How Revlance learns what each shopper wants
Four layers, zero guesswork. Built to run in DTC storefronts without a data science team.
Five signals most personalization tools discard
Revlance captures behavioral events at the session level — not just clicks and purchases. Most recommendation engines only look backward at transaction history. We look at what a shopper is doing right now, in the current session, and treat that as the primary signal.
- Scroll depth pattern — how far and how fast reveals browse intent vs. hunting
- Hover duration per product tile — lingering without clicking is a strong consideration signal
- Add-to-cart and remove events — bidirectional intent across the same session
- Cross-session return velocity — how quickly a visitor comes back, and what changed
- Category affinity drift — when a shopper's interest pattern shifts between sessions
A preference vector per visitor — not a segment
Each session builds a preference vector specific to that visitor — not a cluster assignment, not a "shoppers like you" segment. The preference model starts influencing recommendations after 3 behavioral events in a session, so new visitors don't sit in a cold-start plateau seeing the same generic results as everyone else.
- Preference vector per visitor, updated after each meaningful interaction
- Category affinity scores weighted by recency and dwell time
- Price band inference from browsed vs. carted product price distributions
- Cross-session continuity — model persists return visit context
- No segment assignment — every visitor is treated as a distinct individual
Every touchpoint, personalized
The same preference model drives recommendations across homepage modules, collection grids, and Klaviyo email blocks. One integration with your Shopify or BigCommerce store, three output channels, consistent ranking logic across all of them.
Reordered in real time based on this session's signals. No manual rules.
Revlance surfaces items most likely to convert above the fold, based on scroll-depth signals from prior sessions.
Recommendations draw from behavioral history, not just last purchase category. Drop into any Klaviyo flow.
Conversion is the signal that matters most
Every purchase, click, and cart abandonment feeds back into the preference model. There's no black box here: your dashboard names the signal that drove each recommendation, and shows you the before/after on conversion rate when the model changes its ranking. If the numbers don't hold, you'll see why.