How a DTC Fashion Brand Increased AOV 28% With Cross-Sell Timing

Harlow & Rye is a DTC womenswear brand based in Chicago. When they came to us in March 2025, their average order value was sitting at $94 — solid for their price point, but they knew there was room to grow. They were running cross-sells manually: a merchandising team selecting "complete the look" items for each product, updating the selections monthly.

Twelve months later, their AOV is $120. The difference isn't magic. It's timing.

The setup: where they were starting

Harlow & Rye's catalog runs about 800 active SKUs at any given time — dresses, tops, pants, outerwear, and a growing accessories line. Their core customer is 28-42, shops primarily on mobile, and tends to browse 4-6 products before purchasing. Average session length is 7 minutes.

Their cross-sell implementation was a static "You might also like" row at the bottom of every product page. Same items for every visitor. The selections were curated but generic — if you were looking at a black blazer, everyone saw the same four companion pieces regardless of what else they'd been browsing.

Add-to-cart rate from those cross-sell recommendations: 2.1%.

What we changed: signal-driven timing

The first thing we addressed was timing. Cross-sells at the bottom of a product page are an afterthought — they appear before the shopper has decided whether they even want the primary product. You're asking someone to consider accessories before they've committed to the main item.

We moved the primary cross-sell trigger to three moments:

Post-add-to-cart modal. The moment a shopper adds an item to cart, they've made a decision. That's the highest-confidence moment to show complementary products. The shopper's buying frame is active. A well-timed "this pairs well with" prompt at that moment feels helpful, not pushy.

Cart page, above the checkout button. Shoppers who reach the cart page are 3-4x more likely to buy than average session visitors. The cart page is not the time to show them unrelated products — it's the time to show them two or three specific items that complete or enhance what they already have.

Post-purchase confirmation. The brand had never used this touchpoint for cross-sells. It's underutilized across the industry. A shopper who just completed a purchase is in a confirmed buying mindset. They've already decided they trust the brand. A "you might want these to go with your order" suggestion on the confirmation page doesn't feel aggressive — it feels like a natural extension of a positive transaction.

What else we changed: personalized pairing logic

Timing without relevance is just better-placed noise. The second change was replacing static cross-sell selections with dynamic personalization.

Instead of "here are four items we picked to go with this blazer," the logic became: here are the items most likely to appeal to this specific shopper based on the price range they've been browsing, the aesthetic signals in their session behavior (classic vs. contemporary styles, their apparent color preferences based on products they've engaged with), and the purchase patterns of shoppers who bought this item before.

For a shopper who'd been browsing minimalist, neutral-tone pieces, the blazer's cross-sells weighted toward white shirts, straight-leg trousers, and simple accessories. For a shopper who'd been engaging with printed pieces and brighter colors, the same blazer's cross-sells weighted toward patterned skirts and bold accessories. Same primary product, completely different supporting cast.

The numbers after 90 days

We ran a 90-day A/B test — properly powered, pre-specified, held for full test duration before calling the result. The variant (signal-driven timing + personalized pairing) vs. control (static bottom-of-page cross-sells):

• Cross-sell add-to-cart rate: 2.1% (control) vs. 6.8% (variant) — +224%
• Revenue per session from cross-sells: $1.87 (control) vs. $4.31 (variant) — +130%
• Overall AOV: $94 (control) vs. $112 at 90 days (variant)
• Cart abandonment rate: no significant change (this matters — aggressive cross-sells can hurt abandonment; these didn't)

The test was statistically significant at 99% confidence across all primary metrics. Sample size: 28,000 sessions per variant over 12 weeks.

The full-year picture

By month 12, Harlow & Rye's AOV had reached $120 — a 28% increase from the pre-ShopPulse baseline. The trajectory continued past the 90-day test window because of two compound effects the test didn't fully capture:

First, the post-purchase cross-sells drove repeat purchases. Shoppers who saw relevant post-purchase suggestions had a 23% higher 90-day repurchase rate than those who didn't. Cross-sell timing wasn't just increasing AOV on the initial transaction — it was shortening the repurchase cycle.

Second, the personalization logic improved over time as the model accumulated more behavioral data specific to Harlow & Rye's customer base. Month 12 cross-sell performance was meaningfully better than month 3, even though the underlying approach was the same.

What this requires operationally

The brand's merchandising team went from spending 15-20 hours per month updating static cross-sell selections to spending 3-4 hours reviewing the model's output and adjusting category-level weighting for new collections. That's a substantial shift in how the team's time is spent — less manual selection, more strategic oversight.

It also required a short integration period to map product attributes at the level of detail the personalization logic needed: fabric type, occasion, style aesthetic, price tier, complementarity scores between categories. That mapping took about two weeks upfront and is largely evergreen.

The underlying principle

The 28% AOV increase wasn't achieved by showing people more cross-sells. It was achieved by showing people the right cross-sells at the moment they were most receptive to buying. Same catalog, same price points, same brand voice. The only thing that changed was the system's understanding of when and what to show each shopper.

That's the actual opportunity most brands are sitting on. Not more aggressive cross-selling. Better-timed, better-personalized cross-selling.