Advertisement bid optimization method based on eCPM conjugate prior distribution fitting technology

Abstract

The invention discloses an advertisement bid optimization method based on an eCPM conjugate prior distribution fitting technology. The method includes: simulating a media platform eCPM by utilizing RTB scoring, a GSP advertisement bidding mechanism and statistical characteristics of bidding exposure data, and then sequentially iteratively solving probability density functions of distribution parameters by adopting conjugate priori according to time after judging the exposure statistical distribution, and finally calculating the optimal bidding according to the mathematical expectation. The method has the beneficial effects that eCPM exposure distribution in an RTB bidding scene is simulated; under the condition that a probability density function obeyed by distribution can be analyzed, function parameters can be used as random variables, and iterative fitting according to a time sequence is carried out through conjugate prior distribution; by calculating the joint probability expectation of parameters, the eclectic and important problem of advertisement bid required by an advertiser in exposure is finally solved; and the method has important practical application values for savingadvertisement cost and obtaining effective exposure.

Description

technical field [0001] The invention belongs to the technical field of Internet advertising, and in particular relates to an advertising bidding optimization method based on eCPM conjugate prior distribution fitting technology. Background technique [0002] Advertising bid is the most important setting in real-time bidding RTB advertising, which directly determines the cost of advertisers. Although the vast majority of advertising bidding adopts the generalized second price GSP model, a high first price often corresponds to a high second price. In addition, the media sometimes adopts an improved GSP that uses the first price, so the actual unit price of the advertisement and the advertising bid remain the same. are positively correlated. [0003] At present, when advertisers or agencies set advertising bids, they mainly adopt two bidding technical solutions: the first one is based on the advertising budget and the advertiser's CTR and CVR experience value, reversely calcula...

AI Technical Summary

Problems solved by technology

[0004] 1. The advertising bid is a fixed price set manually according to the budget, which cannot be flexibly adjusted according to the real-time competition of the advertising space;

[0005] 2. There is a lack of advertising competition information guidance during manual adjustments, and price adjustments are blind;

[0006] 3. Often due to the low price, it is impossible to obtain enough exposure, resulting in the failure of advertising

[0008] 1. Advertisers and the media are in an antagonistic relationship. In order to earn more income, it is difficult for the media to wholeheartedly improve the cost performance of advertising;

[0009] 2. The actual advertising conversion price is often higher than the expected price of the advertiser, and statistics show that the difference between the two is as high as 10-30% in many cases;

[0010] 3. Some media platforms have a high threshold when adopting this method. The OCPA or OCPC model requires advertisements to run for a period of time and obtain sufficient conversion data, which greatly limits the privacy protection and method choices of advertisers.

Images