The most common anxiety I hear from merchandisers when I describe what Curated For You does is some version of: "So where does that leave me?"
It's a fair question. If the product grid is being rebuilt per visitor automatically, if the sort order is updated every session without manual intervention, what is the merchandiser actually doing?
The honest answer is: the same things they were always supposed to be doing, but without the parts that were never a good use of their time.
What Merchandisers Are Actually Good At
Before getting into what changes, it's worth being specific about what a skilled merchandiser actually brings to a retail operation.
A good merchandiser understands the product assortment at a granular level. They know which items are trend-driven versus core staples, which SKUs have margin issues that affect how aggressively they should be promoted, which vendor relationships come with commitments that require specific products to get page-one exposure. They understand the seasonal calendar and how it interacts with inventory levels. They can look at a product image and predict, with reasonable accuracy, whether a particular customer segment is going to respond to it.
None of that knowledge lives in a database. It's judgment built from experience, vendor conversations, and watching what happens over time when specific decisions get made. That judgment is not something a personalization model replaces. The model doesn't know that you've committed to moving 500 units of a particular jacket before the end of the season. The model doesn't know that a vendor relationship depends on a specific brand getting consistent first-page exposure. The model doesn't know that your new candle line is positioned to hit a trend that will peak in Q4.
Business logic is still human work. It always will be.
What Merchandisers Were Doing That They Shouldn't Have Been
But here's the part worth being honest about: a significant portion of what merchandisers at mid-size retailers actually spend their time on is manually maintaining category sort orders and product positioning that could be automated much more accurately.
Weekly sort order reviews. Monthly manual promotions of specific SKUs to page one. Manual demotion of slow-moving items. These tasks exist because the platform doesn't do them automatically, and someone has to make the store look like it was arranged with intention. A skilled merchandiser can make good decisions in these reviews. But they're working with category-level aggregate data: which products are selling, which have high return rates, which have inventory issues. They're not working with per-shopper signal. They can't be, because per-shopper signal at the individual level isn't visible to a human reviewing a category dashboard.
The result is that most manual category management is a reasonable approximation of what a well-personalized category page would look like for the average shopper. For repeat buyers with strong taste profiles, it's still a one-size-fits-all answer.
The Shift: From Maintenance to Strategy
When personalization handles the per-session sort order, the merchandiser's role shifts. The time that was spent on manual sort maintenance gets redirected toward decisions that benefit from human judgment.
Override rules are one of these. The personalization model is good at surfacing what a specific shopper is likely to buy based on their history. It is not good at knowing which products need fixed placement for business reasons. Merchandisers define those override rules: "This vendor's new collection gets position 1-3 on the Women's category page for the next three weeks." "This clearance line needs priority placement until inventory drops below 50 units." "Our collaboration with this brand gets featured positioning regardless of individual shopper history."
Assortment decisions are another. Which products to carry, which to discontinue, which new items to introduce: these are still merchandising calls. A personalization model can surface which product attributes are resonating within different shopper segments, providing data to inform those decisions. But the decision itself, including the vendor relationships and margin considerations that go into it, is human work.
Category taxonomy is a third area. How categories are structured, how broad or narrow product groupings are, which items live in which sections: these decisions shape the data the personalization model has to work with. A merchandiser who understands their assortment deeply can structure categories in ways that make personalization more effective.
We built Curated For You from a merchandising background, not a machine learning background. That distinction matters. The tools a merchandiser needs to interact with a personalization system are different from what an ML engineer needs. The goal was to put meaningful controls in the hands of the person who knows the business, not to create a black box that requires technical interpretation.
What the Toolkit Looks Like in Practice
For a merchandiser working with Curated For You, the day-to-day workflow looks roughly like this:
The model handles session-by-session product ranking within each category. The merchandiser sets and maintains override rules for products that have specific placement requirements. The merchandiser reviews category-level performance reports that show how different shopper segments are responding to which product attributes, which informs assortment planning. And when something looks wrong, whether a new product isn't getting the exposure it needs or a seasonal item is underperforming, the merchandiser has the controls to intervene.
The manual weekly sort review goes away. What replaces it is a higher-level monitoring task: watching the signals the model surfaces, setting rules where business requirements demand them, and using the freed-up time for the assortment and vendor work that was always the more valuable use of a merchandiser's expertise.
The Fear Is Real, the Conclusion Is Wrong
I understand why merchandisers look at personalization tools and worry about their role. The narrative around automation in retail has not been kind to people who work in operations, and a lot of that narrative overstates what current tools can actually do.
The merchandisers who work best with Curated For You are the ones who understand their assortment deeply enough to set smart override rules and interpret performance data meaningfully. That depth of knowledge doesn't become less valuable when the sort order is automated. It becomes more valuable, because the tool amplifies good merchandising judgment and makes poor judgment more visible.
A merchandiser with strong product knowledge and a working personalization system will consistently outperform a team relying on manual sort order alone. That's the actual outcome. Not replacement. Amplification.