I’m trying to separate hype from reality on product recommendations (“Recommended for you”, “You may also like”, bundles/upsells, etc.).
From conversations with a few brands, the results sound mixed. Some saw little/no lift. I’m not sure if that’s due to low traffic/data, weak placement, poor testing, or just that recommendations aren’t a big lever vs merchandising.
If you have tested recommendations/personalization what kind of things DID improve conversion rate, AOV, or revenue per session?
Even rough directional results are helpful (e.g., “no change”, “small lift”, “noticeable lift”)
@simplecheckout, if you are looking to improve your AOV I imagine you already implemented “You are $x from FREE Shipping“? You may test raising it by just a hair over your AOV and test with meaningful traffic. As for “Recommendations” and “Personalization” they need to actually be meaningful. You can also A/B test with different wording for those sections. Also bundling with a discount can help you move more inventory. Maybe shipping 1 item or 2 is just about the same cost for you, so having customers adding an additional product to their order at a discount doesn;t cost you more to ship both together.
The honest answer is that recommendations are a real lever, but the default implementations available to you aren’t great, and you may need to test what works best for your product line or brand.
A few things I’ve seen actually move numbers, in my experience across a few thousand stores:
Placement matters more than algorithm. A “frequently bought together” module on the cart page or in a slide-out cart consistently outperforms the same module on the product page. By the time someone’s looking at their cart, they’ve already committed to buying and they’re in “what else do I need” mode rather than “am I buying this” mode. Product page recommendations compete with the primary purchase decision and often just create decision fatigue - often hurting overall conversion instead.
“Customers also bought” outperforms “you might also like.” The first implies social proof and a curated selection. The second feels like the store is guessing. Wording changes like this routinely produce measurable lifts for almost zero effort. People want to know what “normal people” do.
Free shipping thresholds are the highest-ROI “recommendation” most stores aren’t using well. Set it just above your current AOV (like 15-20% above whatever that number is today, regardless of what that number is today), show progress toward it prominently, and the recommendations basically sell themselves because customers are now actively looking for something to add. Even better if you can make the recommendation cost something just a bit higher than what they need to hit the shipping threshold.
Bundles work when they solve a problem, not when they’re just a discount. “Everything you need to get started with X” converts extremely well. “Buy 3 random things and save 10%” mostly doesn’t. I’ve seen it work, but I wouldn’t do it myself.
The brands that report “no lift” usually have generic Shopify-default recommendations showing loosely related products on the product page and nowhere else. It’s just not a fair test. You have to try things and engage in ongoing experimentation to find the exact right combination for you. And you probably need to pay for an app or two to do it properly, which some people are too stubborn to do 
Let me know if you have any questions!
From what I have seen across a few Shopify stores, product recommendations can help, but their impact really depends on how relevant they are and where they’re placed. The mixed results you’re hearing about are accurate.
What worked better was complementary products (items that naturally go together), bundles or “frequently bought together” offers, showing them near the Add to Cart or in the cart drawer.
These setups usually deliver a small to noticeable lift in AOV, while conversion rates remain stable.
In our case, instead of relying on personalization, we manually grouped related products and kept the bundle simple using the Wizio Bundle Shopify app. The improvement wasn’t huge, but it was consistent and easier to control.
Overall, recommendations work best when they feel like smart merchandising, not just automated suggestions.
The biggest gains come from better product info & options clarity and faster paths to the right product. When customers clearly understand variants, compatibility, and use-cases—and could quickly narrow to the right item—conversion improvs. Generic recommendations on top of unclear product data usually led to no change.
Once clarity and navigation are solid, recommendations and bundling helps with the margins. Without that foundation, it’s hard to move the needle.
