Visitor Attribution
Visitor attribution is how Elevate identifies visitors, assigns them to experiment variations, and tracks their actions throughout their session. This page explains how the system works.
How Visitors Are Identified
When a visitor arrives at your store, Elevate assigns them a Visitor ID — a randomly generated unique identifier stored as a browser cookie.
Key details:
The Visitor ID is anonymous — it's a random string, not tied to any personal information
It's generated the first time a visitor reaches your store while Elevate is installed
If a visitor clears their cookies or uses a different browser/device, they'll get a new Visitor ID and be treated as a new visitor
Elevate also tracks a Session ID stored in session storage. This resets when the browser session ends, allowing Elevate to distinguish between separate browsing sessions from the same visitor.
How Visitors Are Assigned to Variations
When a visitor lands on a page with an active experiment, the Elevate script runs the following process:
Check for existing assignment — The script checks whether the visitor has already been assigned to this experiment. If so, they see the same variation as before.
Check eligibility — If the visitor is new to this experiment, the script checks:
Does the visitor match the audience targeting rules (device, country, traffic source, etc.)?
If traffic isolation is enabled, does this visitor fall within the allocated percentage?
Assign a variation — For eligible visitors without an existing assignment:
Experiments: A random number is generated and mapped to a variation based on the traffic allocation percentages (e.g., 50/50, 70/30)
Personalizations: The visitor is always assigned to the personalized experience (non-control variation)
Store the assignment — The assignment is saved to a cookie. This ensures the visitor sees the same variation on every subsequent visit.
This entire process happens in milliseconds, before the page content renders.
How Actions Are Attributed
Once a visitor is assigned to a variation, all subsequent actions are attributed to that experiment and variation. Elevate captures all visitor events throughout the shopping journey — page views, cart interactions, checkout activity, completed purchases, and more.
Full-Journey Attribution
Elevate uses full-journey attribution — if a visitor views a tested page and later completes a purchase (even of a different product), the entire order is attributed to the experiment variation they were assigned to.
This captures the downstream behavioral impact of the experiment. A change on one page can influence the visitor's overall buying behavior — not just purchases of the specific product being tested.
What's Excluded
Certain order types are automatically excluded from attribution:
Draft orders — Created manually in the Shopify admin, not from real visitor behavior
Subscription contract orders — Recurring charges from existing subscriptions, not new purchase decisions
Order Attribution via Cart
When a visitor who's part of an experiment adds items to their cart, Elevate stores attribution data in Shopify's cart attributes. This data travels with the cart through checkout and is attached to the resulting order.
This is how Elevate connects completed orders back to the correct experiment and variation — even if the order is processed after the visitor's session has ended or if the visitor completes checkout on a different page than the one being tested.
Concurrent Experiments
A visitor can participate in multiple experiments simultaneously, as long as the experiments don't conflict.
However, if Traffic Isolation is enabled on any experiment, visitors assigned to that isolated experiment will not enter other isolated experiments. Non-isolated experiments are unaffected — visitors can be in multiple non-isolated experiments at the same time.
Cross-Device Limitations
Elevate's attribution is cookie-based and device-specific. If a visitor browses on their phone and later purchases on their laptop, these are treated as two separate visitors. This is a fundamental limitation of client-side analytics that affects all A/B testing tools.
For this reason, experiment results reflect the behavior patterns of the majority of your visitors — most of whom complete their journey on a single device.
Last updated