When Ad Creative Gets Cheap, the Scarce Skill Is Knowing What Not to Say

For most of the years I ran paid acquisition, the limiting factor was never the media budget. It was creative. Every new angle, every new hook, every seasonal variation had to be briefed, shot or designed, copywritten, and approved before it could run. Across more than a decade of building e-commerce businesses, I put more than NT$400 million into paid channels, and the single most common reason a campaign underperformed was not that we picked the wrong audience or the wrong bid strategy. It was that we simply did not have enough good creative in the pipeline to test, so we kept running the same few ideas long after they had gone stale.

That constraint is gone. Any marketing team today can generate more ad variations in an afternoon than my teams used to produce in a quarter. Headlines, images, video cuts, landing page copy: all of it can be produced at a volume that would have been unthinkable when I was building those businesses. This is a genuine change in what a marketing team can do, and I do not think most teams have absorbed what it actually means for how they should spend their week.

The instinct is to treat the freed-up hours as more production capacity. Make more variants, test more angles, ship more often. I think this is close to the opposite of what should happen. When creative was expensive, the cost itself acted as a filter. Something had to be worth building before you built it, and that forced a team to think hard about the customer before a single asset got made. Take that cost away and you remove the filter, not the need for it. The team that used to spend most of its week producing things now has to spend that time deciding what any of it should say, and very few teams have actually built that muscle, because they never had to.

The concrete change I would make is where the freed hours go. They should not go into more production. They should go into two things a cheap-creative team tends to neglect: direct contact with customers, and an explicit, written list of claims the brand will not make. On the first, I mean actual conversations, actual reading of support tickets and returns and reviews, done by the people who write the briefs, not summarized for them secondhand. Knowing the customer well enough to write a brief that says something true and specific is a slower kind of work than generating variants, and it does not get easier just because production got faster. On the second, unlimited variants make honest positioning harder, not easier, because nothing stops a team from generating a claim that tests well and happens to overstate what the product does. When creative was scarce, a weak claim often died from lack of resources to produce it. Now it survives unless someone deliberately kills it. A short, specific list of what the brand will not claim, reviewed before a batch of variants goes out rather than after one goes viral for the wrong reason, is boring work, but it is the discipline that scarcity used to provide for free.

In practice this means restructuring the week so that a fixed block of time, not whatever is left over, is protected for customer contact and for reviewing what the team is about to say, before the volume goes out. The teams I built at their peak had more than 200 people, and the ones that performed best were not the ones that produced the most creative. They were the ones where the people writing briefs actually understood the customer, and where someone was willing to say a claim was not going out even though it would probably work. That judgment does not scale automatically just because production does. I now spend most of my own operating time on AI-enabled operations across a portfolio of sites, and the lesson has been consistent across very different businesses: production speed is not the same skill as knowing what deserves to be said, and treating them as one skill is where I have seen teams go wrong.

I should be honest about the limit of this argument. I am describing a pattern from paid acquisition and e-commerce, and I have not run this exact playbook inside a large in-house brand marketing organization, so the specific weekly structure I am proposing may need real adjustment outside the context I know best.