Product affinity segmentation for broadcast campaigns

Product affinity segmentation is an Architect AI tool you can use in your broadcast campaigns to predictively identify contact segments with different degrees of affinity for a product attribute e.g. brand, category, fabric, colour. 

Product affinity segmentation can help you test assumptions and progressively make changes to your content strategy.

Setting up product affinity segments

To get started, reach out to your Customer Success Representative to discuss product affinity segmentation.

They will work with you to understand your strategy and which attributes you're most interested in exploring.

Choose up to 20 attribute names (e.g. black, blue, grey, cotton, polyester, etc.) with any combination of attribute types (e.g. colour, material, etc.).

See: Choosing your product attributes for product affinity segmentation

Your Customer Success Representative will check your product data to make sure it's ready for setup and configure the selected attributes for you.

How it works

Ometria’s product affinity segmentation model takes into account:

• Purchase behaviour, 
• Visit data, and; 
• Abandoned basket data. 

Product affinity segments are generated based on inferred attribute affinities, i.e. past behaviour is used to predict a customer's interests.

Once you've chosen your attributes and product affinity segments have been enabled for your account, each of your contacts is assigned an affinity score for each attribute:

• Established affinity
• Growing affinity
• Emerging affinity 

The more activity a customer has (e.g. visits, orders, baskets) related to an attribute, the more established their affinity for that attribute will be. 

Orders give the strongest indication of affinity.

Use cases

The tool works best if you have a clear marketing calendar for the next 6-12 weeks. 

This will give you time to plan and work with your Customer Success Representative on the best approach. 

Examples: 

• Cross-selling - e.g. if customers have an established affinity for Nike, will they also like Adidas?
• More impactful retention messaging - e.g. Target at risk customers with “We miss you” messaging that directly relates to their established interests.
• New lines or limited edition products - e.g. "We are launching new colours for this iconic piece."
• Testing assumptions  - e.g. "We think customers who have purchased this product in the past - and therefore show an established affinity for this attribute - will buy it again." 
• Replenishable products - e.g. if I target leads with replenishable products, will they be more likely to buy them than other kinds of product?

Create a campaign with product affinity segments

Create new broadcast campaign and set up the following where applicable:

• Add the starting contacts 
• Add the excluded contacts 
• Exclude unengaged contacts 
• Exclude contacts currently in automation campaigns 

Select Start in the Product affinity segments panel:

Build your product affinity segments by selecting: 

• One attribute type from the drop down menu e.g. Category, Colour, Brand, Fabric 
• Select up to three attribute affinities: emerging, growing and established
• Select your combination of attribute names and affinities 

Then click Create segments: 

You might want to include a segment to 'catch' any remaining contacts (i.e. a rest of base segment).

This segment shouldn't have any filters or conditions.  

Click on the + button underneath the last product affinity segmentation segment: 

You can estimate your campaign reach by clicking on the Estimate audience button.

Continue to the Content screen and add the relevant template to the segments. 

Create the rest of your broadcast campaign as usual.

See also:

• Test and preview broadcast campaigns
• Scheduling and sending broadcast campaigns

Reporting

Use the Variants tab in broadcast campaign performance to report on your product affinity segments. 

You can't save product affinity segments, but you can set up the following filter and save it as a dynamic or static segment:

• Received specific broadcast variant - is > [X]