How do I do cross-sell and upsell recommendations without being annoying?

The key is to make recommendations relevant, timely, and optional. If they feel like help instead of pressure, they’ll usually be welcomed.

What works

1. Recommend only when it truly fits

Base recommendations on:

• What the customer is already buying
• What similar customers tend to add
• The customer’s current stage or goal

Example:

• Good: “Since you’re buying trail-running shoes, many runners also want blister socks.”
• Bad: “You bought shoes, so here are 12 unrelated accessories.”

2. Keep the number small

Show 1–3 strong options, not a wall of products. Too many choices makes it feel pushy and creates decision fatigue.

3. Use the right moment

Best times are:

• On product pages
• In cart or checkout, sparingly
• After purchase, in confirmation or follow-up email
• In post-purchase flows based on usage

Avoid interrupting high-friction moments unless the add-on is clearly helpful.

4. Frame it as a benefit, not a tactic

Lead with the customer outcome:

• “Complete the setup”
• “Protect your purchase”
• “Get better results”

Instead of:

• “People also bought…”
• “You may like…”
• “Upgrade now!”

5. Make the “no thanks” path easy

Never hide the skip option. A low-friction “Not now” or dismiss button builds trust and reduces annoyance.

6. Match the recommendation to the customer

Use segments like:

• New vs. repeat buyer
• Budget vs. premium shoppers
• Professional vs. casual use
• Product compatibility

A premium upsell to a price-sensitive customer will feel tone-deaf.

7. Don’t overdo discounts

Constant “limited-time offers” can train customers to ignore you. Use urgency only when it’s real.

8. Test and measure annoyance signals

Track:

• Click-through and conversion
• Cart abandonment
• Unsubscribes
• Refunds/returns
• Support complaints
• Time to purchase

If engagement drops when recommendations appear, they’re probably too aggressive.

A good rule of thumb

Ask: “Would this still be helpful if I removed the sales goal?”

If yes, it’s probably a good recommendation.

Simple formula

Relevant item + right time + short list + clear value + easy opt-out

If you want, I can also give you:

• examples for ecommerce,
• examples for SaaS,
• or a recommendation UX checklist.