Cross-Device Attribution

Analytics

Also: Cross-Device Tracking · Multi-Device Attribution

What it doesConnects the same person across devices
MethodsDeterministic and probabilistic
Watch forPrivacy changes shrinking the signal
Depends onLogged-in user data or modelling

Quick definition

Cross-device attribution is the process of connecting a single customer's behaviour across multiple devices, such as a phone, laptop and tablet, so their full journey from first touch to conversion is visible as one path rather than several unrelated ones.

Try it: how attribution models redistribute credit
Phone social adPaid social (mobile)
Laptop searchOrganic search (desktop)
Tablet emailEmail (tablet)
Desktop checkoutDirect (desktop)

A customer touches all four, then converts.

Credit distributed

Phone social ad
0%
Laptop search
0%
Tablet email
0%
Desktop checkout
100%

Last-click credits the desktop checkout and ignores every device that came before. Mobile looks useless. It isn't.

How it varies across Australia

Most Australian businesses have incomplete cross-device visibility. The gap is largest for businesses without a logged-in product or CRM that ties device sessions together. Modelled approaches help fill the gap but signal quality has declined as privacy restrictions have tightened.

See data and tracking scores across Australian industries

Two ways to stitch devices together

Deterministic matching

Matches devices using a consistent identifier like a login or email. Accurate when the customer is logged in across devices.

Requires a logged-in product or CRM
Probabilistic matching

Estimates device connections using signals like IP address, location and behaviour patterns. Less accurate but broader reach.

Degrading as privacy rules tighten

What it actually means

Most customer journeys now span more than one device. Someone sees a social ad on their phone during lunch, searches on their work laptop that afternoon, and converts on their home computer that evening. Without cross-device attribution, your analytics sees three separate visitors, three separate sessions, and three separate channels each claiming their share of credit.

Cross-device attribution tries to stitch those sessions back into one journey. There are two ways it does this: deterministic and probabilistic. Deterministic matching works when the customer is logged in on each device, so you have a consistent identifier. Probabilistic matching works when they're not, using signals like IP address, browser fingerprint, timing patterns and location to estimate that two sessions belong to the same person. Deterministic is accurate. Probabilistic is educated guesswork.

The challenge, and it's a real one, is that both methods are being squeezed. Apple's App Tracking Transparency changes, tighter browser privacy rules and the decline of third-party cookies have all reduced the data available for device stitching. The result is that cross-device attribution is getting less accurate at exactly the point when journeys are more fragmented.

This matters for any metric built on top of attribution: CPA, ROAS, channel mix decisions. If a meaningful share of your conversions come from multi-device journeys that your attribution model can't follow, every channel number is distorted.

Most attribution reports show four journeys. Cross-device attribution reveals it was one person the whole time.

How it shows up

Cross-device attribution shows up in your analytics as the difference between the number of converting users and the number of sessions involved in those conversions. If you have 200 conversions but Analytics shows 340 sessions in the converting paths, device fragmentation is part of that gap.

It also shows up in platform-reported versus CRM-reported conversion counts. Google and Meta both do their own cross-device stitching within their walled gardens, which is why their numbers often look better than your server-side data. Their model includes more cross-device signal than your analytics tool does, but it only works within their ecosystem.

The Australian context

Australian mobile usage is high relative to desktop for browsing, but conversion still leans toward desktop in many categories including finance, B2B and high-consideration retail. That pattern creates a specific cross-device problem: mobile gets most of the upper-funnel exposure but desktop gets the conversion credit, which makes mobile look weak in last-click attribution.

The Privacy Act amendments and the evolving guidance from the Office of the Australian Information Commissioner (OAIC) are also pushing Australian businesses toward consent-first data collection, which affects how much cross-device signal is available. Building a first-party identity layer is increasingly both a legal obligation and a measurement necessity.

Where people get this wrong

Assuming single-device attribution is accurate enough.If a meaningful share of your customers use more than one device before converting, single-device attribution systematically misattributes credit to the final device regardless of where the real influence happened.
Trusting ad platform cross-device numbers as the full picture.Google and Meta's cross-device matching only works within their own ecosystem. Journeys that cross platforms, apps and direct channels are invisible to any single platform's stitching.
Treating probabilistic cross-device matching as deterministic.Probabilistic matching is a statistical estimate, not a confirmed identity. Decisions made on it should carry that uncertainty explicitly. Acting on it as though it's certain produces overconfident channel budgeting.

Related terms

Common questions

How does cross-device attribution work without third-party cookies?

Without third-party cookies, the reliable method is first-party identity: getting customers to log in or provide an email early in their journey so you can match sessions yourself. Probabilistic methods that relied on third-party data are degrading. Walled gardens like Google and Meta maintain their own cross-device graphs but those only cover activity within their platforms.

Does Google Analytics do cross-device attribution automatically?

Google Analytics 4 includes cross-device reporting for users who are logged into a Google account and have opted in to sharing data. For users who aren't logged in, GA4 uses modelling to estimate cross-device behaviour. The modelled numbers are directionally useful but not individually accurate.

Should I invest in a Customer Data Platform to fix cross-device attribution?

A Customer Data Platform (CDP) helps if you have meaningful first-party data to unify and the volume to justify the cost. The underlying problem is identity, not tooling. If customers don't log in or identify themselves, a CDP has nothing to stitch. Fix the identity capture first.

How much of my conversion data is affected by cross-device gaps?

Varies by industry and audience. Categories with high mobile browsing and desktop conversion, such as finance, B2B and high-consideration retail, are most affected. A rough diagnostic: compare the number of converting users in your analytics to the number of sessions in those converting paths. A large gap suggests device fragmentation is a real factor.

Keep exploring

About New Rebellion

New Rebellion is a marketing intelligence consultancy. We build tools, score Australian businesses on how their marketing actually performs, and publish Debrief every day. This dictionary is part of how we work in the open.

How we think →