> ## Content Index
> Fetch the complete content index at: https://conversion.club/llms.txt
> Use this file to discover other available public pages before exploring further.

# How can product recommendations increase order value?
- URL: https://conversion.club/playbook/product-recommendations-order-value/
- Published: 2026-10-29T09:14:26.000Z
- Updated: 2026-10-29T09:50:01.000Z
- Description: Build genuine complementary pairings for bestsellers, place them where they help and compare revenue per visit and margin against a control group.
- Author: Jeremie Werner
- Tags: Increase average order value, #playbook, Cart, Shopify, GA4

Product recommendations raise order value when they offer a genuine complement where the shopper can still add it. Start with manual rules for bestsellers, one clearly defined block per placement and a control group. An expensive recommendation engine should follow a demonstrated need, rather than substitute for product knowledge.

## The benchmark

Baymard distinguishes alternatives, which help people choose, from complements, which help complete the purchase. In its 2014 study of 50 large stores, only 42% offered both on product pages. Its 2021 cart research found irrelevant suggestions on 52% of desktop sites; a single doubtful recommendation could cause participants to ignore the rest. These are dated research samples, not current market-wide prevalence estimates.

The source's October 2026 Shopify documentation snapshot allows up to 10 related and 10 complementary recommendations per product. New products without purchase history can receive collection-based suggestions that are too broad. The source says description-based automatic analysis was limited to English-language stores; verify current language and account support rather than applying that limitation universally.

| Placement         | Useful recommendation                                             | Avoid                                         |
| ----------------- | ----------------------------------------------------------------- | --------------------------------------------- |
| Product page      | Alternatives and complements in separately named blocks           | One mixed block with no explanation           |
| Add-to-cart/cart  | Relevant, lower-cost complements, especially required accessories | Alternatives reopening an already-made choice |
| Confirmation      | A complement or replenishment for a future order                  | A long unrelated list                         |
| No search results | Related products or previously viewed items                       | Only generic bestsellers                      |

Placement can matter as much as selection. A baby-products brand in Le Club showed its “Complete your order” block only at the bottom of product pages, while roughly 65% of visitors viewed no product page at all. A book accompanying its flagship item was absent from recommendations, although one customer in 10 bought it spontaneously. The proposed cart test had not yet run. That is a diagnosis and test idea, not a measured uplift.

## The rule

Manually define pairings for the 20–50 products responsible for most sales, then consider automation for the long tail. Choose the placement before the tool. Cart alternatives can reopen a completed decision; see [upsell versus cross-sell placement](https://conversion.club/playbook/upsell-or-cross-sell-which-goes-where/).

Explain the reason: “Wear with,” “Compatible refill” or “Required batteries.” Baymard's research supports labels that help people understand the suggestion.

Keep the selection manageable and optional. In a field audit of an intimate-health brand, the authors completed a sensitive diagnostic only to receive a €542.30 routine behind one “Add everything” action. The audit criticized the abrupt jump from personal disclosure to a large bill. Let shoppers select or deselect each item and see the total update. The amount is the source's observed euro example, not a benchmark basket.

## The steps

1. **Find useful pairs.** Inspect 12 months of orders for products bought with bestsellers, then validate that the association represents a real complement. Record contribution: a cheap accessory can raise order value but reduce profit if it fails to cover picking and packing. Use the [order-margin method](https://conversion.club/playbook/how-do-you-calculate-the-contribution-margin-of-an-order/).
2. **Write manual rules.** In Search & Discovery, configure product recommendations and decide whether manual choices replace or supplement automatic suggestions. Confirm the current controls and whether the desired block uses those rules.
3. **Inspect the component.** Use a complementary-products block where the theme supports it. For cart recommendations, check the API request's `intent=complementary`. A fragrance brand found its cart carousel always showed the same list, even recommending the bottle already in the cart. A styled carousel is not proof of contextual logic.
4. **Instrument placements.** The source describes publishing a custom theme event with `Shopify.analytics.publish`, then passing appropriate data through a custom pixel to GA4 as `view_item_list` and `select_item`. Give each placement a stable `item_list_name`, such as `product_recommendations`, `cart_recommendations` or `confirmation_recommendations`, and preserve the list context on later cart additions. Check existing theme/app tracking first to avoid duplicates and honor applicable privacy settings.
5. **Measure and compare.** For each placement, inspect click rate, add-to-cart rate, average order value and contribution from orders including suggested items. Then test presence against absence using a suitable [Shopify experiment](https://conversion.club/playbook/how-to-a-b-test-on-shopify-with-or-without-an-app/), with revenue and profit outcomes alongside interactions.
6. **Maintain the rules.** Review unclicked pairs monthly, allowing for exposure volume. The source says native complements exclude unavailable and already-carted products; verify the actual behavior of your version, custom lists and third-party apps. Remove irrelevant or incompatible pairings promptly.

## The pitfall

An app's “revenue generated by recommendations” commonly counts orders after someone clicked and bought a suggested product, including purchases that would have happened anyway. More exposure can make this attributed figure grow without improving the business.

Hide the block from a randomized control group and compare [revenue per visit](https://conversion.club/playbook/should-you-steer-by-revenue-per-visit-instead-of-conversion-rate/) and contribution. With insufficient traffic, matched-period comparisons can provide an estimate, with explicit seasonal and assortment limitations.

## What to tell your team

> “The cart block gets clicks from 7% of visitors, but we do not know its incremental revenue. Before paying for an engine, we will define rules for 30 bestsellers and compare six weeks of revenue per visit against a control without the block.” These figures are fictional.

For related placements, see the [confirmation page](https://conversion.club/playbook/what-should-go-on-your-order-confirmation-page/) and [on-site search](https://conversion.club/playbook/useful-ecommerce-site-search/).

## Sources

- [Baymard: alternatives and complements, 2014](https://baymard.com/research-articles/product-page-suggestions?ref=conversion.club).
- [cart cross-sells, 2021](https://baymard.com/research-articles/product-recommendations-cart?ref=conversion.club).
- [Shopify: Search & Discovery recommendations](https://help.shopify.com/en/manual/online-store/search-and-discovery/product-recommendations?ref=conversion.club).
- [recommendations API](https://shopify.dev/docs/api/ajax/reference/product-recommendations?ref=conversion.club).
- [publishing pixel events](https://shopify.dev/docs/api/web-pixels-api/emitting-data?ref=conversion.club).
- [GA4: ecommerce measurement](https://developers.google.com/analytics/devguides/collection/ga4/ecommerce?ref=conversion.club).