Asks Cooee's recommendation service, in real time, what this particular visitor is most likely to respond to. You control the answer by weighting eight signals against each other.
AI-powered, and computed live
This is Cooee's AI-powered product source, highlighted in the type list.
Its products are worked out live, per visitor, at the moment the campaign is shown, rather than read from a list prepared earlier. That is what lets it react to what the visitor has done in the current session, and it means a newly added product can be recommended as soon as it exists.
The trade is a service call in the display path, where the other sources read your catalogue or your store's own data directly.
Options
Eight sliders, each 0 to 5. Zero switches a signal off; leave every slider at zero and the service falls back to its own defaults.
Based on how products look
Signal | Recommends |
Looks Similar | Products that look alike by image analysis — same style, colour or pattern. |
Looks Different | Visually varied products, to add discovery and avoid near-identical items. |
Based on what this shopper is predicted to do
Signal | Recommends |
Likely to Buy | What this shopper is predicted to purchase, learned from the purchase patterns of similar shoppers. |
Likely to View | What this shopper is predicted to browse. |
Likely to Cart | What this shopper is predicted to add to cart. |
Based on what shoppers actually did together
Signal | Recommends |
Bought Together | Products genuinely purchased in the same order by past shoppers. |
Viewed Together | Products viewed in the same browsing session. |
Carted Together | Products added to the same cart, whether or not the order completed. |
The middle group is predictive — it generalises from shoppers who resemble this one. The last group is observational — it only reports what actually happened. Predictive signals reach further on thin data; observational ones are more literal.
Entry point
Generic, locked. The recommendation is built from who the visitor is and what they have done, not from the page they are on, so it works anywhere on the site.
Filters
Set | Available |
Primary — static | No |
Secondary — static | No |
Secondary — dynamic | No |
No filter set runs, so the form shows no filters. Constrain the output through the weights instead. If you need explicit product conditions, Similar is the source built for that.
Good to know
Weights are relative, not absolute. Looks Similar at 4 against everything else at 1 gives the same blend as 4 against 1 — it is the ratio that matters.
A brand-new visitor has little for the predictive signals to work from, so the observational ones carry more of the result early in a session.
This is the source to reach for when you want the selection to differ per visitor. Every other source shows the same products to everyone in the same context.