How AI Upsell in DMs Increases Shopify Average Order Value by 23%

Every Shopify store owner obsesses over two metrics: conversion rate and traffic. But there's a third metric that's often more impactful and far easier to improve: average order value (AOV).

Getting a customer who's already buying to add one more item to their cart is dramatically cheaper than acquiring a new customer. And in 2026, the best place to do it isn't on your checkout page — it's in the conversation.

Why DM-based upselling outperforms on-site popups

Traditional upselling happens on your product page or checkout: "Customers also bought..." widgets, post-purchase popups, bundle offers. These work, but they face two problems:

• Banner blindness — customers have learned to ignore on-site recommendation widgets. Click-through rates on "You may also like" sections average just 1.8%
• No context — a widget doesn't know why the customer is buying. It can only match by purchase history or category. It can't read intent

DM-based upselling is fundamentally different. When a customer messages your Instagram or WhatsApp asking about a product, they're actively engaged in a conversation. They're asking questions, expressing preferences, revealing their intent. That's the perfect moment to suggest complementary products — because you actually know what they want and why.

The psychology behind conversational upselling

There are three psychological principles that make DM upselling so effective:

1. The commitment principle

When someone has already invested time asking questions about a product, they've made a psychological commitment. They're far more likely to add items to their purchase than a cold visitor browsing your site. Robert Cialdini's research shows that small commitments (like asking a question) dramatically increase the likelihood of larger commitments (like buying more).

2. Personal recommendation effect

A product suggestion in a conversation feels like a recommendation from a friend, not an ad. "Since you're getting the running shoes, you'll definitely want these moisture-wicking socks — they're designed for the same impact zones" hits differently than a generic "Frequently bought together" widget.

3. Social reciprocity

When an agent provides genuinely helpful answers to product questions — sizing advice, material comparisons, use-case recommendations — the customer feels a subtle obligation to reciprocate. Adding a suggested item to their cart is a natural way to "return the favor" for helpful service.

How AgentFlow's Sarah agent upsells in DMs

Sarah has real-time access to your entire Shopify catalog — products, collections, tags, inventory, pricing, and active discounts. When a customer asks about a product, Sarah doesn't just answer the question. She looks for the upsell opportunity.

Example 1: The complementary product

Customer (Instagram DM): "Hey, does the Classic Running Shoe come in size 11?"

"Yes! The Classic Running Shoe is available in size 11 — and we actually have the new colorway (Storm Grey) that just dropped this week. It's in the same collection. A lot of runners also grab our Performance Running Socks ($12.99) — they're designed with the same cushioning zones as the shoe. Want me to create a cart with both?"

Sarah identified the product, confirmed availability, mentioned a new variant (potential upgrade), and suggested a complementary product from the same collection — all in one natural response. She even offered to create the cart, reducing friction to zero.

Example 2: The bundle opportunity

Customer (WhatsApp): "What's the difference between the Basic and Pro yoga mats?"

"Great question! The Basic ($29.99) is 4mm thick — perfect for beginners. The Pro ($49.99) is 6mm with extra grip coating, which really matters if you're doing hot yoga or flow sequences. Most of our yoga customers also love the Cork Yoga Block ($18.99) and the Carry Strap ($9.99) — I can bundle all three for free shipping. Which mat are you leaning toward?"

Sarah answered the question thoroughly, then naturally introduced two add-ons that make sense together. She framed it as a bundle with a benefit (free shipping), not a hard sell.

Example 3: The upgrade nudge

Customer (Messenger): "Is the Basic Wireless Charger compatible with iPhone 15?"

"Yes, the Basic Wireless Charger ($19.99) works perfectly with iPhone 15! Just so you know, we also have the Fast Wireless Charger ($34.99) that charges 2x faster — if you're someone who charges overnight it won't matter, but if you're a quick-charge-before-going-out type, the Fast version is worth it. Both are in stock!"

Sarah confirmed compatibility, then positioned the upgrade naturally by connecting it to the customer's use case rather than just listing specs.

How Sarah decides what to suggest

Sarah's upsell logic is based on your actual Shopify data, not generic rules:

• Same collection — products in the same collection are natural complements
• Shared tags — if the customer's product and the suggested product share tags like "running", "outdoor", or "summer", they're likely a match
• Price range — Sarah suggests add-ons that are 20-40% of the main product's price (the sweet spot for impulse additions)
• Inventory priority — Sarah favors products with higher inventory levels, helping you move stock
• Active discounts — if a complementary product is on sale, Sarah highlights it as extra incentive
• Purchase patterns — over time, Sarah learns which products are frequently bought together in your store specifically

The numbers: what to expect

The key insight: customers who receive a relevant product suggestion in a DM conversation don't feel sold to. They feel helped. The suggestion is contextual, timely, and personalized — it's what a great human salesperson would do, except Sarah does it at scale, 24/7.

Setting up DM upselling (2 minutes)

1. Connect Shopify — Sarah gets access to your full product catalog, collections, and pricing
2. Connect DM channels — Instagram, WhatsApp, and/or Messenger
3. Set upsell rules — max products to suggest per conversation, price range limits, excluded products
4. Customize tone — casual, professional, playful — Sarah matches your brand voice

That's it. Sarah starts upselling in DMs immediately. You can review all conversations and upsell performance in your daily Slack briefing.

Upselling without being pushy

The biggest fear store owners have about AI upselling: "Will it annoy my customers?" The short answer is no — if done correctly. Sarah follows three rules:

• Always answer the question first — the customer's question is the priority. The upsell comes second, naturally woven into the answer
• Maximum one suggestion per conversation — Sarah never stacks multiple upsells. One relevant suggestion, max two products
• Respect "no" — if a customer doesn't engage with the suggestion, Sarah doesn't push again. She moves on

The result: upselling that feels like good customer service, because it is.

← Back to all articles