Last Touch

Last-touch attribution — also called last-click attribution — assigns 100% of the conversion credit to the final marketing touchpoint a user interacted with before converting. It’s the simplest attribution model ever invented and the one that ruled digital marketing reports for two decades. GA4 retired it as the default in late 2023, replacing it with data-driven attribution, but last-touch still appears in nearly every analytics tool, every paid-media platform, and every executive dashboard. This guide covers what last-touch does, how GA4’s “paid and organic last-click” variant differs from the original, when last-touch still makes sense, and how to switch off it without losing your reporting baseline.

What is Last-Touch (Last-Click) Attribution

Last-touch attribution gives full credit for a conversion to whichever channel delivered the final pre-conversion interaction. If a user saw a display ad on Monday, searched the brand on Wednesday, opened an email on Friday, then typed the URL directly on Saturday and bought, last-touch credits 100% of the revenue to direct. The first three touches receive nothing.

The terms last-touch and last-click are used interchangeably in practice. Last-click strictly counts only click interactions (paid, organic, email). Last-touch additionally counts non-click interactions like impressions when they’re tracked. In GA4 and Google Ads the UI label is “last-click” and view-through impressions are excluded by default.

The model is deterministic — given the same conversion path, it always produces the same credit allocation. That auditability made it the universal default in Universal Analytics, Adobe Analytics, and almost every third-party platform from 2005 to 2023. Modern attribution reports replaced it because deterministic doesn’t mean accurate.

How Last-Touch Works: Credit Allocation Example

A worked example makes the mechanics obvious. Suppose 170 purchases happened this week across one e-commerce property. The customer paths feeding those conversions look like this in aggregate:

Under last-touch, those numbers are the entire revenue story — and the marketing team optimizes budget against that split. What it hides: the upper-funnel display campaign that drove 60% of brand searches feeding paid search, the organic article that 70% of email subscribers read first, the social posts behind the direct visits. None of those upstream channels appear in the report.

Mechanically, GA4 walks the user’s event stream backward from the conversion to find the most recent traffic source, applies that source’s UTM parameters or referrer to the conversion record, and aggregates at the channel/campaign level. No cross-session weighting, no probability math.

Last-Touch vs First-Touch vs Linear vs Time-Decay vs Data-Driven

The same four-touchpoint path scored under five attribution models produces five different revenue splits. The chart below highlights last-touch in red so the contrast with multi-touch models is unmistakable.

Last-touch and first-touch are single-touch models — only one channel ever gets credit. Linear, time-decay, position-based, and DDA are multi-touch models that distribute credit across the full path. The single-touch family is structurally incapable of reflecting upper-funnel work, which is why it falls apart in any journey longer than a single session.

Last-Touch in GA4: Sunset and Replacement (DDA Default Since 2024)

Last-click was the GA4 default until October 2023. Three changes between 2023 and 2026 reshaped how the model appears in the platform:

• Late 2023. Google switched the default reporting model on every new GA4 property to data-driven attribution. Existing properties were migrated automatically over the following months.
• April 2024. Cross-channel rule-based models — first-click, linear, time-decay, position-based — were removed from GA4. The only remaining cross-channel options became data-driven attribution and “paid and organic last-click”.
• 2025–2026. The “last-click” option in GA4 today is a non-direct variant: it skips direct visits when assigning credit and walks backward to the previous source. This is closer to old-school “last non-direct click” than the pure last-touch logic that ruled UA.

You can still set “Paid and organic last-click” as the reporting model in Admin → Attribution settings → Reporting attribution model. Doing so applies retroactively to all reports — no re-processing needed. But Google Ads’ bidding signals are not affected by this UI choice; ad bidding always uses DDA-derived signals plus its own platform-level rules. See our breakdown of the attribution model changes for the full timeline.

When Last-Touch Still Makes Sense (Short-Cycle, Direct-Response)

Last-touch isn’t useless. It’s a legitimate choice in narrow scenarios:

• Single-session impulse purchases. Categories with average path length under 1.3 touches (impulse fashion, small-ticket goods) have no upper-funnel signal to allocate. Last-touch matches DDA at a fraction of the complexity.
• Mobile-app installs and one-click conversions. Click ad, install in-session — multi-touch has nothing to distribute.
• Direct-response ROAS reporting. If the campaign goal is “drive a click that converts now”, last-click ROAS is the metric the campaign was designed to maximize.
• Audit baseline alongside DDA. Run both in parallel in GA4’s Model comparison. Treat last-click as the floor, DDA as the directional truth.
• Low-volume accounts. Properties under 300 conversions per type per 30 days produce unstable DDA weights — last-click is more honest than a noisy data-driven model.

The pattern: last-touch works when the journey is genuinely short, the conversion genuinely direct, or DDA genuinely starved of data. It fails everywhere else.

Limitations: Underestimates Top-of-Funnel, SEO, and Display

Five structural gaps appear in every account that runs last-touch as the primary model:

• Display and programmatic vanish. Display rarely closes conversions in-session — it introduces the brand. Last-touch shows it unprofitable even when it’s the reason later channels work.
• SEO content undercredited. Long-tail blog content sits mid-funnel. Users learn from it, then convert weeks later via paid search or direct. Last-touch sends those dollars to the closer.
• Email skewed both ways. Newsletters to existing customers collect credit for purchases that would have happened anyway. Prospecting emails that introduce new subscribers get nothing.
• Direct traffic over-credited. When a user types your URL or uses a bookmark, last-touch credits direct — even though the upstream cause was a podcast, offline ad, or invisible referral.
• Brand search inflates paid performance. Branded keywords convert at 5–10× the rate of non-brand. Last-touch gives that inflated rate to paid search, masking the incremental value of brand-discovery channels.

Avinash Kaushik’s Occam’s Razor argues this distortion is the largest single source of mismeasured marketing ROI in modern analytics — bigger than cookie loss, bigger than walled gardens.

Last Non-Direct Click: GA4’s Variant

The exact model GA4 uses today when you select “last-click” is technically last non-direct click. The rule: when the final touchpoint is direct, the model skips it and walks backward to the most recent non-direct source. This patch addresses the over-crediting of direct traffic noted above.

Concretely: if a user’s path was Email → Organic → Direct → Conversion, pure last-touch credits direct. Last non-direct click credits organic search instead, on the assumption that direct visits represent a return user whose original acquisition source matters more.

The non-direct logic uses the cookie or first-party cookie stitching that ties returning sessions back to the original acquisition source. Without that stitching (think Safari ITP cutting cookies after 7 days), the model degrades — direct visits remain direct, and the credit lands where pure last-touch would have placed it. This is one reason cross-device and cross-session continuity matter so much for any rule-based model.

Last-Touch in Paid Channels (Google Ads, Meta Ads Conversion Settings)

Inside the major paid platforms, attribution is configured separately from GA4 — each owns its own settings for ad-bidding decisions:

• Google Ads. Default is data-driven for new conversion actions. Last-click remains an option per conversion action in Tools → Conversions → Settings → Attribution model, and affects which clicks Smart Bidding treats as a “conversion”.
• Meta Ads. Default attribution window is 7-day click + 1-day view, last-click implicit. Adjustable to 1-day click, 7-day click, or 7-day click + 1-day view. No multi-touch model in native UI.
• LinkedIn Campaign Manager. Last-touch within a 30-day window — only option. Multi-touch reporting requires third-party MTA tooling.
• TikTok Ads Manager. Last-click within 7-day or 28-day windows. Same single-touch limitation as Meta.

The practical implication: GA4 may report DDA as the cross-channel truth, but paid platforms are still bidding against last-touch. Reconciling those views is one of the harder problems in modern analytics — see the Bounteous explainer on attribution modeling.

Switching from Last-Touch: Migration Checklist

Moving off last-touch is a procedural change, not a technical migration. GA4 recomputes credit on the fly — no historical re-processing. The work is stakeholder management and cross-tool reconciliation:

1. Run the model comparison report (Advertising → Attribution → Model comparison) for the last 90 days, comparing last-click against DDA. Expect 15–40% per-channel revenue reshaping.
2. Pre-brief stakeholders on which channels will look better and which will look worse. Display, SEO content, and upper-funnel social typically gain. Direct, brand search, and remarketing typically lose.
3. Update budget allocation rules to reflect the new credit weights, and redirect prior last-click ROAS thresholds to DDA-based equivalents.
4. Adjust paid-platform bidding signals. Google Ads accepts DDA-aligned conversions. Meta and LinkedIn don’t — accept that paid bidding remains last-touch and account for it in the split.
5. Keep a parallel last-click view as the audit baseline. Any DDA story has to be reconciled against it.
6. Define a 90-day re-evaluation window to let DDA stabilize before judging channel-level results against pre-switch baselines.

Reporting Last-Touch Despite GA4’s Default (BigQuery, Custom Models)

Last-touch reporting is straightforward to reproduce despite GA4’s DDA default:

• GA4 UI. Set “Paid and organic last-click” as the reporting attribution model in Admin → Attribution settings — applies retroactively to all reports.
• GA4 + BigQuery export. Pure last-touch (including direct) is roughly 30 lines of SQL: group events by user_pseudo_id, walk backward from the conversion event to the most recent session_start with a non-null traffic_source, apply that source/medium.
• Custom MTA platform. Funnel.io, Adverity, or Northbeam ingest GA4 + paid-platform data and let you switch attribution models per report.

The BigQuery route is the most flexible — implement any variant (pure last-touch, last non-direct click, last paid click, last non-brand click) without waiting for Google’s UI. Combine with cross-device tracking and server-side ingestion via Measurement Protocol for any attribution stack reporting on macro conversions with meaningful revenue weight.

Frequently Asked Questions