Marketing Attribution: How to Track What's Driving Revenue

Marketing attribution is the process of determining which advertising touchpoints deserve credit for driving a conversion. When someone sees a Google ad on Monday, a Meta retargeting ad on Wednesday, and installs your app on Friday after an Apple Search Ad, which campaign gets credit for the install? Attribution answers that question.

The answer depends on which attribution model you use, and each model tells a different story.

Attribution Models and What They Measure

Last-touch attribution assigns 100% of conversion credit to the final ad a user interacted with before converting. It's the default for most mobile measurement partners (MMPs) and ad platforms. It's simple to understand and easy to optimize against, which is why it's standard. The downside: it severely undervalues any ad that ran earlier in the user's decision process, and it massively overstates the value of retargeting campaigns that reach users who were already going to convert.

First-touch attribution does the opposite: all credit goes to the first interaction. This overvalues awareness campaigns and ignores everything that happened afterward.

Linear attribution splits credit equally across all touchpoints. If a user saw three ads before converting, each gets one-third of the credit. Better for understanding full-funnel contribution but harder to optimize against in real time.

Time-decay attribution gives more credit to touchpoints closer to the conversion and less to earlier ones. More intuitive for longer purchase cycles where recent interactions genuinely matter more.

Data-driven attribution (available in Google Analytics 4 and some MMPs) uses machine learning to estimate how much each touchpoint actually contributed based on conversion path data. It requires substantial volume to work well — typically 1,000+ conversions per month — and it's a black box, which makes it hard to explain to stakeholders.

For most mobile app campaigns, last-touch with a 7-30 day click window is the right starting point. It's what ad platforms optimize against, which means your campaign bidding and attribution are aligned. Moving to multi-touch or data-driven attribution becomes valuable once you're running enough campaigns across enough channels that you need to understand their relative contribution.

Attribution Windows

An attribution window defines how long after an ad interaction a conversion can still be credited to that interaction. The industry defaults are:

Click-through window: 7 days (Meta default), 30 days (Google default) View-through window: 1 day (Meta default), varies by platform

These defaults matter because they affect how much credit your campaigns receive. A 30-day click window credits your campaign with any conversion that happens within a month of a click, even if the user installed 25 days later without seeing another ad. A 7-day window is more conservative.

Mismatched windows between your MMP and your ad platform are a major source of attribution discrepancies. If Google reports conversions on a 30-day window but your MMP tracks on 7 days, the numbers will never match. Standardize your windows across platforms before drawing conclusions from attribution data.

How iOS Privacy Changes Affected Attribution

Before iOS 14.5, mobile attribution relied on device identifiers (IDFA) to match ad clicks with app installs. Apple's AppTrackingTransparency framework, introduced in 2021, requires users to opt in to tracking. Most don't. Average opt-in rates across apps are 20-30%, which means attribution data for iOS users is incomplete by design.

Apple's SKAdNetwork (SKAN) framework provides an alternative: aggregated, privacy-safe conversion data with a 24-72 hour reporting delay and no user-level information. It tells you that a campaign drove X installs, but not which users those installs came from. This makes it impossible to build lookalike audiences or run personalized retargeting on iOS — two major tactics that relied on user-level attribution.

The practical consequence is that iOS campaigns look less efficient on paper because fewer conversions are attributable. Teams that haven't adjusted for this often underinvest in iOS and over-invest in Android, where GAID-based attribution still works.

Probabilistic vs. Deterministic Attribution

Deterministic attribution relies on a definitive match between an ad interaction and a conversion: a click ID, a device ID, or a first-party login. It's accurate when it works. When it can't match (because of privacy restrictions, cross-device behavior, or cross-browser sessions), conversions go unattributed.

Probabilistic attribution fills the gaps using statistical modeling: if a user whose IP address, device type, and operating system match a user who clicked your ad converts 12 minutes later, that conversion is probabilistically attributed to the campaign. It's less accurate than deterministic matching but more complete — fewer conversions fall into the "direct" bucket where they get misattributed to organic or other channels.

X-Ray handles both, connecting ad platform data, MMP signals, and first-party behavioral data to give you the most complete attribution picture available under current privacy constraints.