There is one perspective on AI agents that I strongly agree with: rather than listening to what they say, pay closer attention to what they do. Apply that yardstick to marketing, and things get a little uncomfortable.
When AI mentions your brand in an answer, is that a sales lead—or a digital round of applause?
I suspect that one of the marketing outcomes most likely to be overrated next is “having your brand appear in AI answers.” Where the name appears, the reasons attached to it, and what the user does next can make an enormous difference in commercial value.
Your Name Makes the List, but the Order Goes Next Door
Suppose you run an air conditioner cleaning business in Taichung. Someone asks AI: “We have cats at home, we can only have the work done on Saturday, and we don’t want the whole house smelling of cleaning chemicals afterward. Any recommendations?”
AI lists five businesses. Your brand ranks first, described as offering “professional service, affordable prices, and excellent reviews.” But the fourth business says: “Saturday appointments available, cleaning products publicly listed, and arrangements for pets confirmed before work begins.”
If I were that pet owner, I would click on the fourth one first.
The first business gets exposure; the fourth gets an inquiry. Yet the month-end report might highlight the first-place ranking in green and label it “increased AI visibility.”
This is the assumption I want to challenge: The more often AI mentions a brand, the closer that brand gets to a sale. There is actually a missing step: What reason does it give for connecting you with this person’s needs?
“Professional and Reliable” Is Safe—and a Tough Sell
Brands have long competed over adjectives: premium, attentive, one-stop. In advertising, these words can build an impression through photos, tone, and repetition. But in an answer comparing options, they do little to explain why you are right for the business at hand.
Imagine two photographers who take corporate assignments. One has “capturing the soul of your brand” on their homepage. The other publishes turnaround times, the scope of commercial usage rights, and a case study on handling reflective metal during a factory shoot.
When a buyer asks, “We need product photos next week, and the images will be used in overseas catalogs,” the latter at least provides verifiable reasons for a recommendation. That does not guarantee AI will choose them. But the former’s polished copy certainly does little to address the buyer’s constraints.
My concern is not making websites read more like spreadsheets. It is putting information that customers would otherwise have to call and ask about somewhere they can find and understand it, ahead of time.
Sometimes the Ugliest Numbers Are the Most Useful
Marketers naturally have reasons to track mention rates. Sales take time, and complete data may not be available; visibility is one of the few signals they can observe early on.
But if budgets follow that number, they may start mass-producing content that is easy to quote but does nothing to help people choose. The brand name appears more frequently, while the reason for recommending it remains nothing more than “a good reputation.”
I would want to break the report down: What kind of need was the question about? Did the answer accurately describe the service limitations? Can someone complete a booking on the linked page? These questions are closer to actual business than a flattering overall score.
And clear information comes at a cost. Stating that you do not accept rush jobs or serve a certain area may immediately cost you a recommendation. To someone chasing exposure, that is a lost point. To a business owner fielding unsuitable inquiries every day, it might mean half an hour saved.
At ten that evening, the air conditioner cleaning business owner opens the dashboard and sees just one inquiry for the day. The customer writes: “I saw your explanation of the cleaning process for homes with cats. I’d like to book Saturday.” Beside it, the monthly report still displays another number: brand mentions are down from last week.