Content Attribution: Models, GA4 Limits and a Per-Page Example

Content attribution is the process of giving individual pieces of content, such as blog posts, guides, case studies and comparison pages, credit for the conversions and revenue they helped produce. An attribution model splits each conversion across the pages a buyer touched, so you can see which content introduces buyers, which content keeps them interested and which content closes.

That sounds like what your analytics already does. It is not. Most tools, GA4 included, credit channels and campaigns, not articles. This guide covers the models, what GA4 can and cannot tell you about a single page, a worked example that puts a dollar figure on four pages under four models, a decision rule for each page, and the places where content attribution quietly breaks.

What Content Attribution Is (and What It Is Not)

A buyer rarely converts on the first page they read. They find a guide through Google, come back a week later from a newsletter, read a case study, then request a demo from the pricing page. Content attribution asks: of that demo, how much belongs to the guide, the case study and the pricing page?

Three terms get mixed up, and the difference matters for what you can measure:

• Channel attribution credits sources like Organic Search, Email or Paid Social. This is what GA4 attribution reports do.
• Campaign attribution credits UTM-tagged campaigns or ad groups.
• Content attribution credits URLs or assets: this article, that case study, the pricing page. Few tools do this out of the box, so most teams have to build it.

The same model logic applies to all three. What changes is the unit you record for each touch. If you only record channels, no model will ever tell you which article earned the credit.

Why Content Teams Need It

Attribution is used to make spending decisions. For content, that means four practical jobs:

• Proving return. A content program that can show conversions and revenue per article defends its budget. One that shows pageviews does not. If you need the ROI formula itself, see content marketing ROI.
• Deciding what to write next. If comparison pages close and guides only introduce, you know which gap to fill.
• Deciding what to fix or cut. Pages that get credit under no model after months of traffic are candidates for a refresh, a merge or removal.
• Mapping the funnel. Credit by page shows how buyers move from awareness content to decision content.

The reason content needs this more than paid media: last-click reporting systematically undercounts it. Content tends to start journeys, and the last click before a conversion is usually something else, like a branded search, an email or a direct visit to pricing. Read only last click and your best introducers look like dead weight.

Content Attribution Models

An attribution model is the rule that splits credit. Single-touch models give one touchpoint 100%. Multi-touch models spread it. Here is each common model, with what it means when the touchpoints are pages:

If you want a deeper look at one model, we have separate guides on first touch attribution and data-driven attribution.

Where the guides disagree: what “U-shaped” means

Many articles define the U-shaped model as 40% to the first touch and 40% to the last. Others define it as 40% to the first touch and 40% to the touch that created the lead. Both are in use. Classic position-based attribution puts the second 40% on the last interaction. HubSpot’s attribution report definitions put it on the lead conversion interaction, then spread 20% across everything else.

For content this is not a footnote. Under the lead-conversion version, a gated template or webinar page gets a big share. Under the last-touch version, the pricing page does. Before you compare reports from two tools, check which definition each one uses.

What GA4 Can and Cannot Do for Content

According to Google’s attribution documentation, GA4 now offers three reporting models: data-driven, paid and organic last click, and Google paid channels last click. First click, linear, time decay and position-based were discontinued in November 2023. All GA4 models also exclude direct visits from credit unless the whole path was direct.

Those models credit channels and campaigns. Here is what you can still get at the page level:

• Landing page plus key events. The Landing page dimension shows the page that started each session. Pair it with key events and you get, roughly, last-session credit by entry page. A guide that brings someone in and leads straight to a sign-up gets counted. A guide read in an earlier session does not.
• User-scoped first source. First user source, medium and campaign record how a user originally arrived. That is user attribution at the channel level, not the page level.
• Path exploration. In Explore, you can follow page sequences before an event. Useful for spotting patterns, not for assigning credit.
• BigQuery export. With raw events you can rebuild any page-level model yourself, including the retired ones.

Channel Attribution Is Not Content Attribution: How to Get Page-Level Credit

The cheapest reliable method does not need a new platform. It records landing pages at the moment someone becomes a lead, then lets your CRM or a spreadsheet apply the model.

1. Define the conversions and give each one a value. Decide what counts: demo request, trial, purchase, qualified lead. Assign each a dollar value so credit becomes money. Our guide on how to calculate conversion value walks through the math.
2. Store the first landing page. On a visitor’s first visit, save the landing URL and referrer in first-party storage, then pass it into hidden fields on every form.
3. Store the converting landing page. Also pass the landing page of the session in which the form was submitted. Now each lead carries a first page and a last page.
4. Log content touches in between, if you can. Email clicks to articles, webinar attendance and case study downloads are recorded by most marketing automation tools against the contact. These become the middle touches.
5. Add a self-reported field. “How did you first hear about us?” with named options (search, a specific article, an AI assistant, a podcast, a colleague) catches what no tracking sees.
6. Apply two models and wait one sales cycle. Start with first touch and last touch, add linear or U-shaped once most leads have three or more touches. Judge pages only after a full sales cycle of data, or long-cycle content will look worse than it is.

If you track organic landing pages already, our SEO conversion tracking guide covers the setup on the search side.

Do not put UTM tags on internal links

Some content attribution guides tell you to add UTM tags to every internal link from a blog post to a landing page so the click is “tracked”. Do not do this. Google’s documentation on campaigns and traffic sources says that new campaign values arriving in the middle of a session are associated with the events collected from that point and are used for event-based attribution. In practice, a sign-up after an internally tagged click can be credited to your own “blog” campaign instead of the organic search or email visit that actually brought the person in. You lose the real source and gain a fake one.

To know which article sent someone to the form, use the page referrer, a hidden field holding the previous page, or a custom event parameter on CTA clicks. None of these overwrite the traffic source.

A Worked Content Attribution Example, Start to Finish

This example is illustrative. The journeys and the lead value are made up to show the arithmetic. A B2B software company has four pages that show up before demo requests:

• A: a how-to guide that ranks in Google
• B: a comparison article
• C: a customer case study
• P: the pricing page

Five demo requests came in this month. Each journey lists the landing page of every session, in order:

For the U-shaped model we use the first-and-last version: 40% first, 40% last, 20% shared by the middle. A two-touch path splits 50/50 and a one-touch path gives 100%. Credit per page, in demos:

How the linear column works for page A: it is half of demo 1 (two touches), a third of demo 2 (three touches) and a third of demo 5 (three touches).

Now assume each demo request is worth $400 (illustrative lead value). Multiply credit by value:

The total never changes. Only the split does. Under last touch, both articles look worthless and the pricing page looks like it produces 80% of demand. Under first touch, the articles introduced four of the five buyers. Linear and U-shaped land in between. Nobody’s behavior changed: only the rule did.

Note demos 1 and 5. Their last session was a direct visit to pricing. At the page level, P still gets the last touch. At the channel level, GA4 would skip direct and credit the previous channel, so a GA4 channel report and your page-level report will not match. That is expected, not an error.

What to Do With Each Page: A Decision Rule

Once you have first-touch and last-touch credit per page, every page falls into one of four patterns. Each has a different action:

Treat this as a sorting tool, not a verdict. A page with zero credit may be helping buyers who convert through a channel you cannot see, which is the next problem.

How to Choose a Content Attribution Model

Four questions settle most cases:

• How long is the buying cycle? Days: last touch is usually fine for closing content, with first touch beside it. Weeks or months: compare at least two models and look at the gap.
• How many conversions per month? Algorithmic models need volume to stay stable. With low volume, a rule you can explain is more useful than numbers that swing month to month.
• Do you have lead and deal stages in a CRM? W-shaped and full path need clean timestamps for lead creation and deal creation. Without them, stay with first, last and linear.
• What will you change based on the answer? If the decision is “which articles to update this quarter”, first touch plus last touch is enough. Build something heavier only if it changes a decision.

A practical rule: if a page performs well under every model, invest in it. If it only looks good under one, question the model before you question the page.

Where Content Attribution Breaks

Every model can only split credit across touches it saw. When a touch is missing, the credit does not go to the right page. It goes to whatever page was visible next.

• AI assistants. A buyer asks ChatGPT or Perplexity to compare tools, your guide is cited, and they come back days later by typing your URL. Some assistants pass a referrer or a UTM tag on clicks. Many sessions arrive with neither and look like direct traffic. See AI conversion tracking for what can and cannot be measured.
• Private sharing. An article pasted into Slack, a group chat or an email arrives without a referrer. That is dark social, and it lands in direct.
• Cross-device journeys. Reading a guide on a phone and requesting a demo on a laptop looks like two people unless they log in or fill a form on both.
• Privacy controls and short storage. Cookie blocking, consent choices and browser limits on stored data cut journeys short, so the first touch you record may not be the real first touch.
• Offline and clickless influence. Podcasts, events, sales conversations and word of mouth shape decisions without any page touch.
• Lookback windows. If your model only looks back 30 days and your cycle is 90, early content falls outside the window. Our guide to the attribution window explains the settings.
• Outside events. Seasonality, a competitor’s outage or a news cycle move conversions that no page caused.

Debugging a page that “gets no credit”

1. Check that forms actually capture the first landing page. Submit a test lead from a fresh browser and read the hidden fields.
2. Check the lookback or cookie lifetime against your sales cycle.
3. Look for internal UTM tags that overwrite the original source.
4. Compare with the self-reported field. If buyers name the article but tracking does not, the gap is in tracking, not the page.
5. Check whether direct traffic to deep pages is rising. People rarely bookmark a long guide; they arrive from somewhere you cannot see.

Check Attribution Against Reality

Attribution describes the order of touches. It does not prove that a page caused a conversion. Critics of attribution, including some well-known growth leaders, argue that budgets should not follow attribution reports at all. The fair version of that argument: use attribution to compare pages you can observe, and use other methods to check cause.

• Self-reported source. The “how did you hear about us” answer is biased by memory, but it is the only signal that catches a page or channel your tracking missed.
• Incrementality tests. Hold something back and measure the difference. For content, that can mean pausing distribution of a content series to one email segment, or comparing similar pages with and without a new CTA.
• Marketing mix modeling. Correlates total spend and output against total results over time. It needs long history and sees no individual pages, but it does not depend on tracking.
• Demand signals. Branded search and direct traffic that rise after a content push, with nothing else changing, are evidence attribution cannot produce.

When the model and the self-reported answer agree, act. When they disagree, the self-reported answer usually points at the origin and the model shows the path.

Content Attribution Tools

Tools split into three groups, and most teams need one from the first group before anything else:

• CRM and marketing automation (HubSpot, Salesforce, Marketo). These attach page views, email clicks and form submissions to contacts, so they can credit content to leads and deals. HubSpot’s reports offer linear, first, last, U-shaped, W-shaped, time decay, full path, J-shaped and inverse J-shaped models and track blog posts and landing pages as assets.
• Web analytics (GA4 and similar). Free and already installed, strong for channel credit and landing-page conversions, limited for multi-session page credit without BigQuery.
• Dedicated attribution platforms. Products built for paid media and ecommerce, plus call tracking tools that connect phone calls to sessions. Many focus on ad spend more than on organic content, so check that they record page-level touches before you buy.

If your content earns its traffic from Google and AI assistants, SEOConversion reports conversions and their value by landing page for organic search and AI assistant referrals, which covers the “which article converts” question without a CRM project.

Other Meanings of Content Attribution

Outside marketing, “content attribution” means crediting the source of content you publish, and several common questions use that sense:

• What the word means. Attribution is saying where something came from or who caused it: the author of a quote, the source of a fact, the reason behind a result.
• Journalism. Attribution names the source of information so readers can judge it: a named person, a document or an organization. Unattributed claims are weaker by default.
• Media and reuse. When you reuse someone’s image, video or text, attribution credits them. Creative Commons recommends the TASL pattern: title, author, source and license.
• Communication and psychology. Attribution theory studies how people explain behavior, for example blaming a situation versus a person’s character.
• User attribution. In analytics, it means identifying the source that first brought a given user.

The marketing meaning borrows the same idea: give credit where it is due, and be honest about what you cannot trace.

FAQ