AI Attributes

Ometria's profile optimisation agent uses your transaction data to learn about your customers' behaviour and predict whether or not they are likely to place another order, when they are likely to order, and how much they are likely to spend.

It does this by comparing the contact's behaviour with normal behaviour across your whole account. If your account is set up by brand and region, the comparison happens inside each brand and region instead.

You can use this data to optimise your revenue by targeting and personalising content for strategically valuable customer cohorts, e.g. one time buyers likely to buy again.

Find out more: How are AI attribute segments calculated?

Ometria's predictive metrics are available in the single customer view, or as segment conditions in the customer filter.

You can use them to improve your marketing spend efficiency and results by targeting customers who are most likely to provide value in the coming year.

These conditions all look at your data to make a prediction for 12 months in the future.

Each condition has additional 'which' filters for more granular results.

If your account is set up by brand and region

Some accounts treat each brand and region separately. 

In these accounts, Ometria predicts separately for each brand and region rather than once across your whole account. The conditions above work in the same way, with one extra step: you choose which brand and region the prediction applies to.

A contact who has ordered with more than one brand or region has a separate set of predicted spend, order likelihood and AOV for each. So they can fall into a band with one brand and not with another.

Using brand and region filters

Pick your condition and band as usual, then select a Brand and a Region. You need to pick both. There's no option to match across all brands or all regions at once. You can only choose the brands and regions in your account view.

• Your segment then contains the contacts whose prediction for that brand and region matches the band you picked. For example, take Predicted 12 month customer spend - which is - Well above average, with a brand and region selected. That returns the contacts predicted to spend well above average with that brand in that region. Contacts predicted to spend well above average with a different brand aren't included.
• To build a segment covering more than one brand or region, add a condition for each and combine them with OR.
• The segments below also use conditions based on past orders. Those cover your whole account, so the brand and region you choose only affect the predicted part of each segment.

Two places to set brand and region

Each predictive metric condition has its own Brand and Region selection, separate from the brand and region dropdowns at the top of the customer filter. The dropdowns choose which contacts you're segmenting. The selection on the condition chooses which prediction those contacts are tested against.

• Set both to the same brand and region and you get contacts in that brand and region, filtered by their prediction for it.
• Set them differently and you get the overlap between the two. For example, Brand A and EMEA at the top with Brand B and North America on the condition returns Brand A EMEA contacts who are also predicted to order with Brand B in North America. That finds contacts who shop across both, so it's usually a much smaller group.

A contact only has a prediction for a brand and region where they've ordered enough times. If you pick a combination your contacts don't shop in, your segment comes back empty even though the condition is set up correctly.

Segment one time buyers who are likely to become repeat buyers

Set up the following:

• Customer frequency band - which is - 1 order

AND

• Predicted 12 month order likelihood - which is - Likely to order

OR

• Predicted 12 month order likelihood - which is - Likely to order multiple times

Create a VIP segment

Set up the following:

• Customer spend band - which is - Above average/Well above average

AND

• Predicted 12 month customer spend - which is - Above average/Well above average

Find your future VIPs

Set up the following:

• Customer spend band - which is not - Above average/Well above average

AND

• Predicted 12 month customer spend - which is - Above average/Well above average

Find loyal customers who will stay loyal

Set up the following:

• Customer frequency band - which is - 3-4 orders/5+ orders

AND

• Predicted 12 month order likelihood - which is - Likely to order/Likely to order multiple times

Find customers who will become loyal in the next 12 months

Set up the following:

• Customer frequency band - [which is not] - 3-4 orders/5+ orders

AND

• Predicted 12 month order likelihood - which is - Likely to order/Likely to order multiple times