Attribution Window

An attribution window (also called a look-back window or conversion window) is the defined time period after an ad interaction — a click or an impression — during which a conversion can be credited to that ad. It determines how long marketing platforms tie sales, sign-ups, or other actions back to paid touchpoints. The window is the primary mechanism platforms use to construct Marketing Attribution reports, and differences in window length across platforms are the leading cause of cross-channel double-counting and ROAS inflation.

---

How it works

Two attribution types

Attribution windows have two distinct modes (Anderson Collaborative, 2026):

• Click-through attribution (CTA) — credits conversions occurring after a user actively clicks an ad. Treated as a stronger intent signal.
• View-through attribution (VTA) — credits conversions occurring after an ad impression with no click. Captures influence from display, video, and social formats.

When both windows are active simultaneously, click attribution takes precedence over view attribution across most platforms, on the basis that active engagement is a stronger intent signal. (Cometly, 2026)

As of 2026, a third type has emerged:

• Engage-through attribution — introduced by Meta in March 2026 to cover non-link-click interactions (reactions, comments, shares, saves, engaged views); fixed at a 1-day window only. (Jon Loomer Digital, 2026)

Window configuration

Window length can typically be set to 1, 7, 14, 30, 60, or 90 days depending on the platform. Shorter windows exclude conversions from customers with longer decision cycles; longer windows capture more conversions but increase the probability of crediting the ad for an organically driven purchase. Practitioners advise aligning the window to the brand's observed purchase cycle rather than accepting the platform default. (YouTube companion content, 2026-03-31)

Click window length and ROAS direction

The ROAS effect of shortening a window is channel-dependent:

---

Platform defaults (as-of 2026-08-18)

Google Ads

Window type	Default	Available options
Click-through	30 days	1, 7, 14, 30, 60, 90 days
Engaged-view	1 day	1, 7, 14, 30, 60, 90 days
View-through (impression)	1 day	1, 7, 14, 30, 60, 90 days

(ALM Corp, 2026)

Every new conversion action defaults to Data-Driven Attribution (DDA), which distributes credit across multiple touchpoints using machine learning (as-of 2026). Changes to conversion windows apply prospectively only — historical data is not retroactively adjusted. (Cometly, 2026)

GA4 change — April 2026: Google restructured GA4's attribution settings: the default lookback window for acquisition conversion events was shortened from 90 days to 30 days; other conversion events retained the 90-day window. Some accounts were silently switched from last-click to DDA with no in-platform notification. Channels contributing to upper-funnel interactions started receiving more credit under the updated DDA logic; last-touch channels received less. (YouTube companion content, 2026-03-13)

GA4's DDA silently falls back to last-click — without any visible flag — if an account has fewer than 400 conversions for a specific key event or fewer than 20,000 total conversions within the lookback window (as-of 2026). (as-of 2026-08-18) (Splinternet Marketing, 2026)

Meta Ads

Default as of Q2 2026: 7-day click + 1-day view + 1-day engage-through. (Jon Loomer Digital, 2026)

January 12, 2026: Meta permanently removed the 7-day view and 28-day view attribution windows from the Ads Insights API. The maximum view window is now 1 day. Advertisers relying on the removed windows saw 15–30% drops in reported conversions. (Dataslayer, 2026)

March 2026: Meta redefined "click-through attribution" to require an actual link click (sending someone to a website, lead form, app, or Messenger). Previously, social interactions — likes, shares, saves, comments — counted as "clicks" for attribution purposes. Non-link-click interactions were moved to the new "engage-through" category with its 1-day window only. (Jon Loomer Digital, 2026)

Prior to January 2026, Meta offered: 1-day click, 7-day click, 28-day click, 1-day view, 7-day view. The 28-day click window had already been removed in a prior cycle. (Conversios, 2026)

Amazon Ads

January 1, 2026: Amazon replaced its blanket 14-day view-through window for Sponsored Brands and Sponsored Display (vCPM) campaigns with a "shopping-signal enhanced last-touch model" that uses machine learning to assess whether an ad view genuinely influenced a purchase before crediting it. Amazon retains legacy "all views" metrics through Unified Reporting for backward comparability. This change reduced view-attributed reported conversions by an estimated 15–30% for affected campaigns. (Amazon Ads, late 2025; ppc.land, 2026)

As of 2026, Amazon effectively runs two parallel attribution models simultaneously — the legacy 14-day view window for backward comparability and the new ML-signal model for optimization and bidding — creating divergent ROAS figures depending on which interface or API endpoint is queried. (AdBeacon, 2026 — vendor source; potential conflict of interest)

Amazon Attribution (off-Amazon measurement for non-Amazon channels driving to Amazon listings) uses a 14-day attribution window for all its reports (as-of 2026). (date unknown — treat as potentially stale) (Amazon Ads Help)

TikTok Ads

Window type	Default	Available options
Click-through	7 days	1, 7, 14, 28 days
View-through	1 day	Off, 1, 7 days
Engaged view-through (EVTA)	N/A (opt-in)	1, 7 days

(TikTok for Business Help, date unknown)

TikTok Shop Ads specifically use a 7-day click / 1-day view window; attribution settings are fixed at the ad group level and cannot be changed after the ad group is published. (TikTok for Business Help, date unknown)

No major TikTok attribution window changes were found for 2025–2026 in this research pass (absence of evidence — not confirmed at primary source).

Criteo

Criteo reportedly uses a 30-day post-click and 1-day post-view window for retail media campaigns by default.

---

Industry standards

IAB/MRC Retail Media Measurement Guidelines (January 2024)

The IAB/MRC Retail Media Measurement Guidelines (published January 2024) state that typical attribution windows for digital measurement are often 3, 7, 14, 28, or 30 days, and that retail media organizations are encouraged to establish consistent and comparable windows for standardized product types. (IAB/MRC, January 2024)

IAB Europe Commerce Media Measurement Standards V2.1 (May 2026)

IAB Europe released Commerce (Incl. Retail) Media Measurement Standards V2 on January 22, 2026, following a public comment period from September–November 2025. It was updated to V2.1 in May 2026. (IAB Europe, May 2026)

Key requirements (as-of 2026-08-18):

• 30-day lookback window as the default for Commerce/Retail Media reporting
• Retailers and ad tech providers must offer flexible, customizable windows within that 30-day envelope
• Windows must be "empirically supported" and "logical with regard to campaign objectives and category sales cycles"
• Expanded five generic timeframes for new-to-brand and new-to-category measurement
• Certification grace period: organizations may comply with V1 or V2 until end of July 2026; after that, certification only available under V2. (ExchangeWire, January 22, 2026)

IAB/MRC Attention Measurement Guidelines (November 2025)

The IAB/MRC Attention Measurement Guidelines (finalised November 2025, developed with 200+ industry members) specify that attribution windows "must be empirically supported" and logical with regard to campaign objectives and category sales cycles. (IAB/MRC, November 2025)

Practitioner consensus

In the absence of a universal IAB or MRC mandate for a specific click or view window outside retail media, published practitioner content gravitates toward 7-day click / 1-day view as a balanced default for ecommerce. (Cometly, 2026)

A Forbes Tech Council piece (November 2025) argued that advertisers must demand consistent, comparable attribution methodologies from retail media networks before setting 2026 budgets, noting that ROAS figures across networks are not built on equivalent window definitions. (Forbes Tech Council, November 12, 2025)

---

ROAS inflation and double-counting

Window-driven double-counting is a structural feature of multi-channel advertising. A concrete scenario reported by practitioners: a shopper sees a Meta video ad on Day 1 (not clicking), clicks a Google Shopping ad on Day 15, and purchases on Day 16. Meta claims the conversion within its 1-day view window; Google claims it within its 30-day click window; Shopify records one order; platform-reported revenue doubles. (AdsX, 2026 — vendor source, potential conflict of interest)

View-through attribution is the largest structural source of cross-channel overcounting. A person served an ad who does not click, and then purchases via any other channel within the view window, is counted as a conversion by the platform that served the impression. (Cometly, 2026)

The mismatch between Google's 30-day click window and Meta's 7-day click window means the same conversion can be claimed by both platforms for up to 7 days of the customer journey, with no native deduplication between them. (Dataslayer, 2026)

Practitioners are advised to wait at least 21 days after any attribution window change before reading ROAS figures as meaningful, because Smart Bidding retrains its credit-signal distribution during this period; daily ROAS can swing 20–30% in either direction without reflecting any real change in business performance. (YouTube companion content, 2026-03-31)

---

Retail media attribution windows and closed-loop measurement

Retail Media Attribution in closed-loop networks differs structurally from open-web attribution. Retailers with loyalty programs (Amazon, Tesco, Walmart) match ad exposures to actual purchase transactions using first-party identity (logged-in user IDs, loyalty card numbers), removing the dependency on pixel-fired events. This makes the attribution window less about probabilistic inference and more about defining which confirmed purchases post-exposure to count.

55% of US advertisers report inconsistent targeting and attribution from retail media networks, reflecting structural differences in how each network defines and measures a conversion (as-of 2026). (AdBeacon, 2026 — secondary citation; original source is EMARKETER, paywalled, unverified)

US retail media advertising reached approximately $58.79B in 2025 and is projected at $69.33B in 2026, growing ~20% annually (as-of 2026). Attribution inconsistency is described as a structural risk to continued budget scaling. (AdBeacon, 2026 — vendor source)

The 2026 retail media stack is moving toward deterministic-first identity — logged-in user IDs, hashed emails, retailer loyalty IDs — with probabilistic methods kept as fallback. (Osmos.ai, 2026 — vendor source)

Amazon as closed-loop benchmark: Amazon's Purchase window is linked directly to verified transaction data in its own retail system, not pixel-fired events — described as the structural differentiator of closed-loop attribution. (Osmos.ai, 2026 — vendor source)

Tesco Media: Powered by Dunnhumby; Tesco Clubcard membership reached 23 million members out of approximately 28.3 million UK households (~81%) as-of 2024.

Walmart Connect (Cannes Lions 2026): Jon Gibs (group director, media insights and measurement, Walmart Connect) stated that retail media credibility now rests on incrementality measurement and independent cross-retailer measurement — not conversion attribution windows alone. He framed the core question not as "what was attributed?" but as "what is the difference between what happened and what would have happened without the media?" — positioning Incrementality Testing as the next chapter beyond window-based attribution. (Circana / Walmart Connect, June 2026)

---

Attribution windows vs. incrementality

Attribution windows are a proxy for the deeper question of causal impact. Window-based attribution assumes the ad caused the conversion, but all it demonstrates is temporal proximity. Incrementality Testing (geo-lift tests, holdout experiments) is the alternative that directly measures the counterfactual — what would have happened without the media.

Google lowered the minimum spend threshold for its geo-lift incrementality testing tool to $5,000 in late 2025, extending access to SMB ecommerce brands (as-of late 2025).

Third-party measurement tools (Triple Whale Compass, Northbeam, Hyros) attempt to reconcile cross-platform attribution by layering Media Mix Modeling (MMM) and incrementality on top of multi-touch attribution (MTA). A practitioner critique from 2024 argues that pixel-based MTA tools cannot resolve the fundamental cross-channel double-counting problem because each tool depends on platform-side credit signals with their own window assumptions baked in. (YouTube, 2024-08-13)

---

Key terms

Term	Meaning
Attribution window	Time period post-interaction within which a conversion is credited to an ad
Look-back window	Synonym for attribution window; used interchangeably by practitioners and platforms
Click-through attribution (CTA)	Credits conversions following an ad click
View-through attribution (VTA)	Credits conversions following an ad impression (no click)
Engage-through attribution	Meta-specific (March 2026): credits conversions following non-link social interactions within 1 day
EVTA	Engaged view-through attribution; TikTok-specific; for engaged-view formats (1 or 7 days)
Double-counting	Same conversion claimed independently by multiple platforms within their respective windows
Closed-loop attribution	Matching ad exposures to actual purchase records via first-party identity (loyalty data, login)
DDA	Data-Driven Attribution (DDA); ML-based multi-touch model; default in Google Ads since 2022
Smart Bidding retraining	Period (typically 21 days) during which Google's bidding algorithm recalibrates after a window change

---

Gaps and open questions

• Criteo, Walmart Connect, Instacart Ads, CitrusAd/Epsilon — specific click and view window lengths not publicly documented in any 2025–2026 source found
• Tesco Media — specific attribution window lengths not publicly available
• Amazon Seller Central vs Vendor Central window split — Feedvisor source has unknown date; verify at amazon.com directly
• Google Ads primary support page not fetched directly — all Google window data sourced via third-party summaries
• IAB/MRC January 2024 primary PDF — full text not extracted in this research pass
• No independent validation of AdsX's 2.3x ROAS overstatement figure