Revenue attribution

9 min

Attribution is how Personizely decides which campaigns and widgets get credit for a goal hit, such as an order. When a visitor reaches a goal, every campaign they saw within the attribution window (7 days by default, configurable from 2 to 30 days) that tracks that goal gets credit, including control groups. Credit isn't split and isn't last-touch: each campaign gets the full conversion and its full value.

How it works

A visitor is exposed. They see a campaign variation, or they're assigned to its control group. Each exposure is recorded with the variation and a timestamp.

The visitor reaches a goal. For example, they buy something. On Shopify and Wix the purchase is recorded server-side from the platform's order data. See Shopify order tracking and profit.

Personizely looks back over the attribution window from the moment of the goal hit. It finds every campaign the visitor was exposed to in that period whose goal list includes the goal.

Each of those campaigns is credited with the goal hit and its value (revenue, plus profit and shipping where tracked). If the visitor was exposed to the same campaign several times, the credit goes to their latest exposure to it.

Widgets are credited the same way, with the extra rules described below.

Attribution happens once, when the goal hit is recorded. The links between that goal hit and the campaigns it credits are stored, and reports read from them.

The attribution window

Setting	Where	Values / default
Attribution window	Settings → Tracking	2–30 days, default 7

The window counts back from the goal hit, so only exposures that happened before the conversion and no more than N days earlier count. A visitor who sees a test on day 1 and orders on day 5 is credited with a 7-day window. If they order on day 9, the order isn't credited to that test.

The window applies to goal hits recorded after you change it. Conversions that were already attributed keep their credit, and older conversions aren't re-attributed.

Choosing a window

Short (2–3 days): impulse purchases and low-consideration products. Less chance of crediting a test for a purchase it didn't influence.

Default (7 days): most ecommerce stores.

Long (14–30 days): high-ticket products with long research cycles, and B2B. More conversions are credited, but more of them are coincidental, which adds noise equally to control and variations.

What gets credited

Campaigns (A/B tests and personalizations)

Every campaign seen within the window that has the goal in its goal list.

Control exposures are credited too. Control is the baseline for comparisons, so it's measured exactly like the variations.

A campaign that doesn't track the goal gets nothing for it, even if the visitor saw it.

Widgets

A widget is credited for a goal it tracks when, within the window, the visitor:

converted in the widget (a Track conversion automation ran, for example on a form submit or button click), or

was in the widget experiment's control group, or

saw the widget, if Attribute on view is turned on for the widget.

Purchases can also be credited to widgets regardless of goals:

Coupon codes: if an order uses a coupon code a widget applied to the cart, or that the visitor copied from a widget, the widget is credited. Shopify and Wix.

Product click-throughs: if the visitor clicked a product in a widget's Products element and then bought that product, the widget is credited. Shopify and Wix.

In a widget A/B test with a control group, turn on Attribute on view. Without it, control visitors are credited on assignment while variation visitors are credited only after converting in the widget, which makes the variations look worse than they are. The report shows a warning when this applies.

Values

Value	Source	Notes
Revenue	The goal hit's value. For orders, the order total in your store currency	For non-commerce goals, the goal's Value setting or the value passed in code
Profit	Order subtotal after discounts, minus the cost of goods from Shopify	Shopify only. Needs Track profit analytics. See Shopify order tracking and profit
Shipping	The order's shipping charge	Added to profit when Calculate shipping price towards profit is on

How attribution shows up in reports

Report date ranges. A report counts exposures inside the date range you select. Conversions are counted up to the end of the range plus the attribution window, so visitors exposed near the end of the range still get credit for what they buy afterward.

Campaign and widget lists show totals (goal hits and value) for the primary goal only. Reports let you look at any of the campaign's goals.

Overlapping tests. A visitor in two running tests that both track Purchase credits the same order to both. Each test's comparison stays fair, because control and variations of each test are equally exposed to the other one. To keep tests fully separate, use Campaign groups (mutually exclusive tests).

Order tags. On Shopify you can tag each order with the campaigns and variations it was credited to. See Shopify order tracking.

Behavior details and limitations

Not last-touch, not split. Three campaigns seen before an order each get 100% of the order. Don't add up revenue across campaign reports: the totals overlap.

Latest exposure per campaign. In a personalization where a visitor matched different variations over time, the variation they saw last before the conversion gets the credit.

The visitor must be recognized. Conversions are credited to the visitor Personizely tracked on your site. If a visitor browses on their phone and buys on their laptop without logging in, the two can't be linked.

Changing the window isn't retroactive. See the note above. To measure a test under a different window, set the window before you start it.

Deployed and ended campaigns keep the credit they already received. A deployed variation isn't tracked, so it gets no new credit.

Outliers count in full. A single very large order is credited like any other. Review them with outlier detection.

Examples

Price test plus a free-shipping bar. A visitor sees variation B of a price test and a free-shipping bar widget that tracks Purchase. They submit the bar's email form and buy 3 days later. Both the price test (variation B) and the widget are credited with the full order.