and the common multiple is 30,000 (the meaning of setting the common multiple is to calculate the estimated ranking of the product under the existing competitive conditions, assuming the number of competitors is the same).
It can be seen that the competitiveness of patchwork and the competitiveness of “patchwork+floral” are basically the same under their respective keyword search results. Considering that patchwork has more search results, it will cost more to use manual advertising to push it to the top column in this category, so the keyword combination “patchwork+floral” is added to the keyword selection of manual advertising. Better value for money.
Among other combinations, although “floral+pocket” had 98 seemingly good keyword rankings at the beginning, when the competitive conditions became the same, the combination of “maxi+floral” performed even better. However, considering that its ranking is still very low, it is not recommended to use manual advertising to push it to the first column of the search bar, so there is no need to click on the option of paying 50% more per click fee.