How AI is changing where keyword strategy starts on Amazon

Every product launch starts in a learning phase. There's no click history, no converting search terms, and nothing for an optimization engine to learn from. Before it can improve performance, it first has to discover how shoppers actually search for the product.

For years, that's been the natural starting point. Advertisers do their homework before launch by reviewing competitor ASINs, digging through Brand Analytics, browsing the category, and applying what they know about the product, but everyone understands that the real keyword strategy won't emerge until campaign data starts coming back.

AI changes where that strategy starts. The advantage isn't that AI knows more than you. It's that it turns what you already know into a launch-ready list in minutes, so campaign data spends its time confirming your strategy instead of discovering it.

The Amazon Search Term Report still does something AI can't

Amazon’s Search Term Report remains the best view anyone has into how shoppers actually search. It captures real queries from real customers, including the unexpected phrasing no amount of category research would have predicted.

That data powers one of the most longstanding workflows in Amazon advertising. Converting search terms are harvested into managed keywords, while underperforming terms are eventually negated once enough evidence has piled up that they aren't working. Some advertisers manage that process manually. Others rely on automation. Perpetua's Keyword Harvesting and Negative Match Rules automate the same workflow, and nearly every serious optimization platform has its own version.

It's an effective system, and it isn't going away, but it has one requirement: it needs campaign data before it can do anything useful.

New launches start with no data, so the campaign has to learn first

The opening weeks of a launch are often spent building coverage that could have been mapped in advance. While manual research gives you a starting list of obvious terms, Auto campaigns are left to fan outward and discover the rest of your category's demand.

That exploration is valuable, but letting a campaign discover demand through spend takes time. Many of the high-intent search terms your campaign has to discover through trial and error are terms you could have targeted intentionally on day one. Until that discovery happens, your campaigns compete against only a fraction of the market demand available to you.

AI changes what pre-launch research can produce

AI is doing a great job at accelerating the pre-launch keyword research that advertisers already perform. Before every launch, an advertiser studies competitor products, reviews Brand Analytics, browses the category, and applies their own understanding of the product. The challenge has been turning that research into a comprehensive keyword strategy quickly enough that it shapes the launch.

Traditionally, this hasn't been the easiest process. The research was manual, scattered across different tools, and time-consuming enough that the resulting keyword list was treated as a starting point rather than a strategic asset. Once campaigns went live, the Search Term Report naturally took over.

With the introduction of AI into the process, instead of asking which keywords have already converted, advertisers can start with a different question: What should this product compete on?

Perpetua's AI Keyword Research is designed around that idea. Advertisers can start with a competitor ASIN or describe the product's features, use cases, and positioning. The system generates candidate keywords before a campaign ever runs, organizing them into head terms, long-tail terms, and exploratory opportunities. Each keyword is paired with Search Frequency Rank so advertisers can judge relative search demand, while existing campaign keywords are flagged to avoid duplication.

The output isn't a finished strategy. It's the same pre-launch research advertisers have always tried to do, completed in minutes instead of hours and organized in a way that's immediately actionable.

Expertise moves to the front of the launch

When generating hundreds of plausible keywords takes minutes instead of hours, the difficult part is no longer finding opportunities. It's deciding which ones deserve investment.

Some keywords justify meaningful budget from day one. Others are worth pursuing because organic ranking compounds over time. Some generate efficient advertising metrics without creating incremental growth. Others are worth watching but not funding.

Those decisions do not, and should not, come from AI. They depend on business objectives, category dynamics, and judgment specific to individual brands.

The expertise of the advertiser didn't disappear when keyword generation became easier. It simply moved earlier in the process, into deciding what the product should compete on before the first dollar is spent.

Campaign data becomes validation instead of discovery

Once campaigns go live, the Search Term Report confirms which assumptions were right. It surfaces customer language you couldn't have predicted. Harvesting expands coverage. Negative matching removes wasted spend. Bid and match-type optimization continue exactly as they do today.

Nothing about that process changes. Instead of originating your keyword strategy, campaign data validates it. It sharpens your assumptions, corrects the ones that were wrong, and uncovers opportunities no amount of pre-launch research could have predicted.

That's the shift AI makes possible. The best launches won't begin with a blank slate and wait for campaigns to reveal what matters. They'll begin with a well-informed hypothesis built from the product itself, then use campaign data to refine it.

The Search Term Report remains as valuable as ever. It simply does its best work confirming a strategy rather than creating one.