Attribution in GA4: Models, Data-Driven & Conversion Path Analysis

Q: What attribution model does GA4 use by default?

GA4 uses data-driven attribution (DDA) as the default for all conversion reports since late 2023. This replaces last-click as GA4's default and applies to every channel — Google Ads, organic, paid search, email, social, referral, display, and direct.

Q: How do I change the attribution model in GA4?

Go to Admin → Property → Attribution Settings . You can change the reporting attribution model (data-driven, last-click, etc.) and the lookback window (30/60/90 days for acquisition). Changes apply to all subsequent reports — historical data is recomputed retroactively for the new model in most reports.

Attribution is how analytics tools assign credit for a conversion across the marketing touchpoints that influenced it — display ads, organic search visits, emails, paid clicks, direct visits. The model you choose determines which channels look profitable and which look wasteful. GA4 uses data-driven attribution by default, replacing the old last-click logic that dominated marketing reports for two decades. This guide explains the six standard attribution models, how GA4’s data-driven approach works, conversion path and attribution window settings, and when each model is the right choice.

What Is Attribution?

Attribution is the method of distributing conversion credit across the user interactions that occurred before a purchase, signup, or other key event. A real customer journey rarely involves a single click — a typical B2B buyer might see a display ad on Monday, search the brand on Wednesday, read an email Friday, and buy directly the next week. Attribution decides how the resulting $100 in revenue is split across those four touchpoints.

Without attribution, every source and medium would be reported in isolation, and you’d have no way to compare how channels actually contribute. Modern analytics platforms — GA4, Adobe Analytics, Piwik PRO — bake attribution directly into their reports so that channel-level numbers reflect a chosen credit model, not raw last-touch counts.

Why Attribution Matters in Marketing Analytics

Attribution is the bridge between ad spend and revenue accounting. Three concrete reasons it matters:

Budget allocation. If you only credit the last click before purchase, channels that introduce customers to your brand (display, social, organic) appear unprofitable and get defunded — even when they’re the reason later channels work at all.
Channel comparison. Comparing Google Ads against organic search, email, and paid social only makes sense when each is measured the same way. Attribution standardizes that comparison.
Forecast accuracy. Revenue forecasts built on last-click data are systematically biased toward late-funnel channels and undershoot when upper-funnel spend changes.

The attribution model is therefore a strategic choice, not a technical detail. Switching from last-click to data-driven typically reshapes channel-level revenue reports by 15–40%, and that reshaping flows into every downstream marketing decision. For deeper context on how attribution feeds revenue analysis, see our pillar on tying revenue to traffic.

The 6 Common Attribution Models

Six attribution models dominate practical use. Each weighs touchpoints differently:

Attribution models comparison bar chart showing how first-click last-click linear time-decay position-based and data-driven attribution distribute credit across four touchpoints — Six attribution models compared on the same conversion path. Each makes a different claim about which touchpoints “deserve” the credit.

Model	How credit is split	Pro	Con	Use when
First-Click	100% to the first touchpoint	Highlights discovery channels	Ignores everything after the first interaction	Measuring brand-awareness campaigns in isolation
Last-Click	100% to the final touchpoint before conversion	Simple, deterministic, easy to audit	Over-credits the closer, ignores upper-funnel	Short, single-session purchase journeys (impulse buys)
Linear	Equal share across all touchpoints	No bias toward first or last	Treats trivial and important touches as equal	Reporting baseline; sanity-check against other models
Time-Decay	Recent touchpoints get exponentially more credit (7-day half-life)	Reflects intent build-up near conversion	Underweights early-funnel channels	Short consideration cycles (1–2 weeks)
Position-Based	40% first + 40% last + 20% split among middle touchpoints	Recognises both discovery and closing	Ignores the relative importance of middle touches	Mid-length B2C journeys with 3–5 touches
Data-Driven (DDA)	Algorithmic — credit derived from observed lift across actual paths	No fixed assumptions, adapts to your data	Black-box, requires conversion volume to train	GA4 default; any account with regular conversion volume

The first five are rule-based — they apply a fixed formula regardless of what the data says. Data-driven is fundamentally different: it learns from your actual conversion paths and assigns credit based on which touchpoints statistically lift the conversion probability. That’s why Google made DDA the GA4 default in 2023.

GA4’s Default: Data-Driven Attribution (DDA)

Since late 2023, every new GA4 property uses data-driven attribution as the default reporting model for all conversions. This was a significant change — GA4’s predecessor, Universal Analytics, defaulted to last-click for over a decade. Existing rule-based models (last-click, first-click, linear, time-decay, position-based) are still available in GA4’s Advertising → Attribution settings, but data-driven is what most reports show out of the box.

DDA in GA4 covers Google ads, organic search, paid search, email, referral, social, display, and direct — all channels are eligible to receive credit. There’s no minimum conversion threshold to enable it, but the underlying machine-learning model needs sufficient data to produce stable weights. For low-volume properties, GA4 falls back to using cross-channel last-click while DDA learns.

How Data-Driven Attribution Works

DDA uses machine learning to compare paths that converted with paths that didn’t, then attributes incremental credit based on which touchpoints made conversion more likely. The simplified mechanism:

Path enumeration. GA4 collects every conversion path from the past 28–90 days — sequences of channels and ads each user touched.
Counterfactual modelling. The algorithm estimates what would have happened if a specific touchpoint were absent. If removing “email” from a path drops the conversion probability by 30%, email earns substantial credit.
Credit distribution. Credit is assigned proportional to each touchpoint’s measured lift across thousands of similar paths.
Continuous retraining. The model retrains as new data arrives, so credit weights drift as channels’ real impact changes.

The result is non-obvious. A channel with high path frequency but low marginal lift (e.g., direct traffic from existing customers) gets less credit than its raw count suggests. A channel that rarely shows up but consistently lifts conversion when it does (e.g., a niche referral) gets disproportionately more.

Conversion Path and Attribution Window in GA4

Two GA4 settings control what DDA sees:

Customer journey path diagram showing four touchpoints display ad organic search email direct with credit assignment overlay for first-click last-click linear and data-driven attribution models — The same four-touchpoint journey, scored by four different models. Data-driven sits closest to “fair” because it’s the only model trained on real lift data.

Conversion path. GA4 defines a path as the chronological sequence of channel/campaign touchpoints that happened before a key event. Paths are visible in Advertising → Conversion paths, where you can see Path Length distribution and how often each channel appears at early/middle/late positions.
Attribution window (lookback). The maximum lookback period for paid-channel touchpoints. GA4 supports 30, 60, or 90 days for acquisition events and a fixed 30 days for engagement/conversion events. Outside the window, touchpoints are ignored. Configure in Admin → Attribution Settings → Lookback windows.

The attribution window is one of the most under-used GA4 settings. The default 30 days is fine for retail, but B2B sales cycles routinely exceed that, and a 30-day window will systematically miss the display impressions and content visits that started the journey. Set it to 90 days if your sales cycle is long.

Choosing the Right Attribution Model

For most accounts the choice is between data-driven (GA4 default) and a rule-based fallback. Decision matrix:

Use data-driven when you have ≥600 conversions per month per conversion type. This is the threshold where DDA’s machine learning produces stable, trustworthy weights.
Use last-click for short-cycle direct-response campaigns (impulse e-commerce, mobile-app installs) where the conversion typically happens in a single session and upper-funnel signals are weak.
Use position-based (40/20/40) when you specifically want to balance brand-discovery and closing channels in a mid-funnel B2C account.
Use time-decay when the consideration cycle is short (1–2 weeks) and recency genuinely reflects intent.
Use linear for reporting baselines and sanity checks — never as your primary model, because it ignores all signal about touchpoint importance.
Use first-click only as a complementary view to highlight which channels open journeys; never as a sole optimization model.

You don’t have to commit to one. GA4 lets you compare reports across models in Advertising → Model comparison, which is invaluable for stakeholders who want to see the same revenue split through last-click and DDA side by side.

Attribution Limitations You Need to Know

Every attribution model — including data-driven — operates on incomplete data. The known blind spots:

Cookie loss. Safari’s ITP and Firefox’s ETP cap third-party cookie life at 7 days; iOS users on Safari lose touchpoint stitching well before any 30-day window expires.
Dark traffic. Direct visits from messaging apps, podcast mentions, and offline word-of-mouth land in GA4 as (direct) / (none), even though they were caused by an upstream channel.
Multi-device journeys. A user researching on mobile and converting on desktop appears as two separate users unless cross-device tracking via Google Signals or User-ID is configured.
Ad blockers and privacy regulation. GDPR consent banners legitimately strip a 5–25% slice of pageview events, depending on geography and consent rate.
Window cut-offs. Touchpoints outside the lookback window simply don’t exist for the model, even if they were genuinely influential.

Treat attribution outputs as directional, not exact. The right framing: “data-driven attribution gives a better channel-level decision than last-click — but neither produces single-source-of-truth revenue accounting.” Combine attribution insights with media-mix modelling and incrementality tests for the highest-stakes budget decisions.

Frequently Asked Questions

What is attribution in marketing?

Attribution is the rule or algorithm an analytics tool uses to assign credit for a conversion across all the touchpoints (ads, search visits, emails, direct visits) that influenced the user before they converted. It determines which channels appear profitable in your reports.

What attribution model does GA4 use by default?

GA4 uses data-driven attribution (DDA) as the default for all conversion reports since late 2023. This replaces last-click as GA4’s default and applies to every channel — Google Ads, organic, paid search, email, social, referral, display, and direct.

What is the difference between first-touch and last-touch attribution?

First-touch (first-click) gives 100% of conversion credit to the first marketing touchpoint a user encountered. Last-touch (last-click) gives 100% to the final touchpoint before conversion. Both are extreme rule-based models — first-touch favours discovery channels, last-touch favours closing channels, and neither reflects the multi-touch reality of most real customer journeys.

How do I change the attribution model in GA4?

Go to Admin → Property → Attribution Settings. You can change the reporting attribution model (data-driven, last-click, etc.) and the lookback window (30/60/90 days for acquisition). Changes apply to all subsequent reports — historical data is recomputed retroactively for the new model in most reports.

What is a good attribution window for GA4?

For typical B2C e-commerce, the default 30 days is sufficient. For B2B and high-consideration purchases (SaaS, real estate, finance), use 90 days. Window length should approximate your real sales cycle — too short and you’ll miss upper-funnel touches; too long adds noise from unrelated channel exposure.

Does data-driven attribution work for low-conversion accounts?

DDA needs sufficient conversion volume to train reliably — Google’s guidance suggests around 300 conversions per conversion type over 30 days. Below that threshold, GA4 still labels reports as “data-driven” but the model falls back to weighted-cross-channel logic similar to position-based until enough data accumulates.

How is data-driven attribution different from last-click in GA4?

Last-click assigns 100% credit to the final non-direct channel before conversion. Data-driven uses machine learning across thousands of paths to credit each touchpoint based on its measured incremental lift on conversion probability. DDA typically credits upper-funnel channels (display, organic) 20–40% more than last-click does, and credits the final paid click correspondingly less.

Attribution model — the specific rule (last-click, linear, DDA, etc.) used to distribute conversion credit.
Conversion — the key event being attributed (purchase, signup, lead).
First-touch — model giving 100% credit to the first touchpoint.
Last-touch — model giving 100% credit to the final touchpoint before conversion.
Source — the origin of a touchpoint (google, facebook, newsletter).
Medium — the channel category of a touchpoint (organic, cpc, email).
UTM — URL parameters that label inbound traffic for accurate attribution.
Campaign — the marketing initiative grouping related touchpoints.
CPA — cost per acquisition, the headline efficiency metric attribution feeds into.

Bottom Line

Attribution decides which channels appear to drive revenue and which look wasteful — so the model is a strategic choice, not a technical setting. GA4 defaults to data-driven attribution for good reason: it’s the only model trained on your actual paths rather than a fixed assumption. Use DDA when conversion volume supports it, set a lookback window that matches your real sales cycle, and remember that no attribution model produces single-source-of-truth accounting — combine it with incrementality tests for budget decisions that matter.