English
Welcome to the next episode of What AI Actually Does in Your Store. Last week we asked whether your store actually personalizes or just segments. This week about how AI scales guided selling.
This series is about what AI does in online stores, based on what e-commerce managers ask, worry about, and get wrong.
E05: How AI handles shoppers search can’t help
A mom buying her son’s first football boots doesn’t know where to start.
She needs someone to ask her the right questions. That’s what a shopping assistant does.
It’s a guided experience that asks specific questions and narrows down the options with each answer.
The questions are tied to filters. Click “my son plays on artificial grass,” and the catalog filters to the right stud type.
By the end of the dialogue, the shopper has one or a few products that actually fit.
Guided selling has existed for years. What AI changed is how fast you can build one.
New workflow
Before LLMs, building a shopping assistant meant writing questions manually, mapping each to catalog attributes, and testing the whole thing. For large catalogs, it could take weeks.
Now you take the “How to choose football boots” article that’s been on your blog for years and generate a draft tree of questions, answers, and filter mappings.
You review it, adjust where needed, and publish.
If you sell across multiple markets, you can even duplicate the assistant and localize it per domain without rebuilding the whole structure from scratch.
Now do it for every category
Most stores that could benefit from guided selling don’t have it.
Before AI, building one properly was expensive and slow relative to the return.
One well-built assistant for one category could take a developer and a content person a week.
LLMs change that math.
A store with twenty categories that need guided selling can now build all twenty without a development project.
The bottleneck shifts from building to deciding which categories actually need it.
Guided selling doesn’t fit every category. If your shoppers come knowing what they want, search beats a quiz. But what counts as search is changing.
Next week: the season finale. The difference between the AI search that’s been running for years and what’s actually coming next.
