Glossary/Attribution window
What is attribution window?
An attribution window is the period following an ad interaction during which a subsequent order is credited to that ad. Marketplace ad platforms define their own windows, distinguish click-through from view-through interactions, and apply their own tie-breaking rule when a shopper touched several ads before buying. The window is a reporting convention, not a physical fact: the same shopper, the same purchase, and the same campaign can produce different attributed revenue on two platforms purely because the windows differ. Every ROAS, ACOS, and cost-per-order figure a seller reads inherits whatever window produced it.
01/Formula
Example
A shopper clicks a Shopee ad on Monday, does not buy, returns through search on Thursday and buys a ฿1,200 item. Inside the window: the order is ad-attributed. Campaign ROAS rises. Outside the window: the order is organic. Campaign ROAS falls. Same shopper, same purchase, same ad. Only the reporting rule changed.
02/In detail
Why do platforms use different windows?
Because the window encodes a belief about how long an ad keeps working, and each platform holds a different one. A marketplace search ad reaches a shopper already close to purchase, so a short window is defensible. A short-form video ad reaches someone who was not shopping at all, so its platform argues for a longer window and for counting views as well as clicks. Neither is wrong on its own terms. The problem appears when a seller compares the two: a longer window with view-through counting will always report more attributed revenue for the same real effect, so the platform with the more generous convention looks like the better channel.
What is the difference between click-through and view-through attribution?
Click-through attribution credits an ad only when the shopper clicked it. View-through attribution also credits ads that were merely displayed, on the theory that exposure influenced the later purchase. View-through is far weaker evidence, because being shown an ad is close to costless as a signal — a large share of people shown any ad for a popular product would have bought it anyway. When a platform reports a combined figure, the view-through portion is the part most likely to be non-incremental, and it is worth reading the two separately whenever the reporting allows it.
How should sellers handle the window in practice?
Three habits cover most of the damage. Fix a window per platform and stop changing it, so that period-over-period comparisons mean something. Never compare raw ROAS across platforms whose windows differ; compare cost per acquired order against unit contribution instead, which is at least denominated in the same currency of truth. And treat total revenue against total ad spend as a sanity check on the sum of attributed figures — when attributed revenue across campaigns approaches or exceeds total shop revenue, the windows are double-counting and the individual campaign numbers cannot all be true.
03/Why it matters
The trap, in one paragraph.
Attribution windows are the most consequential invisible setting in marketplace advertising. Sellers make budget decisions by comparing reported ROAS across campaigns and platforms, and those comparisons silently assume a common measurement standard that does not exist. A channel can win the budget on nothing more than a more generous crediting rule.
Common mistake
Reading attributed revenue as caused revenue. Attribution answers “which ad touched this order last, within the window?”. It never answers “would this order have happened anyway?”. Those are different questions, and only a holdout test answers the second one.
04/In DataGlass
How Attribution window is used in DataGlass.
DataGlass reads each platform’s attributed orders as the platform reports them and then re-prices those same orders with real economics, so the attribution convention affects which orders are in scope but never what an order is worth. Where the attributed figures across a shop stop reconciling with total order data, the discrepancy is surfaced rather than absorbed.
05/Sources
- [1] Attribution (marketing) — overview
Background on attribution models, lookback windows, and the last-touch convention most marketplace ad platforms apply by default.
- [2] Shopee Ads Thailand — ROAS definition
Shopee reports ROAS from ad-attributed sales; the attribution rules that determine which orders enter that numerator are platform-defined and documented separately from the metric itself.
- [3] TikTok for Business — Help Centre, ad metrics
TikTok documentation of click-through and view-through attribution settings and their effect on reported conversions.
Related terms