The new generation of Jingdong marketing 360 intelligent A/B Split Experimental platform -Youtou Lab recently launched a split experiment function in their CPD advertising platform. Having helped advertisers realize the intelligent creative A/B experiment without staring at the data, the intelligent traffic allocation algorithm can also greatly improve the click through rate of CPD advertisement.
During the period of Jingdong 618 Shopping Day this year, through the trial use by many brands, the AI optimization algorithm – haave significantly improved click through rate. Many brands were impressed by the split and contrast experimental function and said it is "a great progress of CPD advertising!"
Trusted by many brands
Nearly half of Unilever, Yili, Mengniu, Hp, Vivo, etc. used the split experimental function during the promotion period of Jingdong 618. After using this function, nearly 70% of the businesses will continue to use the experimental function in the subsequent CPD advertising.
Mengniu (a Chinese dairy brand) feedback after using the CPD advertising split experiment function and saying "the split experiment gave us significant brand exposure via CPD advertising, which initially solves the problem that CPD advertising couldn't be optimized. We saw significantly improvements in the overall click through rate."
High usage ratio to improve click through rate
During the period of Jingdong 618 this year, the vast majority of CPD advertisers used the split experiment function to optimise their creatives. According to statistics, the consumption of CPD advertising using split experiment function accounts for 83% of the total consumption of CPD advertising. Among these CPD advertisements that open the experimental function, 46% of the advertising campaign's CTR has increased by 10%, and 22% of the advertising campaign's CTR has even increased by more than 20%.
In view of the positioning of functional intelligence experiment and the obvious improvement effect of CTR, many brands are interested in the underlying principle of splitting and comparing functions. In fact, the core technology of intelligent experiment comes from the published research paper:
Algorithm being selected by the top international AI conference
In February this year, AAAI (the Association for the Advance of Artificial Intelligence) was held in the United States. The research paper <Online Evaluation of Audiences for Targeted Advertising via Bandit Experiments> published by Jingdong and Stanford University " was selected successfully.
The core principle of split experiment is the practical application of this research paper:
l The system adopts the industry-leading reinforcement learning model, which can monitor the effect of each creative in real time during the whole experiment;
l At the beginning of the experiment, the intelligent algorithm will play creatives randomly to fully explore and understand the real effect of each creative;
l After learning each creative to a certain extent, the model will continue to update the playback strategy to play the better creative with greater probability;
l In the case of ensuring the fairness of the experiment, the loss and cost of the experiment are reduced to the greatest extent, and the experimental benefit is improved.
Intelligence, data and technology are the core of Jingdong marketing 360 to provide professional, high-quality and efficient marketing solution for brand merchants. In terms of how innovative intelligent algorithm improves marketing efficiency, Youtou Lab will continue to provide more abundant experimental scenarios for businesses: it can not only provide intelligent A/B experimental services for the creative, segmentation, bidding strategy, landing page, etc. in the advertising efficiency measurement, it can also support businesses to compare user behavior differences between those who have seen ads and those who have not seen ads, which is scientific and accurate to measure the impact of advertising on consumers and brands.
Ref: Jingdong Marketing 360