An attribution model is the rule that assigns credit for a conversion, such as a purchase or a sign-up, to the different touchpoints the user went through before completing it. In digital marketing and online performance measurement, this model decides which channel, ad, keyword or visit is credited with the final result. Without a shared rule, every advertising platform tends to claim credit for the entire conversion, which makes comparisons between channels unreliable.
The choice of attribution model directly affects how analytics data is read, how budgets are evaluated and how campaigns are optimized. A model that rewards the last click will make closing channels, such as branded search or remarketing, look more effective, while a model that spreads credit across several touchpoints will also give value to discovery and consideration activities. Understanding these differences is essential to interpret reports correctly and to make investment decisions that match business goals.
What it means to attribute a conversion
Attributing a conversion means establishing a causal, or at least probabilistic, link between a user action and a measurable result. In a typical customer journey, the user may come into contact with the brand through organic search, a paid ad, a social post, a newsletter or a direct visit. Each of these contacts is a touchpoint and contributes, to a different degree, to the final decision.
The attribution model is therefore the rule that distributes credit among these touchpoints. Some models give all the credit to a single contact, others split it according to fixed percentages or to data-based criteria. The choice depends on your goals, the length of the sales cycle and the availability of adequate measurement tools.
The main attribution models
Several models are well established in digital marketing practice. Each follows a precise logic and produces different readings of the same conversions. The most common ones are described below.
Single-touch models
Single-touch models assign the entire conversion to a single touchpoint. The last-click model gives everything to the last interaction before the conversion, and it is the default rule on many platforms. The first-click model instead assigns credit to the first interaction, giving value to discovery activities. There are also variants that reward the last non-direct click or the first non-direct click, excluding direct visits to the site.
Multi-touch models
Multi-touch models distribute credit across all the interactions in the journey. The linear model divides credit equally among the touchpoints. The time-decay model assigns higher percentages to the contacts closest to the conversion. The position-based model, also known as U-shaped, gives a significant share to the first and last contact and splits the rest among the ones in between.
Data-driven models
Data-driven models use statistical or machine learning methods to estimate the real contribution of each touchpoint. These approaches require significant data volumes and advanced tools, but they can adapt to complex journeys and to behavior that changes over time. They are often paired with experimental measurement techniques, such as incrementality tests, to verify the causal effect of activities.
Custom models
Custom models let you define your own rules, assigning specific weights to certain channels or events. They are useful when the purchase journey has particular features, for example a long sales cycle or a central role for a specific channel. Customization, however, requires careful governance, so that the rules reflect verifiable data rather than preconceived beliefs.
How the most common models compare
The table below summarizes the main features of the most common models, showing how each one distributes credit and which goals it fits best.
| Model | Credit distribution | Strength | Main limitation |
|---|---|---|---|
| Last click | 100% to the last contact | Simple and immediate | Undervalues discovery channels |
| First click | 100% to the first contact | Gives value to the attraction phase | Ignores the role of the final phases |
| Linear | Equal shares among contacts | Balanced view of the journey | Does not distinguish the importance of contacts |
| Time decay | Higher weights to recent contacts | Reflects proximity to the conversion | Penalizes the initial contacts |
| Position-based | Strong weights at the start and end of the journey | Recognizes the key phases | Requires careful configuration |
| Data-driven | Statistically estimated weights | Adaptive and analytical | Needs large data volumes |
What the choice of model depends on
Choosing an attribution model is not a purely technical decision: it depends on business goals, the structure of the purchase journey and the quality of the available data. A useful model is one that helps answer a concrete question, for example which channel generates valuable customers or where to intervene to improve return on ad spend.
- Define the measurement goal clearly, distinguishing between conversion volume and conversion value.
- Map the touchpoints actually present in users’ journeys, including online channels and, where possible, offline interactions.
- Check the quality and completeness of the analytics data, verifying consistency across sources.
- Compare at least two different models to see how the distribution of credit across channels changes.
- Pair the reading of models with experimental techniques, such as controlled tests, to estimate the real incremental effect.
- Document the rules you adopt and share them with the teams involved, to avoid conflicting interpretations of reports.
- Review the chosen model periodically, because user behavior and the channel mix change over time.
- Communicate results transparently, always making clear which model was used and what its limits are.
Frequently asked questions about the attribution model
What is the difference between attribution and incrementality measurement?
Attribution assigns credit for a conversion to the observed touchpoints, while incrementality measurement estimates how much of that conversion would have happened even without a specific activity. The two perspectives are complementary and are often used together to assess the real effectiveness of campaigns.
What is the most common attribution model?
The last-click model is historically the most common, because it is the default rule on many advertising platforms and analytics tools. Its popularity comes from its simplicity, not necessarily from its accuracy in representing complex journeys.
Does the attribution model change historical data?
In most cases the model does not change the raw data, but it changes how conversions are assigned to channels in reports. Some tools let you apply different models to the same data, offering alternative readings of the same activity.
Why do different platforms show different numbers?
Every platform tends to claim credit for conversions according to its own rule and its own attribution window. The differences also come from tracking limits, unavailable cookies and users who browse across multiple devices, factors that make direct comparisons unreliable.
Is attribution useful for small businesses?
Smaller organizations can also benefit from a simple model, as long as it is applied consistently and accompanied by a critical reading of the data. The important thing is to avoid drawing definitive conclusions from a single rule.
What are the main limits of attribution?
The limits mainly concern the difficulty of observing the user’s entire journey, the presence of untracked interactions and the tendency to confuse correlation with causation. For this reason attribution should be treated as an analysis tool, not as definitive proof of a channel’s effectiveness.
Dealing with this?Do you know which attribution model to use to understand the real contribution of your Google Ads campaigns?The wrong model can make you believe some conversions come from channels that did not actually generate them, leading you to spend your budget badly. I set up GA4, GTM and advanced conversions to apply the attribution model that best fits your funnel and give you reliable data.Discover conversion tracking
Related entries:
Marketing attribution model Attribuzione Attribution modelling Attribution modeling Attribution modeling in marketing