From performance marketers and brand marketers to growth marketers and content marketers, there is likely no other job role that has been as segmented and rapidly changing with the times as that of a marketer.
So, now that we have entered the AI era, how is the role of marketers changing?
There is someone who has flexibly adapted to these changes and expanded their capabilities. That person is Kang Seul-gi, CEO of Connect AX. Leveraging her data analysis skills, Seul-gi has worked at agencies and media companies, and is currently continuing with projects related to AX consulting, experimenting with the various possibilities for marketers to expand their careers.
She is also the perfect candidate to be invited as a speaker for the 2027 Marketing Seminar: Brand Content Surviving the AI Era, planned by the Elifunt Team this coming October . We met with Seulgi in advance, who has been witnessing the changes in marketers' roles from the closest perspective—from the field of marketing to the organization's AX transition—and is driving greater growth.
ConnectAX CEO Kang Seul-gi, Elifunt's Hong Yu-kyung (who oversaw this seminar), and myself!
Q. Hello, CEO Kang Seul-gi! Please introduce yourself. 🙂
Hello. I am Kang Seul-gi, CEO of Connect AX! I worked at a marketing agency for about 10 years, and afterwards, I switched roles to a data analyst at a media company to gain experience in data-driven marketing.
Currently, I am independently conducting AI marketing and data-related projects, and 'AI After Work'to help working marketers connect AI and data to their actual work.
Q. You have experienced changes in the role of a marketer across various organizations. In the era of AI, what competencies should marketers focus on more?
A. The excuse that you couldn't do it because of a heavy workload no longer works. (Laughs) It is no longer the ability to do everything yourself that matters, but the capability to design the way you work.
In the past, execution was often delayed due to the heavy workload and busy schedule of marketing tasks. However, AI is now handling a significant portion of content creation, data organization, and repetitive work. Therefore, it seems that the excuse "I couldn't do it because I was busy" will become increasingly unacceptable.
now is not the ability to do everything themselves, but the capability to identify repetitive manual tasks and design work that can be automated.
Regarding automation, rather than approaching it with the goal of building a finished system from the start, it is more realistic to first put existing manual tasks into an automation loop and continuously improve them at the team level. The crucial starting point here is the brief. You must be clear about what to build, why to build it, and what metrics will be considered success so that the results remain stable even when utilizing AI. If goals and metrics constantly shift, speed may increase, but the direction may actually become unclear.
Which mountain to climb becomes more important! @marketer.ai.seulki
Q. There are cases where even planning is entrusted to AI; what decisions must marketers take full responsibility for until the end?
A. You must take responsibility for the experiment itself—defining the brand's problem and testing the hypothesis—even for the failure.
Marketing begins with formulating a hypothesis, even when there is no certainty, such as, "The problem with this brand right now is likely this." While AI can organize data and suggest various action plans, it is ultimately up to humans to decide which problem to solve. If a hypothesis is vague, it is difficult to identify what went wrong when a campaign fails.
In the AI era, most brands will be able to create more content faster. Therefore, it is difficult to differentiate yourself simply by producing a large volume of blog posts or shorts as in the past. Ultimately, what matters is how accurately you define the problem and how honestly you interpret the results. You must verify whether your hypothesis was correct, or if it was incorrect, which premises were wrong, in order to improve your next steps.
Q. Even if AI assists with execution, retrospectives seem to be a different matter. How can retrospectives be conducted for inexperienced juniors or projects that are difficult to judge?
A. That is why I think the roles of seniors and the organization are becoming more important. There must be a culture where failure can be honestly acknowledged for genuine insights to remain.
Inexperienced juniors are prone to accepting a positive evaluation of a campaign as a genuine success, and accepting a negative evaluation as a failure without question. Therefore, in the early stages, seniors should look beyond simply evaluating the results and examine how the problem was defined, what hypotheses were formulated, and what was verified during the execution process .
Both agencies and brands often find it difficult to directly discuss failures within their organizations. When reporting, they tend to emphasize only the result—such as "conversion rates improved by 1.3 times"—while omitting the fact that the campaign's fundamental hypothesis was actually incorrect. If this pattern is repeated, no insights will remain within the organization. Given the shortened execution cycle, it is crucial to review performance on a weekly basis; however, it is important to document the hypothesis and lessons learned alongside the numbers, rather than reporting only the numbers.
Q. In what order should a marketer just starting a career build their expertise?
A. Currently, it is important to have the experience of directly solving problems within an organization while holding authority and responsibility. Proactive problem-solving is more important than learning and mastering technical skills.
In the early stages of your career, whether new hires are granted actual decision-making authority. If you are limited to repetitive and restricted tasks at your first company, it may be difficult to work proactively even if you later move to a rapidly changing organization. Conversely, gaining experience in defining problems, selecting tools, and evaluating results during the first two to three years builds the fundamental skills to respond to change.
Between the next three and seven years, you can build a portfolio at well-known companies or on projects, and after seven years, you can expand into roles where you can demonstrate leadership. The key is not to spend a long time mastering the use of specific tools. It is more important to continuously accumulate experience in formulating, executing, and reflecting on strategies in a changing environment.
Q. What kind of system is needed for the know-how or expertise members have gained through AI to become an asset for the organization?
A. This is also the most difficult part of organizational AX: recording and documentation. Not only overall work but also individual experiences must be converted into data step by step.
The biggest bottleneck in AX conversion is often documentation rather than technology. If the schedule, verification steps, and systems and data utilization from campaign planning to completion are not organized step-by-step, it is difficult to entrust automation to AI. In practice, marketing workflows often remain only in the hands-on experience of those in charge and are not documented.
If work processes are not recorded, onboarding becomes difficult, and data and know-how are lost along with the employee when they leave. There are also instances where past performance and ideas cannot be reviewed because Excel files managed by individuals are lost. Therefore, it is crucial to store work processes and historical data in a single solution and ensure that anyone can verify previous decisions and results.
If you want to assess your organization's current AI adoption status and future direction, refer to Seulgi's article, '4 Stages of AI Adoption'!
Q. As the way individuals work has changed, it seems there are many things that organizations need to change as well. Where should companies invest more right now?
A. I think we really need to think about sustainability now. Instead of getting bogged down in short-term results, we must invest in systematically accumulating the organization's data and work context.
As the pressure to generate revenue within two months and produce results within three increases, organizations are more likely to postpone establishing a long-term foundation. This is a major reason why work systems are often not documented and data is not accumulated.
Short-term results are important, but you must also consider whether the execution process serves as a foundation for future tasks. To automate work and focus more on performance, the organization must manage what was done, what results were obtained, and what decisions were made prior to that process.
Q. What characteristics do you see in companies or organizations that are successfully implementing AX?
A. Organizations with dedicated AX departments actually change more slowly. The companies currently leading the way start by creating an environment where individual members can immediately utilize AI in their work.
Simply having a dedicated AX department introduce tools and encourage their use does not effectively bring about change in the field. Organizations that leverage AI effectively empower all members to use AI tools to directly improve their own work, while the development team takes on the role of supporting deployment and technical advice. The key is not to separate the use of AI into a project for a specific department.
However, the issue of individual-created automation tools becoming isolated within the organization must be addressed. Governance roles are necessary to manage tools used jointly by teams and the company, and to establish standards for workflow and data connectivity. While AX starts with improving individual productivity, it can yield greater results when transformed into an organizational asset.
Q. Conversely, what are the characteristics of organizations that do not achieve results even after introducing AI?
A. It would be an organization that infinitely increases the volume without sufficiently defining the customer's problems and questions.
As an example, content creation is a repetitive task for marketing teams. You might expect traffic to increase simply by publishing blog posts daily using AI. However, it is difficult to drive customer choices based on the quantity of content alone. What is important is to thoroughly understand what questions customers ask the moment they consider a service, and what information and messages potential customers respond to.
Automation only yields results if you first organize customer questions and problems, and then design the appropriate content specifications and production processes. If you compete solely on quantity without understanding the customer, AI will merely produce content in the same direction faster.
Q. You mentioned that records and data are important for AI. What should teams lacking marketing assets build first right now?
A. You must first connect basic behavioral data—from customer acquisition to purchase, repurchase, and churn—and create an environment where it can be quickly verified.
Early-stage organizations often lack organized GTM event tags or metric measurement environments, and sometimes fail to fully utilize basic tools like Google Analytics or Google Search Console. While these issues may not become apparent during rapid business growth, a lack of data emerges to determine what needs improvement once growth slows.
The first thing needed is an environment that connects where customers came from, what messages they responded to, what their experience was after their first purchase, and how repurchase and churn occurred. It is also important to create dashboards and alerts to quickly identify anomalies. Since search and website visit data only show customers who have already developed a certain level of interest, it is necessary to examine surveys and external data as well to understand their more advanced needs.
Q. What principles must brands adhere to to be chosen by customers even in the age of AI?
A. Instead of making AI automation or the technology itself the goal, you should place understanding what customers think and want at the center of your marketing.
There are instances where AI automation or GEO is launched as a new project following a specific event, such as a marketing conference or seminar. However, if you spend time chasing trends and adopting fad technologies like this, you may end up losing the time to actually listen to the voice of the customer and build a positive brand impression.
In fact, from a customer's perspective, what AI tools a brand uses is not a crucial reason for purchasing. The priority is understanding what problems customers are facing, what messages they respond to, and why they should choose our product. Technology should be a means to help execute this deliberation more quickly and accurately.
Q. As you prepare your marketing strategy for 2027, what is one thing the organization can do right now?
A. Work backwards on this year's projects and draw workflow diagrams for individuals, teams, and the organization. Then, identify any duplicate tasks or gaps and implement them using an AI system.
The starting point is to organize the tasks, durations, solutions used, and necessary data for each step by working backward from the campaign launch. Once each team member creates their own workflow and combines them at the team level, you may identify tasks that are being repeated by multiple people or essential work that no one has assigned.
By extending team-based workflows to the entire organization, you can identify both unnecessary tasks and those that need to be newly created. Moving beyond a mere document to implementing this flowchart into an actual business system connected with data and tools is the first step toward becoming an AI-native organization.
