Nội dung
- AI is excellent at optimization, but only within the scope of the goal it is given
- When metrics improve but the business may not
- Don’t just give automation a direction; set boundaries too
- What to turn off before turning AI on
- Automating a guess is still just a guess
- A perspective for the Vietnamese market
- References
Automation and AI are helping marketers cut down on a great deal of manual work: from ad bidding and lead scoring to suggesting next actions in CRM. But the lesson highlighted by MarTech is especially noteworthy: technology can optimize extremely fast, but it cannot fix a vague objective on its own. For Vietnamese marketers, this matters in particular when many campaigns are still assigned with phrases like “increase ROAS,” “increase sign-ups,” or “reduce CAC” without clear boundaries.
AI is excellent at optimization, but only within the scope of the goal it is given
According to MarTech, today’s ad platforms, CRM systems, and AI assistants can automatically manage many tasks and provide increasingly useful recommendations. However, they cannot “read business intent” if the input objective is too broad. When given a single metric, the system will try to improve that metric in the most efficient way possible, even if that path does not create real value for the business.
In other words, automation does not make a strategic problem clearer. It only makes the wrong decision happen faster and look more convincing.
When metrics improve but the business may not
The article offers many very practical examples. If you ask AI to raise ROAS, the system may prioritize audiences that are easy to convert, such as customers already familiar with the brand, branded search, or remarketing groups. ROAS does go up, but new growth may not be significant.
Similarly, if you simply push for more sign-ups, a campaign may generate many low-quality leads; and if you force CAC down, the system may narrow its reach to focus only on the easiest-to-convert group. The common pattern is this: the metric improves, but the business objective may drift off course.
This is where marketers need to stay especially alert. A beautiful dashboard does not mean a good campaign. If you only look at output metrics and ignore growth quality, AI can help optimize “how to measure,” but not “what needs to be achieved.”
Don’t just give automation a direction; set boundaries too
MarTech’s most important message is this: a vague goal is not a goal. The phrase “we need higher ROAS” is really only a direction, not a destination that can be safely operated toward. Automation needs a clear “playing field,” meaning it must know not only what counts as success, but also what threshold means failure or requires a pause for review.
The article gives the example of a brand running ads with an 8x ROAS. If the real goal is to acquire more new customers, the business may accept ROAS dropping to 5x as long as new-customer volume rises accordingly. But a floor must be defined in advance; if performance falls below that level, the campaign must stop and be reassessed. That way, AI can proactively expand audiences and test lower efficiency in exchange for meaningful growth.
The key point is that the trade-off must be decided by humans before the machine starts optimizing. Otherwise, the team may either tighten the campaign too much to protect a number that is no longer relevant, or let the system burn budget without anyone having set a stop point.
What to turn off before turning AI on
MarTech also emphasizes another layer of the issue: not every AI feature is safe to switch on. In industries with legal or compliance constraints such as insurance, healthcare, or finance, allowing AI to expand too freely can create risks that the dashboard will not warn you about in advance.
For example, a system may rewrite ad copy in a way that no longer fits internal approval processes, or expand brand queries into approaches that should not appear. Therefore, before turning on AI Max or any similar automation mechanism, the business needs to clearly decide what must be locked down: text customization, brand keywords, or targeting groups that must not be touched.
This is not a sign of distrust in technology. On the contrary, it is a way to create an operating framework that is safe enough for AI to deliver its power within the allowed scope.
Automating a guess is still just a guess
In the CRM section, the article notes that many automated workflows are built very carefully: triggering emails, assigning tasks, sending reminders, and nurturing customers step by step. But the fundamental question is often overlooked: does that action actually produce a better result?
If there is no data proving that customers who do action A are more likely to stay, then automating process A simply turns an assumption into a recurring workflow. In other words, automation does not make a hypothesis more true. It only makes that hypothesis run faster, more consistently, and with less effort.
For marketers, this is an important reminder: before automating, check whether you are optimizing something that has already been proven, or merely scaling an untested belief.
A perspective for the Vietnamese market
In Vietnam, many businesses are entering the phase of “turning on AI quickly so they don’t fall behind,” but many teams still lack a sufficiently tight goal framework. In that context, the lesson from MarTech is highly applicable: don’t just give the system a single KPI and expect it to understand the company’s growth strategy.
Instead, Vietnamese marketers should define three clear layers in advance: the primary goal, the acceptable threshold, and the stop condition. For example, when running performance campaigns, it should be clear which kind of growth is prioritized: new customers, profit margin, or order size. When using AI for CRM, it is necessary to verify which workflows truly affect retention, rather than automating everything just because it “looks effective.”
As AI becomes more powerful, the advantage does not lie in automating the most, but in knowing how to set the right boundaries for automation. Businesses that clarify their goals before handing work over to machines will be the ones that use AI better, more safely, and more sustainably.
See more marketing news and guides at https://marketing365.vn.
Follow more updates from Marketing365 to stay on top of the latest marketing trends.
Read more articles in the same category Digital Trends.
This article focuses on marketing automation with a perspective for the Vietnamese market.
