Product recommendation instances predict and display products your contacts might be interested in purchasing.
Ometria has seven recommendation engines for use in your campaigns.
For product recommendation criteria, rules, and product data requirements, see Product recommendations overview.
This page explains how to create or edit a product recommendation engine.
Note for new clients
Product recommendation engines work based on a model which needs to be trained based on the data in your account. Our model trainers run overnight and may take a while to finish, depending on the recommendation engine you select. The engines can only provide recommendations once the model has been trained.Once you've worked through these steps, you might want to read: How to add a product recommendation block to a template.
Recommendation engine types
There are seven different recommendation engines.
Some engines are designed for use in broadcast campaigns, others for automation campaigns, as indicated below.
|
Icon
|
Engine description |
Type |
 |
Top products - Recommends the best-selling product. Choose to recommend the best-selling products (by number of orders placed) in the past 7, 14 or 28 days. |
No context recommendation engines These can be generated for any contact, without needing to know which products they have interacted with. You can use these engines for both broadcast and automation campaigns.
|
 |
Latest products - Recommends the products most recently added to your online store. Products are only ‘latest products’ the first time they are created in Ometria. If you update a product on your site (e.g. to add new colours to the product line) they will not be listed as ‘latest products’. |
|
 |
Similar products - Recommends products most similar to the products the contact has interacted with, based on common attributes. For automation campaigns, the type of interaction depends on the triggering event. E.g. If the trigger is an order, then the recommendations are based on the product purchased. If the trigger is a visit to a webpage, the recommendations are based on the products viewed. |
Product based recommendation engines These can be used where a contact has interacted with a product/set of products, e.g. abandoned basket or ordered. These are known as context products. We recommend you use these engines for automation campaigns only, as they require data from the contact in order to generate useful recommendations. |
|
|
Bought this, bought that - AKA 'product up-sell'. Recommends the most commonly purchased products by customers that purchased any of the context products. |
|
 |
Viewed this, bought that - AKA 'product cross-sell'. Recommends the most commonly purchased products by other customers in the same session. |
|
|
|
Recently viewed - Recommends products that the contact has viewed in the past 7, 14 or 28 days. These recommendations display in order of most recent. |
Profile based recommendation engines These engines generate recommendations based on the contact's behavioural history (products they've purchased or viewed). We recommend you use these engines for automation campaigns only. Note: You can use these engines in broadcast campaigns, but contacts without the relevant data available (e.g. recently viewed/recently purchased) will receive the fallback engine you choose.
|
 |
Personalised - Recommends products with attributes the contact has shown most interest in (by viewing or purchasing a product with those attributes). See below for more information. |
Personalised product recommendations
This engine recommends products that have attributes that the customer has shown the most interest in (by either viewing or placing orders for products with matching attributes).
This is called attribute affinity, which you can check for individual contacts in the contact details screen.
Personalised product recommendations are generated based on:
- Customer frequency - the number of times a contact has viewed or purchased a product with a specific attribute
- Attribute weight - If a product attribute is applied to a large number of products it doesn’t tell us as much about the contact’s affinity for the product itself, so it is given a lower weighting than an attribute which applies to fewer products.
The weight of an attribute is calculated as:
Customer frequency x Attribute weight = Product score
The personalised recommendation engine recommends products with the highest score.
Create or edit a product recommendation instance
From the Product Recommendations screen, select Create new:
The New Product Recommendation screen displays:
Enter a Recommendation name, then select a recommendation engine.
This is because these are ‘no context recommendation engines’, meaning that no context data about the interaction is required from the contact to generate recommendations.
Broadcast campaigns typically send to a broader group of contacts who may not have interacted with your brand in some time.
Next, configure the Settings for your product recommendations:
|
Field |
Description |
|
Default store |
Set the store (e.g. UK, Germany, US) to pull recommended products from. |
|
Time period |
Set a time period for the engine to search for products:
Note: This parameter is only available for 'Top products' and 'Recently viewed'.
|
|
Restrict to price range |
Set a 'from' and 'to' price so that the contact will only receive the products within that range. Note: Be aware that the restrictions you set here will not apply to the fallback model you select (if any).
|
|
Restrict to attribute(s) |
Restrict recommendations by attribute or category. Select a category from the dropdown, then select your attributes. You can switch between AND/OR logic for product recommendation attributes:
You can use AND or OR, but you can’t use both in combination. Note: Be aware that the restrictions you set here will not apply to the fallback model you select (if any).
|
|
Exclude product(s) |
Click and select any products you wish to block from this recommendation. These products will not be recommended to your contacts. |
|
Exclude attribute(s) |
Click and select any categories or attributes you wish to block from this recommendation. Products with these attributes will not be recommended to your contacts. Note: Be aware that the restrictions you set here will not apply to the fallback model you select (if any).
|
|
Include products that have been previously purchased |
Select this to recommend products even if they have already been purchased by the contact. Previously purchased products are excluded from recommendations by default. Note: Be aware that the restrictions you set here will not apply to the fallback model you select (if any).
|
|
Include out of stock products |
Select to recommend products flagged ‘out of stock’ by your eCommerce platform - useful for stock which is made to order. Note: This option will only display if you have checked the corresponding flag in General Settings.
|
|
Include inactive products |
Select to recommend products flagged ‘inactive’ by your eCommerce platform. Note: This option will only display if you have checked the corresponding flag in General Settings.
|
Fallback model
| Field | Description |
| Fallback model |
This is used in the event that the primary model (i.e. the recommendation engine you selected) cannot provide recommendations. The available engines are Top products and Latest products, as these should always produce results. You can restrict your fallback model to a specific product attribute if required. |
| Restrict to attribute(s) |
Restrict your fallback recommendations by attribute or category. Select a category from the dropdown, then select your attributes. You can switch between AND/OR logic for product recommendation attributes:
You can use AND or OR, but you can’t use both in combination. Note: Be aware that the restrictions you set here will not apply to the fallback model you selected (if any).
|
Select Save to finalise.
Preview your recommendation
Once you are happy with your chosen recommendation engine and the parameters, you can select Preview to see how the recommendations look to recipients.
The preview options are slightly different depending on which type of recommendation engine you are using (no context, product based or profile based).
No context recommendation preview
Select the relevant store (or ‘any store’ for all available products) and click Preview.
The results display in the Recommendation Preview window:
Product based recommendation preview
Select as many products as you like from the Choose product field, then choose your store (if necessary) and click Preview.
The results display in the Recommendation Preview window:
Profile based recommendation preview
Choose your store (if necessary) and enter a contact’s email address to test, then click Preview.
The results display in the Recommendation Preview window:
Comments
0 comments
Please sign in to leave a comment.