Attribution models: How to evaluate the actual value of your marketing channels

You invest in Google Ads, run meta campaigns, and engage in content marketing. The conversion happens—but which channel actually influenced the purchase decision? Attribution models promise clarity by assigning conversions to different touchpoints. But the reality is more complex: modern customer journeys span multiple devices, channels, and weeks—while classic attribution captures only a fraction of this journey. This article shows how different attribution models work, where their limitations lie, and which model is suitable for which use case.

Written by
Published on
Last modified on

Christopher Smid-Sawall
December 30, 2025
December 30, 2025

What is attribution and what makes it so challenging?

Attribution assigns conversions to different marketing touchpoints and answers the key questions:

Which channel contributed to the purchase decision?
Which touchpoints did users see before conversion?
Which touchpoints deliver the most conversions and which generate the most revenue?

A touchpoint refers to any point of contact where users come into contact with your marketing measures—whether it's clicking on an ad, viewing a social media post, or visiting your website.

The challenge: Purchasing decisions rarely result from a single contact. A person sees your display ad, later researches on their smartphone, compares offers on their desktop, and finally makes a purchase after a Google search. Each of these touchpoints could have contributed to the decision—but which one was actually decisive for the purchase? Attribution models attempt to answer this question using various attribution rules.

What attribution models are there?

Single-touch models: One channel receives the entire allocation

Single-touch models focus on a single point of contact in the customer journey. This simplification makes them easy to understand and implement, but ignores all other touchpoints.

Last click attribution

The best-known and most widely used single-touch model is last-click attribution. The last click before conversion receives 100 percent of the attribution. The underlying assumption is that the final touchpoint triggered the purchase decision. A user clicks on a Google ad, lands on your website, and makes a purchase immediately—Google Ads receives full attribution.

The problem becomes apparent when you take a closer look at the customer journey: the same user may have become aware of your product weeks earlier through a social media campaign, spent several days gathering information via various channels, and finally searched specifically for your brand name. Last click attribution assigns 100 percent to the final brand search and completely ignores the original social media campaign. Awareness and consideration channels are systematically underestimated.

First click attribution

The opposite approach is first-click attribution: the first click receives 100 percent of the attribution, while all subsequent touchpoints are completely ignored. The focus is therefore on those channels that generate initial attention and bring users into contact with the brand in the first place.

A display campaign that established the first touchpoint three months ago receives the entire attribution—even if ten further marketing contacts took place after that and the actual purchase impulse originated from a completely different channel.

This model is particularly suitable for evaluating awareness measures in the upper funnel, but tends to systematically overestimate their importance for the final purchase decision.

Last Non-Direct Click Attribution

Last non-direct click attribution is a variant of the last click approach and, until recently, was the most commonly used model and the standard in Google Universal Analytics. This attribution model assigns full credit to the last click before conversion, with one important exception: direct visits are filtered out. The reasoning behind this is that direct visits often come from users who already know the brand and are specifically looking for it. Filtering is intended to make marketing-generated touchpoints more visible.

Multi-touch models: Multiple channels share the assignment

Multi-touch models distribute the conversion value across multiple touchpoints along the customer journey. These models represent more complex purchasing decision-making processes much more realistically than single-touch approaches—but they require more data and more sophisticated interpretation.

Linear attribution

The simplest multi-touch approach is linear attribution, where attribution is distributed evenly across all touchpoints. Each channel receives the same percentage share, regardless of its position or proximity to the conversion. With five touchpoints, each receives 20 percent; with ten touchpoints, each receives ten percent.

This even distribution takes all contact points fully into account. No channel is systematically favored or disadvantaged. However, the weakness is obvious: a random display impression weeks ago receives the same weighting as the final product search shortly before the purchase.

Linear attribution assumes that all touchpoints contribute equally to the purchase decision—an assumption that rarely applies in practice. The advantage is that the model makes the entire customer journey visible without favoring individual phases. Linear attribution is a good starting point for companies that want to get a complete picture of their marketing activities first.

Time decay attribution

Time decay attribution abandons equal weighting and introduces a time component. Touchpoints close to conversion are given greater weight. The time interval determines the attribution level according to an exponential function. A click two days before the purchase is rated significantly higher than a contact two weeks before.

This model is based on the assumption that the last interactions before the purchase are the ones that actually determine the purchase decision. This logic is understandable for e-commerce with short purchase decision processes, impulse purchases, and time-sensitive offers. The focus is clearly on final conversion drivers—those channels that trigger the immediate impulse to buy.

The systematic consequence: awareness channels that generated initial attention weeks before the purchase are undervalued—even if they had a significant influence on the purchase decision. Time decay works well for short decision-making processes, but underestimates the importance of early touchpoints in longer purchasing processes.

Position-based attribution

Position-based attribution combines the logic of first click and last click into a hybrid approach. The first and last touchpoints each receive 40 percent of the attribution. The middle touchpoints share the remaining 20 percent evenly.

This model considers both initial awareness and final conversion proximity to be particularly important. The assumption behind this is that the first contact is what initially draws attention to the brand, the last contact triggers the actual purchase decision, and the touchpoints in between maintain interest.

Position Based is particularly suitable for long customer journeys with many touchpoints – typically in the B2B sector or for products that require explanation and involve an extensive research phase. The advantage lies in the explicit evaluation of both the awareness and conversion phases. The middle touchpoints are taken into account, but are weighted significantly less.

Incremental attribution: measuring causal advertising impact

All attribution models presented so far—whether single-touch or multi-touch—are rule-based and measure correlations. They analyze which touchpoints were present before a conversion, but cannot answer whether the advertising actually caused the conversion.

Incremental attribution takes a fundamentally different approach: controlled experiments with test and control groups are used to measure how many conversions would not have taken place without advertising. The test group sees your ads, while the control group does not. The difference in purchasing behavior shows the conversions actually caused by advertising.

This method provides the most accurate insights into advertising effectiveness, but requires greater effort and sufficient campaign volume. The key advantage is that it reveals which campaigns actually generate additional sales and which are only credited with conversions that would have happened anyway. Platforms such as Meta and Google now offer automated solutions for incremental measurement that enable test setup and evaluation.

How to choose the right model for your company

Choosing the right attribution model depends largely on the structure of your customer journey. There is no universally best model—only the one that is right for your specific context.

Analysis of your customer journey

The duration of the purchasing decision process fundamentally determines the model selection. An impulse purchase in e-commerce differs from a B2B sales process lasting several months. First, determine the average number of touchpoints before conversion. Tools such as GA4 display this information directly in the attribution area.

Next, identify the dominant channels in different funnel phases. Which channels generate initial attention? Which ones accompany the research phase? Which ones trigger the final purchase decision? This analysis shows whether your customer journey has a clear structure or is evenly distributed across all touchpoints.

Also, consider cross-device user behavior. If your target audience frequently switches between mobile and desktop devices, you can expect larger tracking gaps. This affects the reliability of complex multi-touch models.

Application scenarios for different models

Time decay for quick purchasing decisions

E-commerce with short decision-making processes benefits from time decay attribution. When users make a purchase within a few hours or days of their first contact, the last touchpoints are actually the ones that influence the purchase decision.

Impulse purchases and time-sensitive offers such as flash sales or limited editions follow this pattern. The focus is clearly on final conversion drivers—those channels that trigger the immediate impulse to buy.

Position Based for complex B2B sales

Long decision-making processes involving many stakeholders require a more differentiated model. In the B2B sector, sales processes often extend over weeks or months with multiple points of contact.

Position-based attribution considers both initial awareness and the final purchase decision to be equally important. Initial contact via LinkedIn content is what makes the company known in the first place. The final decision may then be made after a personal conversation or a demo request. Both touchpoints are decisive for the purchase—the intermediate touchpoints maintain interest.

Use model comparison in GA4

Google Analytics 4 enables the parallel application of multiple attribution models, allowing you to directly compare how conversion attribution changes between different models. This comparison shows which channels perform better or worse under different evaluation logics and provides valuable insights for strategic budget decisions.

Systematically analyze the attribution differences per channel: Which channels are gaining importance under time decay? Which ones are losing importance under last click? These differences reveal the actual role of individual channels in your customer journey. Observe the change in key event attribution over several weeks, as meaningful patterns only become visible with a sufficient data base. The prerequisite for reliable comparisons: sufficient conversion data across multiple sessions and different touchpoints.

Use these insights specifically for your budget allocation: if a channel performs consistently well across all models, this justifies higher investments. If, on the other hand, a channel only performs well under a single model, you should critically question whether this model actually accurately reflects reality. Model comparison in GA4 thus becomes a practical tool for understanding the strengths and weaknesses of different attribution approaches for your specific case and making informed marketing decisions.

Why is traditional attribution reaching its limits?

Marketing channels are constantly competing for conversion attribution, with the key question being: Which channel actually influenced the purchase decision? Attribution attempts to answer this question by analyzing touchpoints—the points of contact where users encounter marketing measures.

The fundamental problem: only measurable factors are taken into account

The fundamental problem with traditional attribution lies in its premise: it requires measurable clicks and page visits, because only what is technically traceable can be attributed. The reality is different: purchasing decisions are often made without a direct click. A person sees a display ad, remembers the brand name, and searches for it specifically days later—the original impression remains invisible in the attribution. Impressions without clicks, offline advertising such as posters or TV commercials, and brand mentions in conversations have a significant influence on purchasing decisions, but are not captured by attribution models.

Tracking restrictions

Tracking restrictions due to cookie blockers and browser restrictions are now the norm, not the exception. Users who activate tracking protection remain largely invisible in attribution.

Cross-device tracking also has its limitations, as assigning different devices to a single person requires either login data or probabilistic models—both approaches have significant gaps. A user who searches on their smartphone and makes a purchase on their desktop often appears as two different people in the attribution.

Invisible touchpoints

Offline touchpoints remain completely invisible: print ads, TV commercials, billboards, and events have a significant influence on purchasing decisions, but cannot be integrated into digital attribution. The measured customer journey shows only the digital part of a more comprehensive reality.

Organic brand searches are often difficult to attribute—when someone searches for your brand name and makes a direct purchase, it remains unclear whether this was a conscious decision or the result of previous marketing measures that may have taken place outside of measurable touchpoints.

Lack of transparency in algorithmic models

Data-driven attribution works with non-transparent evaluation logic: the algorithm decides which touchpoints are weighted and how, based on its own criteria, making this black box difficult for strategic decisions—you don't know why the system favors or disadvantages certain channels.

The key insight: model results are approximations, not absolute truths, because every attribution model works with incomplete data and assumptions, a limitation that should be explicitly taken into account in every strategic decision.

The limitations of attribution models:

No model captures the entire customer journey
Tracking restrictions cannot be completely resolved technically
Offline touchpoints remain systematically invisible
Attribution provides guidance, not absolute certainty

Conclusion: Using attribution as a strategic tool

Attribution models evaluate the contribution of individual marketing channels to conversion. Choosing the right model depends fundamentally on the business context, the customer journey, and the available data.

No model is universally correct—only suitable or unsuitable for the specific application:

Single-touch models such as Last Click or First Click are easy to understand and implement, but provide an incomplete picture of complex customer journeys.
Multi-touch models such as Time Decay, Position Based, or Linear provide a more differentiated representation of reality, but require more data and a more precise interpretation.
Incremental attribution provides the most accurate insights into actual advertising impact, but is primarily feasible for larger campaigns with sufficient budgets.

Attribution provides guidance for strategic decisions, but does not replace the need for continuous review. Whichever model you choose, adjust your choice regularly as your business model, marketing strategy, or user behavior changes.