# What should recommendations do when the product they would suggest goes out of stock? | KilterMuse

Source: https://kiltermuse.com/blog/2026-10-17-recommendations-out-of-stock-product/

KilterMuse answers


October 17, 2026

**Short answer:** A recommendation is a promise that this product is right for you and available now. When the suggested item is sold out, the promise breaks twice: the shopper cannot buy it, and the store looks like it was not paying attention to its own shelves.

## Check stock where the recommendation renders

Inventory known at the data layer and inventory shown on the page drift apart whenever feeds lag. The render layer is where the shopper meets the recommendation, so that is where the final stock check belongs, with size and variant level accuracy for products where one size is the whole decision.

Variant detail matters more than product level flags. A shoe in stock in every size except the shopper size is out of stock for that shopper, and recommendations that ignore the distinction send people to a page whose only job is to disappoint them.

## Substitute by intent, not by category

When the first choice is gone, the best substitute is the product shoppers actually took instead: the same need, a comparable price, and a similar look or specification. Co purchase and co view data usually knows this answer better than a category rule that grabs the nearest item on the shelf.

Say what happened, briefly. A note that the original pick sold out and this is the closest match reads as service. Silently swapping in a weaker product reads as a bait, and shoppers who land on the product page can tell the difference in seconds.

## A stockout is a signal, not just a gap

The product a system wanted to recommend going out of stock tells the merchandising team two useful things: demand models liked it, and supply did not keep up. Feed repeated recommendation stockouts into buying and allocation reviews, because they are a ranked list of missed sales written by your own engine.

For the individual shopper, offer the capture: a back in stock alert tied to that product and variant. The alert converts a dead end into a future visit, and its sign up volume is the cleanest measure of how much demand the stockout stranded.

## Keep some recommendations stock proof by design

Anchor slots can favor products with deep stock and stable availability: proven sellers the store can actually fulfill at volume. Save the fragile, fast moving products for contexts where a stockout is less costly, such as editorial content, rather than the checkout adjacent slots where a dead recommendation blocks a sale in progress.

This is a portfolio decision. Every recommendation slot mixes persuasion value against stockout risk, and different slots should hold different mixes instead of one global rule pretending all placements matter equally.

## Measure stranded demand, not just clicks

Track how often rendered recommendations were out of stock, how substitutes performed against the originals, and how many alert sign ups each stockout produced. Those three numbers together tell you the cost of the problem and whether the substitution logic is earning its keep.

Review the worst offenders monthly. A short list of products and variants usually accounts for most stranded demand, and fixing availability or feed freshness for that list recovers more revenue than another round of algorithm tuning.

Keep reading: [How do you personalize email product blocks without rebuilding every send?](https://kiltermuse.com/blog/2026-10-17-personalize-email-product-blocks-sends), [How do you personalize replenishment reminders so shoppers reorder at the right moment?](https://kiltermuse.com/blog/2026-10-17-personalize-replenishment-reminders-reorder-timing), and [How do you personalize product recommendations during a new product launch?](https://kiltermuse.com/blog/2026-10-17-personalize-recommendations-new-product-launch) cover the nearby cases in this series.

## Frequently asked questions

### Why do out of stock recommendations still appear?

Because inventory data and the page the shopper sees drift apart when feeds lag or stock is only checked at product level. The check belongs at render time, at the size and variant level the shopper actually needs.

### What makes a good substitute recommendation?

The product shoppers actually chose instead of the sold out item: same need, comparable price, similar specification or look. Behavior data finds it more reliably than a simple same category rule, and a brief note explaining the swap builds trust.

### Should stores hide that the first choice sold out?

No. A short, honest note that the pick sold out and this is the closest match reads as service. A silent swap to a weaker product reads as a bait once the shopper reaches the product page.

### How should merchandisers use stockout data?

As a ranked list of stranded demand. Repeated recommendation stockouts, plus back in stock alert sign ups, show buying and allocation teams exactly where availability is costing sales.