AI is undoubtedly the hottest topic at the moment. In the past year or so, we have witnessed OpenAI's "soaring" from ChatGTP to SORA, and many industries and enterprises have raised how to embrace AI and make good use of AI as a higher strategic priority.
It is not clear what the future will look like, but it must be a huge energy comparable to the invention of the steam engine, electricity, and computers—this is the unanimous conclusion.
In China's cross-border e-commerce industry, whether it is a platform, a service provider or a merchant, how to take advantage of the trend is an important issue in front of us. Looking at the proven AI application results so far, they mainly focus on product selection decisions, content generation, advertising delivery, customer service and other fields. Especially in the field of advertising, it can be said that it is an important incision to create new increments by helping businesses achieve the upgrade from efficiency improvement to value creation through AI-driven data technology.
However, unclear issues such as the underlying algorithm logic, functional boundaries, application methods, data security, and how to integrate with existing processes are preventing more businesses from using AI tools to solve operational problems. To this end, XMARS, the intelligent advertising platform of SparkXGroup, organized an in-depth seminar with the theme of "Amazon AI Advertising Operation", gathering many operation experts to exchange Amazon advertising trends, merchant operation pain points, AI advertising operation underlying logic, AI advertising operation practical experience, etc.
01AI seizes a new high ground in advertising operationsAccording to Shen Junhua, senior partner development manager of Amazon Ads, on the basis of increased advertising inventory, diversified demand from sellers, and increasingly rich solutions, Amazon Ads will usher in strong growth in the global market in 2023, among which, the Chinese market is also growing rapidly, and the active Chinese advertisers in 2023 are more than 13 times that of 2017.
At the same time, Amazon Ads will usher in five major trends in 2024: first, the application of generative AI, such as automating the creation of ad creatives to reduce production costs; the second is to reach a wider audience and enhance brand awareness through streaming TV advertising; The third is the integration of clean room data to enhance the accuracy of advertising and improve the conversion rate; Fourth, machine learning improves relevance to accurately reach the target audience and improve advertising effectiveness; Fifth, interactive advertising enhances user engagement and enhances brand experience.
However, in the context of intensifying global competition, merchants are also faced with challenges such as complex marketing strategies, the need to integrate multiple advertising products and strategies, and the difficulty of proving marketing value. At the level of advertising operation, there are common pain points such as lack of goal management capabilities, lack of effective advertising strategies, lack of systematic data analysis and decision-making optimization capabilities, and lack of energy to implement strategies and track advertising effects.
The advancement of AI technology is like a light shining into the field of marketing, which has solved many problems. For example, based on business data or advertising data, AI can better determine what kind of budget should be used at the next point in time, and optimize or deliver ads according to existing business goals. In addition, the content of AIGC can also achieve more personalized and real-time landing page copywriting and advertising creative; By optimizing marketing data to deliver more accurate information to consumers, you can form a complete closed loop from insight to action, making advertising more efficient.
Even with SOPs, the performance levels of different operations (personnel) still vary greatly, and it is important to improve the average level through AI. ”
Amazon Ads operations is a very complex body of knowledge, new sellers need to use tools to get started quickly, and experienced sellers need to use tools to do repetitive low-value work. ”
From focusing on ACOS value, to focusing on advertising structure, and now to paying attention to data models, the dimension of indicators is increasing, and companies need to iterate on advertising operation ideas by contacting market talents and technology trends. ”
Regarding the value of AI in Amazon Advertising operations, a number of merchants and operation experts expressed positive opinions at the seminar. They have always believed that from the current point of view, it is a way to greatly improve efficiency by using AI to run data, verify and optimize the direction of advertising operations.
Taking XMARS as an Amazon AI advertising optimization SaaS as an example, in Amazon Advertising operations, what AI does is similar to daily operational work, or imitation and expansion of human operations. Its action space mainly involves the optimization of bidding, budgeting, targeting, and advertising structure.
02Meet AI advertising operations through Xmars, Chief Solution Officer of SparkXglobal, pointed out in his sharing that since 2017, AI has entered a period of rapid development, accompanied by the birth of the two concepts of "machine learning" and "deep learning". The core of the former is to let the machine automatically learn and find the rules, which is different from the system that executes according to manual rules; The latter requires a large amount of labeled data to train the model. XMARS works by using machine learning to optimize Amazon Ads delivery and find the optimal ad delivery strategy.
Kun, head of the data science team at XMARS, explained how XMARS uses AI to optimize Amazon Ads. In summary, data and algorithms are the two core elements of XMARS's operation, that is, by collecting and processing large amounts of data, with the help of machine learning, modeling**, expert system adjustments, and other steps to achieve accurate bidding, targeting, budgeting, and optimized ad structure.
Specifically, there are three levels to how XMARS works under the hood:
1. Use "fast and complete" data to provide data analysis and decision-making for AI optimization. XMARS has comprehensive data** including Amazon-related data, XMARS-unique data (such as product information, keyword information, industry information, etc.), party data (i.e., customer CRM or other self-owned data), and can obtain hour-level advertising data and business result data to make ** and judgments.
2. Optimize ads with the help of machine learning model training. Through advanced algorithms (such as 20 kinds of algorithms, an "expert system" to supplement and adjust the logic for complex scenarios), extremely fast computing power, and the conversion rate of user advertising, XMARS automatically provides bidding and budget optimization strategies accordingly.
3. Humans and AI partners to achieve better and more personalized operational results. In terms of ad creation, merchants set targets to promote growth, maintain stable orders, and campaign impulse, and the AI intelligent recommendation engine realizes intelligent bidding, budgeting, and targeted recommendations of keywords and competing products. In terms of ad optimization, for existing ads and ad campaign groups, the AI intelligent optimization engine can automatically optimize bidding, targeting, and budget based on goals.
In addition, explainability AI and personalized AI are a major feature of XMARS. Explainable AI, which helps merchants intuitively understand the reasons behind AI adjustments through data analysis that supports decision-making; Personalized AI is gradually trained and interacted by adjusting the value of the merchant (for example, by adjusting the AI execution frequency, adjustment amplitude, data review window, etc., to set the degree of "aggressive" and "conservative" AI), so that the AI is more and more in line with the personalized needs of the merchant.
03How to properly understand and apply AI? Summarize and accumulate the actions and processes of merchants in the daily operation of Amazon Ads to achieve batch operations and rule-based automation - this method does not involve AI, but only a machine automation process, that is, through simple mechanical execution to improve efficiency. Based on the business goals and marketing goals of merchants, AI makes further judgments, and actively creates, optimizes, and improves ads in a closed-loop manner.
In other words, AI can be seen as a super partner and partner for merchants in their operations. The work done by AI is very close to the work done by humans, but with a large amount of accumulation and complete algorithm logic behind it, AI can ensure that each action is more accurate, timely and in line with future business goals than humans.
However, when actively engaging in cutting-edge technology, strengthening AI knowledge learning, and choosing the right AI tools to assist advertising operations according to their own business needs, businesses also need to correctly understand the application of AI: First, a large amount of data is crucial to the effectiveness of AI tools, so sufficient data is the foundation of everything. Second, you shouldn't expect too much from quick results, as AI takes time to learn and optimize, and it's a gradual adjustment process.
As Daniel Dong, founder of Yifeng Business School, shared at the seminar: using AI well requires a process, and the most important thing is to manage expectations, spend energy to understand it, and wait patiently for the effect.
Daniel Dong shared the three stages after the introduction of XMARS with his own case: at the beginning, the expectations were high, the data tended to be positive, but the effect did not meet expectations; Later, the expectations began to be adjusted, the expectations were lowered, and the frequency of manual intervention was reduced, and after a period of data accumulation, the effect gradually approached the goal. Now, after about 20 days, the ACOS has been successfully reduced from 30%+ to 20%+, and finally the ideal synergy with XMARS has been achieved.
According to him, after using XMARS himself, the cycle of running out of the effect is 14-25 days, and the success rate is about 80%. In his view, AI energy is a powerful supplement to Amazon's daily operations: "At present, we use XMARS to do many things that cannot be done manually, such as time-sharing price adjustment, time-sharing budgeting, lock ranking, etc., which allows our operations to take away from trivial work and do more strategic work." ”