Abi Davies
Posted 04 October 2017

7 Product Recommendation Engines (and When to Use Them)

Recommendation engines are difficult to get right (and not something you want to get wrong—unless you don’t mind ending up in Private Eye’s “Malgorithms” column).

But this doesn’t mean you should shun product recommendations altogether; on the contrary, in an industry dominated by personalisation, your brand should be investing more time and energy in making them accurate (and in no way annoying).

In this blog post, we’ve taken a look at seven different recommendation engines that can help you do just that. Whilst each one can be used in different ways for different emails (broadcast or automated), there are certain instances where specific engines lend themselves particularly well to specific campaigns; where this is the case, we’ve made a note using a ⭐ . 

7 product recommendation engines to use in your email marketing

Based on your best/latest products

1) Top products

product recommendation engine example_top products

⭐ A product recommendation engine that displays your brand’s “Top products” is great for any campaign where you have limited or outdated customer data (making personalisation difficult); for example, a welcome series.

This engine is also good as a fallback model for other engines if they don’t have enough customer data to operate.

2) Latest products

product recommendation engine example_latest products

⭐ Like “Top Products”, “Latest products” is also a favourable recommendation engine for campaigns that are not supposed to be personalised. However, that said, even if you do have enough customer data to run a more bespoke engine, “Latest products” is a good option if your goal is to promote your brand’s latest range and/or highlight what’s new in-store.

Product-based recommendations

3) Similar products

similar products engine 2.png

⭐ Unlike the above two examples, a product recommendation engine based on “Similar products” needs clear product data (attributes and categories) to actually be able to work.

As it strives to display products that are similar to those a consumer has already shown interest in, it’s usually used for any campaign that’s based on activity: i.e. browse abandonment, basket abandonment and post-purchase.

4) Bought this, bought that

bought this bought that product recommedations engine example

⭐ As the name suggests, the “Bought this, bought that” engine takes a product that a customer has bought and looks at the purchase activity of those who bought the same item—using any other products they purchased as recommendations. This engine works on the premise that certain customers have similar tastes and spending habits.

This product recommendation engine can be used in a post-purchase campaign to upsell a customer; in order to ensure it’s definitely an upsell (and not a cross-sell), you can customise your engine to only show products within a specific price range (…more information on how to do this can be found at the bottom of this post!).

5) Viewed this, bought that

viewed this, bought that product recommendation engine example

⭐ The “Viewed this, bought that” engine is often used for welcome campaigns, where a new contact has viewed certain products but not yet made a purchase. Again, it draws on the purchasing habits of similar customers to display the products a recipient is most likely to want to buy.

Profile-based recommendations

6) Recently viewed

recently viewed engine example

⭐  “Recently viewed” marks a transition to more personalised, profile-based recommendation engines. Displaying the actual products that have been recently viewed by a recipient, this engine is similar to a browse abandonment campaign.

7) Personalised

personalised engine 2.png

This engine gets a special star because it’s a special engine.

Drawing on *all* of the customer data available, it’s likely to bring about the best (and most bespoke) results for each recipient.

However, in order to use a fully personalised engine such as this you need a certain amount of customer data—preferably brought together and stored in one, central place.

For example, in our platform we create detailed customer profiles so that a marketer can gain a true understanding of an individual’s tastes and needs.

Each profile has its own dynamic “recommended products” section, which updates according to any interaction that a customer has with your brand.

personalised product recommendations in ecommerce marketing

✏️ A note on customisation 

To make an email campaign’s product recommendations even more precise, you can filter the above engines by:

  • Fallback model: Choose a product recommendation engine to fall back on, just in case your more personalised engines don’t have enough data to work with.
  • Category/attribute: You can limit your recommendations to a specific product, category or brand. Likewise you can blacklist certain products you don’t want to appear.
  • Price range: If you only want to recommend products of a certain price, you can restrict your recommendations to a specific price range.

price range.png

  • Purchase history: Show how well you know your customers by preventing any products they’ve previously bought from being suggested.

tickbox remove items that have been bought.png

The mistakes to watch out for

Going back to what we said at the start of this post, product recommendations can sometimes hinder rather than help a brand—acting as a source of embarrassment and humiliation.

Whilst the above recommendations are pretty much a foolproof way of getting it right, here are two things to watch out for when you’re putting any type of recommendations together:

– Incorrect product details 

There’s a mid-season sale on, but the products recommended in your email template have the old price. Eek.

To stop this happening, make sure your product recommendation engine updates in real-time (or near real-time). You could also try previewing the recommendations a few times (for different recipients) to check everything is correct.

– Recommending (seemingly) random products 

Recommending totally irrelevant products to a contact can happen when an engine draws on similar attributes from other buyers who bought something similar to the customer, based on factors such as: price point, patterns of purchase, category, size, colour, etc.

Basically: at an algorithm level it makes sense, but at a product level it simply doesn’t.

The result? An email recommending dog leads to cat-lovers, terrible rom-coms to discerning film buffs or a “beginners French” textbook to a native French speaker…

Our advice: If you’re concerned personalised recommendations risk drawing on information that is too broad, filter your recommendations to only fall within a specific category.

To read more about potential recommendation fails (and how to solve them!) you can check out our blog post on the subject here.

Like what you're reading? Never miss a post by joining our mailing list

Ometria is committed to protecting and respecting your privacy, and we’ll only use your personal information to administer your account and to provide the products and services you requested from us. You may unsubscribe from these communications at any time. For information on how to unsubscribe, as well as our privacy practices and commitment to protecting your privacy, please review our Privacy Policy.

Book a Demo

Take the first step toward smarter customer marketing

Having Ometria has completely transformed our approach to CRM.
Credence Pym, CRM Manager at Sigma Sports

Report abuse

We just need few more details to get started

Let's get started