Nội dung
- Synthetic data: a new path when traditional customer research is slow and costly
- Agentic AI is changing martech economics and technology infrastructure
- The most valuable AI skill may only take 10 minutes a week
- When best-of-breed stacks become too complex to manage
- A perspective for the Vietnamese market
- References
AI is not only changing how marketers create content or run ads, but also directly affecting customer research, system architecture, and martech operating costs. Four new MarTech articles paint a fairly clear picture: businesses that want to move fast with AI must learn to use synthetic data, rethink agentic AI models, keep tools updated, and control the complexity of integrated stacks.
For Vietnamese marketers, these are very practical issues: from doing market research faster in a context where real data is hard to access, to the “AI investment” problem where integration and operating costs rise beyond expectations.
Synthetic data: a new path when traditional customer research is slow and costly
The first MarTech article highlights a familiar paradox in modern marketing: businesses need to understand customers more deeply, but traditional research methods such as surveys or focus groups are becoming slower, more expensive, and less able to cover hard-to-reach customer groups. In addition, privacy requirements and consent rules also make it more difficult to access detailed data.
In that context, synthetic data is seen as a way to shift the focus from “collecting more data” to “creating more useful insights.” According to MarTech, AI can generate synthetic data with statistical representativeness, reflecting the characteristics of the real dataset. As a result, marketers can test audience reactions, validate communication ideas, or evaluate scenarios before spending budget on campaigns, creative content, or product development.
The notable point is that synthetic data does not fully replace field research, but acts as an early testing layer. For decisions that require speed, such as adjusting messaging before launch, testing a product concept, or redesigning the customer journey, it can be a tool that reduces risk and supports more informed decisions.
Source: MarTech — Where synthetic data fits into customer research
Agentic AI is changing martech economics and technology infrastructure
The second article shows a very notable shift in the AI market: as vendors move from a “pay as you go” model to token-based pricing, the cost of complex AI tasks can rise much faster than initially expected. This is especially important as agentic workflows — AI-driven processes that can execute multiple steps automatically — begin to become part of everyday marketing work.
MarTech explains that when AI is connected to business systems, it is no longer just a chatbot answering individual questions. It can pull data from CRM, analyze campaign performance, search the web, and generate personalized reports within the same workflow. That power comes from tool calling, meaning AI’s ability to access external systems via APIs or protocols such as MCP.
But the more tools are called, the more tokens are consumed. AI agents also use significantly more tokens because they must continuously carry task history, internal reasoning, and tool data at each processing step. From an operations perspective, this creates major pressure for martech teams: if the infrastructure is not designed properly, AI can increase costs faster than productivity.
Source: MarTech — Agentic AI is rewriting martech economics and infrastructure
The most valuable AI skill may only take 10 minutes a week
In the third article, MarTech pushes back against a common habit: trying to “lock in” one AI tool, learn it deeply, and stay loyal to it for the long term. According to the author, the pace of AI evolution today is too fast to maintain the kind of loyalty to a single platform that was possible before.
Instead of looking for the perfect tool, the more important skill is recognizing when a new technology truly changes what is possible. That skill does not require becoming a full-time AI expert; it mainly requires staying updated regularly. MarTech suggests a simple rhythm: every week or two, spend about 10 minutes scanning what is new, enough to spot which features, models, or approaches are opening up real opportunities.
This message is especially relevant for marketers and operations teams in Vietnam, where many businesses are still in the AI experimentation stage. In that phase, the advantage is not in memorizing one tool, but in knowing when to try something new, drop something old, and integrate AI into workflows flexibly.
Source: MarTech — The most valuable AI skill takes 10 minutes a week
When best-of-breed stacks become too complex to manage
The final article goes straight to a familiar marketing operations problem: should teams choose the best tools for each job, or use a more unified ecosystem to reduce complexity? MarTech describes this as the moment when “best-of-breed” runs into a “complexity wall.”
Every custom API connection between legacy CRM systems and modern AI tools is a potential point of failure. It creates technical debt and requires teams to monitor, patch, and update continuously. For many marketing ops leaders, the time spent keeping these integrations running is starting to eat into time that should be devoted to strategy, growth optimization, or improving customer experience.
The lesson is that when evaluating tools, subscription fees alone are not enough. Total cost of ownership must be considered, including the “integration tax” — meaning technical time, maintenance effort, and operational risk. This is a very necessary perspective for businesses expanding their martech stack without strong enough integration capabilities.
Source: MarTech — When best-of-breed stacks become too complex to manage
A perspective for the Vietnamese market
Taken together, these four stories point in the same direction: AI in marketing is entering a “real operations” phase, where value no longer lies in trying a chatbot for fun, but in generating insights, automating workflows, and optimizing total system costs. For Vietnamese businesses, this is the time to move from scattered experiments to evaluating AI as part of the growth infrastructure.
- For market research: synthetic data can help test ideas quickly before large-scale surveys, but it still needs to be compared with real-world data and the context of Vietnamese users understood.
- For martech operations: measure token costs, integration costs, and maintenance costs from the start, rather than looking only at the tool’s purchase price.
- For marketing teams: maintaining a short but regular AI update habit will be more effective than chasing every new tool in a scattered way.
- For business leaders: a system architecture that is lean enough for AI to create value is needed, rather than piling on tools that leave teams overwhelmed by connections.
In other words, the AI game in marketing is not just about choosing which model to use, but about choosing how to organize data, processes, and technology in a sustainable way.
See more marketing news and guides at https://marketing365.vn.
Follow more updates from Marketing365 to stay up to date with the latest marketing trends.
Read more articles in the Digital Trends category.
This article focuses on AI trends with a perspective for the Vietnamese market.
References
- Where synthetic data fits into customer research
- Agentic AI is rewriting martech economics and infrastructure
- The most valuable AI skill takes 10 minutes a week
- When best-of-breed stacks become too complex to manage
