The Similar Choices tab lets you explore up to 20 products that are similar to any selected choice, all without leaving Assortment Plan. Use it to identify like-for-like substitutes, spot gaps in your assortment, and make more informed planning decisions backed by similarity signals.
Use Cases
Finding comparable products when planning a new choice or placeholder.
Validating that a new style fills a gap rather than cannibalizing existing choices.
Reviewing attribute and price overlap before finalizing an assortment.
Flagging irrelevant results as
Not Similarto improve future recommendations.
Accessing the Similar Choices Tab
In Assortment Plan, select the choice you want to explore.
Click the arrow to open the
Offering Setbottom sheet.Or double-click the Similarity Index value.
Navigate to the
Similar Choicestab.
What You'll See
The tab displays up to 20 similar choices alongside three summary metrics and per-choice similarity signals.
Similarity Index
The Similarity Index is the average similarity score of the top 5 most similar choices. It represents the overall strength of the similarity signal for the selected choice — a higher score means stronger, more reliable matches.
Average Similarity
Average Similarity is the average score across all 20 returned choices. Use this alongside the Similarity Index to understand whether strong matches are concentrated at the top or spread across the full list.
Number of Products
By default, the tab returns 20 choices. If placeholders are hidden (toggled off from the three-dot menu in the top left), fewer choices may appear since hidden placeholders are excluded from results.
Per-Choice Similarity Signals
Each returned choice displays three similarity signals:
Signal | What It Measures |
Descriptive Similarity | Semantic similarity across product title and category values — matches meaning rather than exact keywords |
Attribute Similarity | Overlap across your configured product hierarchy attributes (e.g., up to 3 levels depending on your configuration) |
Price Similarity | How close ticket price, retail price, and cost are between the two choices |
How Similarity Index Is Calculated
The Similarity Index combines three inputs:
Descriptive similarity — semantic similarity across product title and category values. This uses meaning-based matching rather than exact keyword overlap, so products with similar descriptions surface even when the wording differs.
Attribute overlap — overlap across your configured product hierarchy attributes. The number of attribute levels used depends on your hierarchy configuration (e.g., up to 3 levels).
Price proximity — how close ticket price, retail price, and cost are between the selected choice and each match, not a single retail-only comparison.
A higher score across all three inputs produces a higher Similarity Index.
Marking a Choice as Not Similar
If a returned choice is clearly not a relevant match, you can mark it as Not Similar directly from the tab. This feedback helps improve future similarity recommendations for your assortment.
What's Coming Next
Similarity signals will expand across the platform in upcoming releases:
Like-for-Like Forecasting — Similarity Index will feed into the upcoming like-for-like forecasting feature.
Cannibalization — Similarity Index will also integrate into the upcoming cannibalization analysis feature.
UI-based attribute configuration — define custom attributes and assign weights to control how each signal contributes to the Similarity Index, going beyond the current product hierarchy levels.
FAQs
Why do I see fewer than 20 choices?
If placeholders are hidden from the three-dot menu in the top left of Assortment Plan, they are excluded from the results. Showing placeholders will restore the full set of up to 20 choices.
What's the difference between Similarity Index and Average Similarity?
Similarity Index is the average score of the top 5 matches and reflects the strongest similarity signals. Average Similarity covers all 20 results. If the two numbers are close, your matches are consistently strong. If Similarity Index is much higher, only the top few results are reliable matches.
Is Price Similarity based on current prices or price history?
Price Similarity uses a single point-in-time snapshot of each choice's ticket price, retail price, and cost, taken from the most recent weekly catalog rebuild — it doesn't average, blend, or time-weight prices. If a choice's ticket or retail price has changed over time, only whatever value is currently stored as of the latest snapshot is used in the comparison.
Can I undo a "Not Similar" mark?
Not Similar marks can be reversed — right-click the choice and select Mark as Similar to undo it.
What is semantic similarity?
Semantic similarity is about finding things that mean the same thing, even if they're worded differently.
Think of it like this: if you search your assortment plan for "women's sneakers," you'd also want to pull up styles tagged as "ladies athletic footwear" or "female trainers" — because they're talking about the same product, just described differently.
Why it matters for planning:
Smarter search — find the right styles, attributes, or reports even if you don't use the exact right term
Grouping like items — automatically clustering products or categories that are conceptually similar, even if their names differ
Q&A tools — when you ask an AI planning assistant a question, it uses semantic similarity to find the most relevant data or answer, not just an exact keyword match
A simple analogy:
It's like how you know that "sell-through" and "ST%" are the same thing, even though they look completely different written out. Semantic similarity teaches a computer to make that same connection — at scale, across thousands of terms, products, and data points.