Fractional Attribution: Stop Giving All the Credit to the Last Click

Fractional attribution is a method of distributing conversion credit across multiple touchpoints in a customer experience, rather than assigning it all to one. Where last-click attribution says the final touchpoint won the sale, fractional models ask a more honest question: which channels actually contributed, and by how much?

It sounds like a technical upgrade. In practice, it changes which channels get budget, which teams get credit, and how you read the health of your marketing programme. That makes it one of the more consequential decisions you can make in your measurement setup.

I have spent a long time managing attribution across large accounts, and the honest truth is that most teams never question the default. They inherit last-click from whoever set up the account years ago, optimise hard against it, and then wonder why their paid search budget keeps growing while everything else shrinks. Attribution is not just a measurement question. It is a budget question dressed up in analytics language.

Why Last-Click Attribution Survives Despite Being Widely Criticised

Last-click attribution has been criticised for decades. It over-credits the final touchpoint, typically branded search or direct, and ignores everything that brought the customer to that point. Everyone in the industry knows this. And yet it remains the default in more accounts than you would expect.

The reason is not ignorance. It is inertia combined with a genuine problem: last-click is simple to explain, simple to optimise against, and produces clean, unambiguous numbers. When you are managing a large team or reporting to a board, clean numbers feel safer than probabilistic ones. I have been in those rooms. The CFO does not want to hear that your attribution model “assigns partial credit based on position decay.” They want to know what drove the sale.

Last-click also flatters the channels that sit at the bottom of the funnel. Paid search on branded terms almost always looks exceptional under last-click, because it captures intent that was built elsewhere. When I was at iProspect, we would regularly see accounts where branded search appeared to be the highest-performing channel by a significant margin. Strip out last-click and apply a fractional model, and the picture shifted considerably. Display, paid social, and email were doing more work than the numbers suggested.

This matters because budget follows attribution. If branded search looks like it is delivering five times the return of your prospecting campaigns, you will keep cutting prospecting and increasing branded. Eventually you hollow out the top of the funnel and wonder why branded search volume starts declining too.

How Fractional Attribution Models Actually Work

There are several approaches to fractional attribution, and they differ in how they decide what weight to give each touchpoint. Understanding the mechanics matters because the model you choose encodes assumptions about customer behaviour, and those assumptions have real consequences.

Linear attribution distributes credit equally across all touchpoints. If a customer interacted with a display ad, a paid social post, an email, and a branded search ad before converting, each gets 25% of the credit. It is democratic but arguably naive. Not every touchpoint does the same work.

Time-decay attribution gives more credit to touchpoints that occurred closer to the conversion. The logic is that recent interactions are more influential. This is more defensible than linear for short sales cycles, but it still penalises upper-funnel activity that may have been genuinely important.

Position-based attribution, sometimes called U-shaped, gives the most credit to the first and last touchpoints, and distributes the remainder across the middle. A common split is 40% to first touch, 40% to last touch, and 20% shared across everything in between. This acknowledges both acquisition and conversion moments, which makes intuitive sense for many businesses.

Data-driven attribution is the most sophisticated approach available in most platforms. Rather than applying a fixed rule, it uses machine learning to analyse which combinations of touchpoints are statistically associated with conversion, and assigns credit accordingly. Google’s data-driven model in GA4 and Google Ads works this way. It sounds compelling, and in many cases it is, but it requires meaningful conversion volume to produce reliable outputs, and the model itself is a black box. You cannot fully audit the logic.

If you want a solid grounding in how GA4 handles event tracking and attribution across channels, the Moz guide to a flawless GA4 setup is worth reading before you start adjusting attribution settings. Getting the tracking right is a prerequisite for any attribution model to produce useful outputs.

What Changes When You Switch to a Fractional Model

The practical impact of switching attribution models is often more significant than teams expect. Channel performance numbers shift. Some channels look better. Some look worse. And the conversations that follow are rarely comfortable.

I have seen this play out directly. When we moved a large retail account from last-click to a position-based model, the paid social team went from looking like they were barely breaking even to looking like a genuinely strong contributor. The paid search team, who had been the heroes of every monthly report, saw their attributed revenue drop noticeably. Neither number was the truth. Both were approximations. But the fractional model was a more honest approximation.

What changes in practice:

• Upper-funnel channels like display and paid social typically see their attributed value increase
• Branded search and direct traffic typically see their attributed value decrease
• Email often gets more credit for its role in nurturing journeys
• Budget allocation decisions become harder to make with false confidence, which is actually a good thing
• Channel teams start having conversations about contribution rather than credit

That last point is underrated. Attribution models shape organisational behaviour as much as they shape budget decisions. When every team is optimising against a last-click model, everyone fights for the bottom of the funnel. Fractional models create at least some incentive to think about the full experience.

This connects to a broader point about what marketing metrics are actually for. Mailchimp’s overview of marketing metrics makes the point well: metrics should help you make better decisions, not just produce impressive-looking numbers. Attribution is no different.

The Limits of Fractional Attribution That Most Articles Ignore

Fractional attribution is better than last-click in most situations. That does not make it accurate. There are structural limitations that matter, and you should understand them before treating any attribution output as ground truth.

It only measures what it can track. Fractional models work with the touchpoints that are visible to your tracking setup. If a customer saw a YouTube ad, heard about your brand from a colleague, read a review on a third-party site, and then converted through paid search, your attribution model sees one touchpoint. The other three are invisible. This is not a flaw in the model. It is a fundamental constraint of digital tracking.

Cross-device journeys are still a problem. A customer who researches on mobile and converts on desktop looks like two separate users to most attribution systems. The touchpoints do not stitch together cleanly. GA4 has improved this with user-based modelling, but it is not solved.

Correlation is not causation. Data-driven attribution identifies patterns in conversion paths, but that does not mean the touchpoints it credits actually caused the conversion. A customer who was already highly likely to buy may have clicked a display ad along the way. The model credits the display ad. But the display ad may have been irrelevant to the decision.

Cookie deprecation and consent changes are reducing data quality. As third-party cookies disappear and consent rates vary by market, the underlying data that attribution models rely on is becoming patchier. A model that looked reliable two years ago may now be working with significantly less signal. This is not a reason to abandon attribution, but it is a reason to hold the outputs more lightly.

When I judged the Effie Awards, one of the things that struck me about the strongest entries was how carefully the teams had separated what they could measure from what they could infer. The weaker entries treated their attribution data as if it were complete. It never is.

If you want to understand how to think about marketing analytics more broadly, including what the numbers can and cannot tell you, the Marketing Analytics hub at The Marketing Juice covers the full landscape from GA4 setup to measurement strategy.

Data-Driven Attribution: More Sophisticated, Not Necessarily More Accurate

Data-driven attribution has become the default recommendation in most platform documentation, and it is genuinely better than rule-based models in many cases. But it carries its own risks, particularly for teams that adopt it without understanding what it is doing.

The core issue is opacity. When Google’s algorithm decides that your YouTube campaign deserves 18% of conversion credit, you cannot audit that calculation. You can accept it or reject it, but you cannot interrogate it the way you can interrogate a position-based model where the rules are explicit.

This matters when the outputs produce counterintuitive results. I have seen data-driven models assign very low credit to email campaigns that, by any reasonable commercial assessment, were doing significant work in the conversion experience. The model was optimising against patterns in the data. The patterns did not capture the full picture. The team cut email budget based on the attribution output, and saw conversion rates drop shortly after.

Data-driven attribution also requires volume. Google recommends a minimum of 400 conversions in a 30-day window for the model to produce reliable outputs. Below that threshold, the algorithm does not have enough data to identify meaningful patterns, and the outputs can be noisy. For smaller accounts or niche products with low conversion volumes, data-driven attribution may be less reliable than a well-chosen rule-based model.

For teams running GA4, Moz’s guide to GA4 custom event tracking is useful context for understanding how event data feeds into attribution models. The quality of your attribution is only as good as the quality of your event tracking.

How to Apply Fractional Attribution Without Getting Lost in the Model

The practical question is not which attribution model is theoretically correct. It is which model produces the most useful signal for your specific business, given your data quality, conversion volume, and channel mix.

Here is how I would approach it:

Start with your business model. A business with a long consideration cycle and multiple research touchpoints before purchase will benefit more from a fractional model than a business where customers typically convert in a single session. If your average customer experience is three touchpoints over two weeks, linear or position-based attribution will tell you more than last-click. If most of your conversions happen in a single session, the difference between models is smaller.

Run models in parallel before switching. Most platforms allow you to compare attribution models without changing your active reporting. Use the comparison view in GA4 or Google Ads to see how your channel performance changes across models before you make any budget decisions based on the new view. The delta between last-click and your chosen fractional model tells you a lot about where your current budget allocation may be skewed.

Use incrementality testing to validate. Attribution models tell you correlation. Incrementality tests tell you causation. If you want to know whether your display campaign is actually driving conversions or just appearing in the paths of people who were going to convert anyway, run a geo-based holdout test or a conversion lift study. This is harder to set up than switching an attribution model, but it is more reliable. I have seen incrementality tests produce results that were genuinely surprising, in both directions.

Do not optimise blindly against the model. Attribution outputs should inform your thinking, not replace it. If a fractional model tells you that your email programme is underperforming, that is a prompt to investigate, not an instruction to cut the budget. Look at the data, talk to the team, consider what the model cannot see, and then make a decision.

Building a coherent KPI framework around your attribution data is part of making this work in practice. Semrush’s guide to KPI reporting covers how to structure performance reporting in a way that connects channel metrics to business outcomes, which is the right context for attribution data to sit in.

Early in my career, when I was running paid search at lastminute.com, we were optimising campaigns almost entirely on last-click revenue. The numbers looked clean and the campaigns looked profitable. It was only when we started looking at the full path data that we realised how much work certain upper-funnel touchpoints were doing. We had been systematically underfunding them because the attribution model made them look weak. Switching the lens changed the budget conversation entirely.

Attribution in the Context of a Broader Measurement Strategy

Fractional attribution is one tool in a measurement stack, not the whole stack. Teams that treat it as the answer to their measurement problems tend to over-invest in model complexity and under-invest in the fundamentals.

The fundamentals are: clean tracking, consistent conversion definitions, a shared understanding of what the data can and cannot measure, and a culture that treats attribution outputs as inputs to a decision rather than verdicts. HubSpot’s argument for marketing analytics over web analytics gets at this well: the goal is to connect marketing activity to business outcomes, and that requires more than a single attribution model.

Marketing mix modelling sits alongside attribution as a complementary approach. Where attribution works at the individual touchpoint level, MMM works at the aggregate level, using statistical modelling to estimate the contribution of different channels to overall business outcomes. Neither approach is complete on its own. The most sophisticated measurement setups use both, and triangulate between them.

The question worth asking is not “which attribution model is right?” but “what decisions am I trying to make, and what data do I need to make them well?” Attribution is a means to an end. The end is better budget allocation, more accurate channel evaluation, and in the end better commercial outcomes.

If you are building out your analytics capability more broadly, the Marketing Analytics section of The Marketing Juice covers everything from GA4 configuration to measurement strategy and how to connect analytics to commercial decision-making.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions