Field reference: 8 algorithms
Sites without real-time behavioral personalization leak revenue, showing static merchandising to shoppers who have already signaled intent, and letting sessions disengage without a relevant next step. Fulcrum's eight algorithms close that gap, filling every recommendation slot with the item each shopper is most likely to act on, computed live from behavioral signal.
At a glance
Each algorithm targets a distinct moment in the shopper journey, from a first anonymous browse to a repeat customer's reorder window. Fulcrum's waterfall engine calls the right one automatically, so no slot is ever left with a static, generic placement.
In detail
01
Matches an item to the closest things you sell, so a shopper looking at one product immediately sees the alternatives worth considering. It needs no prior behavioral history on the item itself, which makes it dependable for long-tail products and for SKUs you added this morning.
Where it appears
02
Association-rule mining over your own order history, surfacing the products your customers genuinely buy in the same transaction. It returns complements rather than substitutes, so the recommendation completes a purchase instead of competing with it.
Where it appears
03
Collaborative filtering across your whole customer base rather than within one order. It reaches wider than a bundle does, into the adjacent categories that share a customer profile without ever showing up in the same transaction.
Where it appears
04
The products people look at in the same session, whether or not anyone bought. That is the consideration set rather than the purchase, and because view signal is far denser than transaction signal it works earlier in the funnel, while a shopper is still comparing.
Where it appears
05
A shopper's own view history, most recent first, held across the session and across return visits. Nothing is modeled here. It removes the friction of re-finding an item and keeps what someone already liked in front of them.
Where it appears
06
What your storefront is demonstrably selling, scoped to the category the shopper is standing in. It is behavior-agnostic on purpose, which is exactly what makes it the fallback: it is always populated, and it is always relevant to real demand.
Where it appears
07
Event-driven, and the event is a shopper leaving with a full cart: exit intent, inactivity, or a session ending without checkout. It re-presents the cart and the items that go with it at the most recoverable moment in the funnel.
Where it appears
08
A timing model for consumables. It works out roughly how long a shopper's supply lasts from their own purchase history, then surfaces the reorder prompt as that date comes around, instead of waiting for them to remember.
Where it appears
How it works
Every recommendation slot on the page gets filled. Fulcrum works through the algorithms that suit that placement and keeps going until the slot is complete, so a shopper carrying almost no behavioral signal still sees something worth clicking.
Which algorithms suit which placement is decided by the placement itself. A product page wants item-to-item relevance, a cart page wants basket completion, and Top Sellers backs the whole arrangement up as the popularity fallback that guarantees fill no matter how thin the signal is.
Two things people ask
Audience here is behavioral: what a shopper is doing and where they sit in the journey, not their identity status. Every shopper, identified or anonymous, receives the same fully personalized experience.
Abandon Cart Triggers and Replenishment can hand off to downstream email reminders, but that is a separate, optional feature. The core value is real-time, on-site session recovery and repeat-purchase prompting.
Bring your product data, and we will show you exactly how each slot fills, placement by placement.
Start Your Free Pilot