Wang Cheng: Companies Embracing AI Technology with a Positive and Open Mind Will Form a First-mower

「Future AI Talk」 is a dialogue program co-sponsored by Marteker and Digital Frontier to explore the impact of the rise of generative AI on pan-marketing technology and marketing automation, in order to help the entire industry explore a new marketing path in the era of AIGC.

 Interviewee: Wang Cheng, Co-founder and CPO of Convertlab

Q: How do you understand AIGC?

Wang Cheng: From the perspective of AI output, AI includes generative AI and synthetic AI. Generative refers to its own characteristics of getting requests and a small amount of information, generating more information, from the surface of generative AI is particularly good at generating content, so it is quickly adopted by enterprises and individuals to get answers to questions or directions.

Synthetic AI gets insights and decisions from massive amounts of information, and our very common example is that companies have a lot of business data for AI to gain insights from; Or provide the complete background information needed to make business decisions to the AI to make decisions. We think the value is higher in synthetic AI. Because generative AI pays attention to creativity, it can have its own unrestrained play, and the synthetic type is to make judgment decisions, requiring the reliability and stability of reasoning, and of course, relatively higher requirements. From the perspective of maturity and risk, the risk of synthetic AI is itself, of course, the technology iteration of the large model itself is very fast, it is possible to solve this problem from the basic model, and at this stage, the existing problems are mainly alleviated through supporting technology management and control.

Q: How do AI and marketing automation fit together?

Wang Cheng: We can broaden our vision and look at our whole marketing cloud concept, from data insights to data-based marketing decisions, including automated execution, basically in the whole marketing cloud link, what role can AI play?

First of all, there are probably several aspects that are feasible: data-related, including data cleaning and governance, data model establishment, label processing, and crowd selection, these aspects can be infiltrated by AI, of course, based on different scenarios, the penetration rate will be different. For example, Convertlab cooperates with some customers to clean a relatively primitive and disordered data source into a more orderly structure to import CDP. Before this work, it mainly relied on manpower, but now AI can be involved, at least manpower combined with AI to complete it, which has greatly improved the work efficiency, and at least doubled the efficiency at present. The future is higher.

In another job, we had AI learn from industry best practices. For example, the clothing, shoes and hats, luxury goods or automotive industries are very different from each other, and the CDP model of each industry has essential industry characteristics. What we are doing now is to let AI learn the best practices of the industry, and let it map to the best target structure of different industries based on the original data structure of the enterprise. Where it used to be done entirely by people with business experience combined with data engineers, it can now be basically semi-automated.

As for data operation work, I want to quickly build a label, which can be created by giving AI an instruction, or delimit the crowd, which is also completed by describing data dimensions and other conditions and saying it to AI in natural language. Now AI technology is more reliable, can complete the work under human assistance and monitoring, and help enterprises to turn relatively large but disordered data into regular data sets in a short time, and make use of it.

If it's deeper insight, which is what we call exploratory analysis, to help customers solve business problems, such as the decline in member sales of a channel last month, how to find the answer? Previously, business experts and data analysts worked together. Now we let the AI do some exploration automatically, looking for more important data signals, and unearth important insights hidden in the data. The work in this area is a little more complicated, and we are progressing smoothly, and we will see the final results over time.

On this basis is the partial MA (marketing automation) part, which has some AIGC scenes we are familiar with, such as the generation of marketing content, including SMS copy, email copy, image generation, and so on. Let me talk about other aspects of AI application, in marketing automation there is a very core is personalized content, that is, different groups of people may have different characteristics, so the same topic to provide many different versions of the content. It used to be too expensive to do this, because each version had to be designed, copywritten, audited internally, and so on. Now automated content generation solves this problem.

Further work is the decision of marketing strategy, but also depends on our AI to learn the best marketing practices of the industry, users only need to issue different business instructions, such as improving the transformation of new customers, reducing the turnover rate, etc., behind these different instructions there are actually AI through industry best practices to reason what kind of marketing strategies and actions to take. This work will then be directly reflected in MA products. Enterprise users only need to tell the MA system what are the indicators of concern at present, the marketing strategy recommendations given by AI, what kind of marketing activities should be done for what kind of customers, and what kind of changes should be promoted.

Q: With the AIGA large model, product interaction is much simpler. Is this simplification likely to lead to less reliance on technical expertise? How should companies balance simplifying interactions with maintaining employee understanding and mastery of technology?

Wang Cheng: This is a significant change, and the threshold for use will be lowered. Of course, the original complete task flow is retained. Future products will offer both normal and expert modes. The use of ordinary mode is low, the user does not need to know too many details to complete the work; Expert mode, or advanced mode, is convenient for advanced users to fine-tune the granularity and set certain functions of the product.

Just like Excel in Microsoft Office, the existing function of designing complex formulas still exists, but it does not require the user to actually operate, but just issue a command, saying what effect needs to be achieved, and the formula will be automatically generated. Future products will achieve this effect.

This will be the main mode for the next period of time, maybe two or three years or even a little longer, when AI technology matures, and even can completely hide the expert mode of the product, for many people all the work becomes very simple, complex use functions do not need to be exposed to the end user. That's where the next phase of the product is going.

Q: The advent of AI means that marketing workflows and team collaboration models have changed a lot. How should companies adapt to this change?

Wang Cheng: In fact, most white-collar workers in enterprises tend to be knowledge workers, and AI will have a big impact. There's this idea that AI is going to take people's jobs, and I don't really subscribe to that. Of course, if some people are very conservative, do not embrace new practices, and do not love learning, regardless of AI, he is also easy to be eliminated.

With an open mind and a willingness to learn, not most jobs are replaced, but all jobs are upgraded. The vertical job content is not suitable for the future development, and the future work may be comprehensive requirements for everyone. Paranoid work such as collecting information is easily handed over to AI to complete, but the final decision judgment, overall coordination and other work, still need to be solved by people. The number of jobs in the Marketing Department will be reduced, many jobs will be merged, and everyone will grow into the current team leader, responsible for comprehensive work, good at many jobs, although not necessarily proficient, but very clear understanding, good at controlling the work from a global perspective. This is what I think most people's work trend will be in the future, where individuals need to learn to collaborate with AI, upgrade to a "super individual", promote the efficiency of the entire team, the entire enterprise and even the entire society, and ultimately improve the productivity of the entire society, thus giving birth to new occupations or positions, or generating new needs.

Q: What are the new opportunities and challenges for brands to engage with consumers through AI?

Wang Cheng: Different companies have different attitudes, some companies are relatively cautious, have been watching, will hesitate to take action; Some enterprises are more open and embrace AI technology with a more positive and open mind. For at least the next two or three years, companies that are more open-minded and more aggressive in adopting new technologies will themselves form a first-mover competitiveness.

Take advertising as an example, if the brand wants to do the placement of short videos, it needs to be very careful to do creativity, make proposals, and carefully consider to ensure the success of the placement. But now with AIGC, some brands realize ideas at a lower cost, can do more different versions of the idea, do small-scale testing, find the best version, and then do large-scale launch. This model is superior to previous models in terms of efficiency, cost and success rate. The sooner a brand adopts this model, the sooner it will have a competitive advantage and use this advantage to build competitive barriers.

For example, personalization, many brands want to treat the customer as God and provide a very personalized service experience from the beginning of the contact with the customer. AI technology can effectively accelerate or deepen personalization.

So to some extent, it is the change of ideas and ideas that comes first. AI technology or solutions and applications are in a rapid iterative process and are not yet very mature. If the brand has to wait until it is very mature and then adopt it, it is the same speed as its peers, and it is not particularly ahead. Brands want to lead, now can be bold, and solution suppliers to try together, earlier than other brands to practice, there will be phased competitive advantage.

Q: AI is also allowing marketing technology companies to redefine products, business capabilities, and ecological landscapes. How should they rethink and adapt their strategies?

Wang Cheng: For Convertlab, for example, we already had an AI strategy at the beginning of this year, and we are constantly thinking about it. Now approaching the end of the year and looking back at the idea at the beginning of the year, there is almost no change in the general direction, and the judgment of some technical feasibility in the middle will be found that some are not the same as envisaged after practice. The past may have been too simple, too optimistic. But the general direction is still very clear, basically the same as at the beginning of the year.

The combination of AI capabilities and Martech has become the norm, not a matter of whether enterprises should do it, but if they do not do it, they will not be qualified Martech suppliers and may be eliminated immediately. This year, there is a third-party survey, the priority of the adoption of AI technology, the combination of marketing scenarios and AI technology, according to the feedback of enterprises, the priority is very high, including content creativity, marketing, sales, customer service and other related functions, basically within the scope of Martech. Of course, AI applications also involve supply chain, finance and other functions, but for enterprises to feel the return of output in a short period of time, they need to wait for some time.

So naturally, Martech is at the forefront of integrating AI technology. So AI is the default foundational technology to consider. Throughout Convertlab's product strategic planning, the combination of AI technologies will be the default option, and AI is everywhere, the new normal for products. We also consider whether the enterprise has to pay more after the product has added AI functions and upgraded, but we think that this is wrong, and the combination of AI in most cases is to let users better meet the original business needs and expectations, and does not exceed the original value proposition of the product, so customers should not be charged more.

To some extent, our products will be enhanced by all-round intelligence. We also have an interaction mode similar to Microsoft Copilot, that is, there will be AI plug-in assistance next to the workflow to make the original workflow perform better. That's one of the scenarios we're considering. The other is that AI is directly integrated into the entire function, embedded in the entire process, which is probably the more mainstream way. In addition, we also provide independent AI application rapid build solutions for users to solve a single point of marketing problems, such as marketing content creation, and can also be used in non-marketing scenarios. These are Convertlab's two AI product strategies: one is the full intelligence of the original product system, and the other is to provide independent AI applications to solve marketing problems in a lighter and faster way.