---
title: "Do personalization and product recommendations work?"
author: "Performetic Ekibi"
url: "https://www.performetic.com/en/blog/do-personalization-product-recommendations-increase-conversion"
published: "2026-11-20T12:00:00.000Z"
updated: "2026-11-20T18:58:25.833Z"
---

# Do personalization and product recommendations work?

> Personalization and product recommendations can lift conversion and basket size when they appear on the right page with relevant products; irrelevant suggestions erode trust. Recommend complementary items on product and cart pages and keep checkout clean. Start with rule based pairings on low traffic stores and measure impact with a controlled test based on revenue per visitor.

## Do personalization and product recommendations increase ecommerce conversion?

Personalization and product recommendations can lift conversion and average order value when they appear on the right page and suggest relevant products. In the wrong place, or with irrelevant suggestions, they distract shoppers and erode trust. No tool guarantees a lift, so measure which recommendation type works on your store with a controlled test.

This post covers the main recommendation types, where each one makes sense, how much data you need and how to measure the real effect.

## What does personalization mean in ecommerce?

Personalization means showing the store differently based on a visitor's behaviour, history or context. Product recommendations are the most common form, but not the only one. Featuring the last viewed category on the homepage, showing returning customers a "pick up where you left off" block, or sending ad traffic to the product group from the ad are all personalization too.

A common confusion: not every recommendation is personalized. A "best sellers" block shows everyone the same products. "Frequently bought together" changes by product, not by person. True personalization uses the visitor's own behaviour. Which approach suits you depends on your traffic and data volume.

## What types of product recommendations are there?

| Recommendation type | Logic | Where it fits | Watch out for |
|---|---|---|---|
| Related products | Alternatives to the viewed product | Product page, out of stock page | Distracts in cart and checkout |
| Complementary products | Items used with the viewed product | Product page, cart, add to cart drawer | Must be genuinely compatible |
| Recently viewed | The visitor's own browsing history | Homepage, category, 404 page | Respect privacy preferences |
| Best sellers | Store wide sales data | Homepage, empty search results | Same for everyone |
| For you | Personal behaviour and similar users | Homepage, email | Needs enough data |

Shopify's developer documentation draws this line clearly: related product recommendations are generated automatically by the platform, while complementary products are set up manually with the Search & Discovery app. There is a practical lesson in that. Rather than leaving complementary pairings to an algorithm, having the team that knows the products set them up often gives more accurate results.

## Which recommendations belong on which page?

A general page by page approach:

1. **Product page:** Complementary products first, then related ones. Price, variant selection and the add to cart button must stay prominent; recommendations belong lower on the page.
2. **Add to cart drawer:** Only a few genuinely complementary items. The goal is a bigger basket, not sending the shopper to a different product.
3. **Cart page:** Complementary items and low priced products that help reach the free shipping threshold. Showing alternatives here invites the shopper to rethink the decision.
4. **Checkout:** In most cases, show no recommendations. The only goal here is finishing the order.
5. **Homepage and category pages:** Recently viewed for returning visitors, best sellers for new ones.
6. **Out of stock product page:** Related products do their most valuable work here, giving the shopper an alternative instead of a dead end.

## Why do irrelevant recommendations hurt?

According to Baymard Institute's cart research, 52% of the desktop sites it benchmarked show cart cross-sells that are unrelated to the purchase or based only on what other customers bought. The same research reports that a single questionable suggestion can make users lose faith in the whole list and ignore it, while compatibility based suggestions are seen as good service rather than a sales push.

Another Baymard guideline notes that users often ignore or misunderstand cross-sell sections when it isn't clear how the suggested products relate to what they are viewing. The practical takeaway: give every recommendation block a clear label. Instead of "You may also like", try "Capsules compatible with this coffee machine".

## How much data does personalization need?

Personalized recommendation algorithms run on behavioural data. On a store with low traffic and a small catalog, a "for you" block often produces results no different from best sellers. In that case rule based recommendations make more sense: pairing complementary products by hand, showing best sellers per category and defining alternatives for out of stock items.

As traffic and order volume grow, algorithmic personalization starts to pay off. Ask yourself one question when deciding: is my data enough for an algorithm to make better pairings than my own team?

## What are the most common personalization mistakes?

These are the mistakes we see most often in campaigns and site audits:

- **Adding recommendation blocks to every page:** Pages get longer, and on mobile the core product information gets pushed down.
- **Recommending out of stock products:** The shopper clicks, can't buy and is twice as disappointed.
- **Recommending items already in the cart:** A missing filter makes the store look anything but smart.
- **Always pushing discounted products:** It brings clicks in the short term but erodes margin.
- **Setting it and forgetting it:** Manual pairings need updating as the catalog changes.

## How should personalization work alongside KVKK?

Behaviour based personalization relies on data collected through cookies and similar technologies. In Turkey this brings disclosure and, where required, explicit consent obligations under KVKK, the national data protection law. Make sure your cookie preference management works with your personalization tools; visitors who don't consent should see general recommendations instead of personalized ones. Get legal advice for your specific setup.

## How do you measure the effect of product recommendations?

Looking at a "revenue from recommendations" report after launch is misleading, because some of those sales would have happened anyway. The right method is a controlled test. Show the recommendation block to one group of visitors and hide it from another, then compare revenue per visitor, conversion rate and average order value. Run the test for at least two full weekly sales cycles, avoid starting during big promotions and look at return rates as well as revenue. If recommended items come back often, the gain only exists on paper.

Example: a store with 50,000 monthly visitors shows a complementary products block on the cart page to half its visitors for four weeks, then compares revenue per visitor across both groups. If the difference isn't meaningful, the block is removed or the pairings are rebuilt.

## How do recommendations affect average order value?

Complementary product recommendations are one of the most natural ways to grow the basket, and they work harder when paired with a free shipping threshold. See our broader [guide to increasing average order value](/en/blog/how-to-increase-average-order-value) for more. On-site search is part of personalization too; showing recommendations on searches with no results cuts losses, as we explain in our [post on site search](/en/blog/ecommerce-site-search-optimization).

## How can Performetic help?

Improving product pages, carts and recommendation logic through testing is part of our [conversion-focused design service](/en/services/conversion-focused-design). If you'd like to see where recommendations help and where they distract on your store, request a [free growth audit](/en/contact).

## Key takeaways

- Product recommendations can lift conversion and basket size on the right page with relevant products; irrelevant ones damage trust.
- Use complementary and related products on product pages, complementary items in the cart and nothing at checkout.
- On low traffic stores, rule based and hand built pairings beat algorithmic personalization.
- Set up behaviour based personalization in line with cookie preferences and KVKK obligations.
- Measure impact with a controlled test, not a "revenue from recommendations" report.

## FAQ

### Where do product recommendations work best?

Product pages and the cart are usually where recommendations create the most value. Product pages suit complementary and related items, while the cart suits complementary items and products that help reach a free shipping threshold. Showing no recommendations at checkout is usually the better choice, because the only goal at that stage is finishing the order.

### Should a small online store use a personalization tool?

On a store with low traffic and a small catalog, algorithmic personalization often performs no differently from a best sellers block. At that stage, pairing complementary products by hand, showing best sellers per category and defining alternatives for out of stock items is more accurate. Algorithmic tools start to make sense as traffic and order volume grow.

### How do you measure the sales impact of product recommendations?

A revenue from recommendations report is misleading, because some of those sales would have happened anyway. Set up a controlled test that shows the recommendation block to one group of visitors and hides it from another, then compare revenue per visitor, conversion rate and average order value. Check return rates for recommended items as well.

### Does personalization need consent under KVKK?

Behaviour based personalization relies on cookie data, so KVKK can bring disclosure and, where required, explicit consent obligations. Your cookie preference management needs to work with your personalization tools, and visitors who decline should see general recommendations instead. It is advisable to ask a legal expert about your specific setup.

## Sources

- [Baymard Institute: 6 ways to improve the relevance of cross-sells in the cart](https://baymard.com/research-articles/product-recommendations-cart)
- [Baymard Institute: Cross-sell product relevance (guideline #813)](https://baymard.com/guidelines/813-cross-sell-product-relevance)
- [Shopify developer documentation: Product recommendations](https://shopify.dev/docs/storefronts/themes/product-merchandising/recommendations)
