GA4 can show that organic search appeared before a purchase and, where events can be connected, that a specific article was viewed before a key event. It also assigns credit under an attribution model. None of this tells you how many extra sales the article created. That is why we keep three questions apart: which channels appeared, which article interactions you can trace, and whether the article changed the sales outcome.
One journey, several different answers
Take one illustrative journey, not a customer result. Someone arrives from Google organic search on a how-to article and leaves without buying. Later, the same person clicks a Google Ads search ad and completes a purchase key event. It is a single journey, yet each GA4 tool sees a different part of it.
Attribution paths shows the channels in converting journeys
Open Advertising, then select Key event attribution paths under the Key events dropdown. The report has an eligibility limit, set out in Google's Attribution paths guide: only users who triggered a key event appear. Someone who reads several articles but never completes a key event has no path here, so the report shows how converting journeys unfolded, not how many people your blog reached.
At channel level, our example becomes Organic Search → Paid Search, provided both touchpoints are connected in the recorded journey. Those labels are categories of the default channel group. Source identifies the search engine or website, medium describes categories such as organic, CPC or referral, and campaign names the marketing campaign. Every one of these dimensions describes where the traffic came from. None of them identifies the how-to article, so calling the organic touchpoint a blog assist would require additional page-level evidence.
The same purchase can receive different credit
The attribution model decides how GA4 shares credit for that purchase. Google's attribution model documentation names data-driven attribution as the default: it uses machine learning to assign fractional credit to touchpoints. The last-click models apply a different rule. Here is what each gives the organic visit in our example, which was followed by a Google Ads click and the purchase.
| Model | Credit rule | Organic touchpoint in the example |
|---|---|---|
| Data-driven attribution | Uses machine learning to allocate fractional credit to touchpoints | May receive a modeled share, including zero. No percentage is assumed here |
| Paid and organic last click | Gives full credit to the last paid or organic channel | Receives no credit because Paid Search came last |
| Google paid channels last click | Gives full credit to the last Google Ads channel clicked before conversion | Receives no credit because the later Google Ads click gets full credit |
You choose the model in Attribution settings, and a change applies retroactively, so credit for purchases that have already happened can shift because of the setting alone. Google's Attribution paths guide also notes that first-click, linear, time decay and position-based models were deprecated as of November 2023, so you cannot switch to first-click attribution to give the first article visit full credit. Under the models that remain, we would not read a last-click result of zero as evidence that the earlier article had no role, or a data-driven share as an exact measure of the article's sales impact.
Use Path exploration to investigate the article itself
To find out whether a particular article was viewed before a key event, go to Explore and open Path exploration. As Google's Path exploration guide describes, a path can start or end at a page or an event, so you can anchor the analysis on the article instead of the channel that brought the visitor.
- Set the how-to article page as the starting point to see the interactions that follow it.
- Or set the purchase key event as the ending point and look back through the earlier steps for article interactions.
- Read each column as one step in the sequence. Its nodes count the users or events at that step.
In our example, starting from the how-to page brings the article itself into the analysis. If an article interaction appears in the displayed path before the purchase, you have page-level evidence of the sequence, which an Organic Search touchpoint alone cannot provide. The node counts describe the interactions shown. They do not calculate the revenue that page created.
BigQuery can trace events, but identifiers still set the limits
To query the sequence itself, use the BigQuery export, which provides raw, unsampled event-level data, as Google's BigQuery export documentation describes. In our illustrative journey, an event query can trace the article page view and the later purchase when both carry the same identifier. That places the page interaction before the purchase in the data itself, instead of inferring it from a channel label.
Everything depends on that identifier. The third-party BQ4GA user-fields guide describes user_pseudo_id as pseudonymous and user_id as an ID assigned to authenticated users, present only when manually configured. Before building an analysis around user_id, confirm that your implementation actually supplies it. Whichever identifier you use, the trace covers only the events connected to it.
Denied consent can leave events without the identifier the trace needs. According to the third-party GA4BigQuery consent-mode guide, under advanced consent mode the cookieless pings sent when consent is denied exclude identifiers such as user_pseudo_id, so those events cannot be connected through that field to the same visitor's article views. The export fields privacy_info.analytics_storage and privacy_info.ads_storage indicate consent status, which lets you segment the data when you examine these gaps.
An article-linked order still is not an incremental sale
Attribution divides credit for a key event that has already happened. A larger organic share under another model is the same purchase divided differently. Tracing the article page view gives stronger evidence that someone interacted with the article before buying, but it leaves a different question open: would this person have bought without reading the article? Neither the channel path nor the event sequence supplies that comparison.
In our MateiTravel work, we created service articles with promo codes built into the content as bonuses, and the first paid order came directly from a blog post. An order that comes straight from a blog post is an article-specific signal, more concrete than an organic channel appearing somewhere in a journey. Even so, it does not establish whether that order would have happened without the article.
We keep each conclusion as specific as the evidence behind it. Channel paths support the statement that Organic Search appeared before purchases. A traced page interaction supports the statement that a named article preceded a purchase. Attributed credit should always be reported with the model that assigned it. A claim about extra sales needs evidence that compares what happens with the article against what would happen without it.
Example of using the shortcode function through Blogent
