Pairs Well With: Cross-sell Recommendations Without a Discount

Introduction

Not every recommendation needs a discount.

Sometimes the product is already priced right, the margin is already thin, and the only thing missing is the suggestion.

The default assumption in most cross-sell planning is:

"What discount do we attach to the add-on?"

A solutions architect asks first:

"Is the problem price, or is the problem that nobody sees the product?"

When the answer is visibility, a discount only gives away margin on a sale that would have happened at full price. This case study designs a recommendation-only cross-sell for a wine accessories store that used Cross-sell with no discount at all.

Merchant Scenario

Consider Thistle & Vine, a fictional Shopify brand selling decanters, glassware and bar tools.

Glassware had a 28% margin, and the team was not going to discount it. But a customer buying a decanter almost always wanted glasses to match, and the store never said so.

The Discount That Was Not Needed

Every cross-sell tool the team had looked at assumed a discount.

Thistle & Vine's problem was not price. It was visibility. The decanter page did not mention glasses, and the glasses page did not mention decanters. They needed a recommendation, not a reward.

Business Objectives

Evaluating the Options

Option 1: Discounted Cross-sell

Option 2: No-Discount Cross-sell With Manual Picks

Option 3: No-Discount Cross-sell With a Collection Pool

Comparison Matrix

The Discount Prime Configuration

A collection trigger, a collection pool, three items, and no discount.

Why Cross-sell Fits

"No discount" is a real reward type in Cross-sell, and it changes what the campaign is.

It becomes recommendation only. The widget suggests items and one-tap adds them at full price. Nothing is applied at checkout. There is no unit cap and no apply condition, because there is nothing to cap or condition, and the discount function ignores the campaign entirely.

What remains is the part Thistle & Vine actually needed: the widget, the placement, and the analytics on impressions, adds and attach rate per offer.

Pool, Not Picks

The offer source was Pool, from the Stemware collection. Stemware changed often: new shapes arrived and old ones sold through. A pool followed the collection without edits.

Items to show was 3. The pool is ranked by 90-day units sold, so the three glasses customers were actually buying surfaced first.

The live preview showed the top of the ranked list, with an Exclude on each row. Two discontinued but still listed glasses were excluded in two clicks. For a catalog where the exact companion matters more than freshness, manual picks win instead, as in Frequently Bought Together, Without Guessing.

Collection Trigger and Both Placements

A Trigger That Was a Collection

The trigger was the Decanters collection, not a product list. Any decanter in the cart, including next season's, triggered the offer.

The campaign builder notes that rule-based triggers update live as products move in and out of the collection. Nobody had to remember to add a new decanter to the campaign.

Both Placements

The product page block showed three glasses under every decanter. The cart card showed the same three to anyone with a decanter in the cart who had not added glasses yet.

Cross-sell never offers the product being viewed, and never offers something already in the cart. So the cart card only ever showed glasses the customer did not have.

Customer Journey

1. A shopper opens a decanter page and sees Pairs well with and three glasses, the top sellers from Stemware over the last 90 days.
2. They tap Add on one glass. It goes into the cart at full price.
3. In the cart, the card shows the remaining stemware they do not have yet.
4. At checkout nothing changes. There is no discount to apply.

What a No-Discount Cross-sell Cannot Do

It cannot make glassware cheaper. By design.

It also gives the discount function nothing to do, which means none of the usual campaign-side protection applies: no max units, no apply mode. For a recommendation that is fine. For anything with a price change it would not be.

And a pool ranked by sales keeps recommending the bestseller even when the merchant would rather push something new. Pinning is a Smart suggestions feature (an Ultimate feature), not a pool feature. Thistle & Vine lived with the ranking.

Metrics That Measure Success

• Glassware attach rate on orders containing a decanter
• Impressions, adds and attach rate per offered glass, from the campaign analytics
• Average order value of decanter orders against the $79 store baseline
• Gross margin on glassware lines, held at full price
• Product page versus cart placement contribution to adds

Common Mistakes

• ❌ Assuming a cross-sell has to include a discount, and giving away margin on products that were already converting.
• ❌ Using manual picks for a fast-changing collection, and recommending glasses that sold out months ago.
• ❌ Setting Items to show to 6 and overwhelming a product page with a row of near-identical glasses.
• ❌ Forgetting that the pool is ranked by sales, and wondering why the new release never appears.

Key Lessons

A cross-sell is a recommendation with an optional discount, not a discount with an optional recommendation.

Turning the discount off left the whole mechanism intact: trigger, pool, placement, analytics. And a collection on both sides meant the campaign kept itself current.

Conclusion

A suggestion at the right moment is worth more than a markdown at the wrong one.

For products that already sell at full price but are not being seen, Cross-sell with "No discount" puts them on the page with nothing given away.

In this scenario, Thistle & Vine's glassware attach rate on decanters rose from 18% to 44%. The price never moved.