Einstein is smart, but it doesn’t know your next campaign

Einstein is cool. Until it recommends a surfboard in your winter campaign.

Salesforce B2C Commerce Cloud gives you two main ways to handle product recommendations: manually, or using AI-powered suggestions with Einstein. And while everyone loves the idea of automation, there are situations where you need to step in.

Because no algorithm knows your customers, products, or goals better than you do.

Let’s break it down.

Manual Product Recommendations: You’re in Charge

Sometimes, the best recommendation engine is still your own brain. With manual recommendations in SFCC, you can configure:

Product Associations
You decide which products are related. That includes upsells, cross-sells, and replacement products. These relationships are defined and maintained directly in Business Manager.

Examples:

• Showing accessories or parts for a specific product

• Suggesting a higher-priced version of the same item (upsell)

• Offering alternatives if a product is out of stock

Search-Driven Rules
You can also create product recommendations using search rules, including:

• Search-Based Rules: Configure a keyword to return a specific group of products

• Category-Based Rules: Show related products from the same category

• Attribute-Based Rules: Recommend products with similar attributes

These manual setups give you full control. No training data needed. No waiting for AI to “learn”. It just works, the way you designed it.

Merchandisers love this because it fits real-world needs:

• Launching a new collection? You want to manually promote it.

• Running a time-sensitive campaign? You want full control over what's shown.

• Regulated markets? AI might show things it shouldn’t.

Manual recommendations let you decide. Period.

Einstein Product Recommendations: Let the Algorithm Work

Einstein does the heavy lifting using machine learning. It analyzes customer behavior, context, and site activity to generate real-time product suggestions. The more data it has, the better it gets.

You don’t need to configure specific rules. Einstein learns and adapts. It’s great for:

• General cross-sells and upsells across large catalogs

• Personalized product carousels

• Increasing AOV and relevance without much manual work

But there’s a catch.

Einstein is a black box.
You can’t control every outcome. You trust it to make the best guess, but guesswork has limits.

Especially if:

• You just launched a product (no data yet)

• You’re promoting a very specific bundle

• You have legal constraints on what can be shown to whom

• You’re targeting B2B clients with unique needs

The Smartest Strategy? Use Both

This isn’t about manual vs Einstein. It’s about knowing when to use each.

Use Einstein to personalize at scale.
Use manual recommendations when precision matters.

You can even combine both:

• Let Einstein handle product carousels on PLPs or homepages

• Set manual cross-sells on PDPs or carts for strategic promotions

You don’t need to pick a side. You just need to know what’s best for the moment.

Final Thoughts

Not everything needs to be automated.

Salesforce B2C Commerce gives you powerful tools, but it’s your brain that drives conversions.
Don’t hand the wheel to Einstein and hope for the best.

Take control when it counts. Automate when it helps.
That’s how the best storefronts stay ahead.