Attribution models: how GA4 credits conversions across channels

Part 1 opened this whole series with a real question neither Google Ads' own tracking nor Meta's own Pixel could fully answer: when a journey spans multiple channels before converting, which one actually deserves credit? This part is the direct answer — attribution modeling, the actual mechanism GA4 uses to split that credit.

A real, multi-touchpoint journey

Day 1: discovers Bright Leaf Coffee via an organic search result
        (the SEO series' own subject) for "why does coffee go stale"
Day 4: sees a Meta Ads retargeting ad (Meta Ads series, part 6), clicks,
        browses, doesn't purchase
Day 9: clicks a Google Ads campaign ad after searching "coffee subscription
        gift," completes the purchase

Three real touchpoints, one eventual conversion — attribution modeling is specifically about how the value of that purchase gets distributed across all three, not just assigned entirely to whichever happened last.

Last-click attribution: the simplest, and the most misleading alone

100% credit → Google Ads (the final click before purchase)
0% credit → Organic Search and Meta Ads, despite both playing a real
             role in the journey

Last-click is the easiest model to understand, and it's what many simpler reporting setups default to — but it systematically undervalues every channel that plays an earlier, real role in a journey, exactly the organic content and retargeting touchpoints from this example. A business optimizing purely on last-click data risks cutting real, contributing investment (the organic content that started this journey, the retargeting that kept it warm) in favor of whatever channel happens to land the final click most often.

Data-driven attribution: GA4's own current default

Credit distribution (data-driven, illustrative): Organic Search 35%,
Meta Ads 25%, Google Ads 40%

Rather than a fixed rule (always credit the first touch, always credit the last), data-driven attribution uses GA4's own machine learning, trained on real conversion patterns across the property's actual data, to estimate each touchpoint's genuine, differential contribution to the eventual conversion — a touchpoint's credit share reflects how much it actually appears to have mattered, based on real, observed patterns, not a fixed positional rule.

Why data-driven is the sensible current default

Google made data-driven attribution GA4's standard model for good reason: it's the model that most closely reflects the real, complex nature of the multi-touchpoint journey shown above, rather than an oversimplified single-touch rule that was always a known, if convenient, distortion of reality. Older single-touch models (first-click, last-click, linear) still exist as options for specific comparison purposes, but they're not the model to default to for real, ongoing decision-making anymore.

What this means for evaluating the Google Ads series' own campaigns directly

The Google Ads series covered ROAS and CPA largely from within Google Ads' own reporting, which — by default, unless explicitly configured otherwise — tends toward its own attribution logic for its own campaigns. Comparing that against GA4's data-driven, cross-channel view of the same conversions often reveals a real, meaningful difference: a channel that looks highly efficient in its own isolated platform reporting can look less impressive once GA4's broader model accounts for the other touchpoints that also contributed to the same conversions.

A practical takeaway: don't defund a channel based on last-click data alone

Last-click view: "Organic Search drove 0% of conversions directly —
consider cutting content investment"
Data-driven view: "Organic Search contributed to 35% of conversion value
as a real, meaningful earlier touchpoint — cutting it would likely hurt
Google Ads' own last-click numbers too, since fewer people would ever
enter the journey in the first place"

This is a real, concrete example of why attribution model choice isn't an abstract technical detail — it directly shapes real budget and strategy decisions, and the wrong model can lead to defunding a channel that was quietly doing real, contributing work the simpler model just couldn't see.

Next: connecting GA4 directly to Google Ads and Meta Ads — the actual integration that makes cross-platform attribution and audience sharing (part 11) possible at all.