Sponsored Article
Google killed your favourite attribution models – now what?
02 April 2025
Marketing success often hinges on understanding not just what works, but why it works. In pay-per-click (PPC) advertising, attribution models are often the answer to this puzzle, revealing which interactions drive customers to conversion. Here, Lorna Wilde, PPC Specialist at Stone Junction, explains how PPC attribution models can help to understand the true value of campaigns and improve decision-making.
For marketing managers, attribution models might seem complex, especially if familiarity with tools like Google Ads is limited. But grasping the basics – how these models work and how they’re calculated – can transform your approach to PPC campaigns.
At their core, attribution models are frameworks that assign credit to different interactions a customer has before completing a conversion. These interactions, or touchpoints, could include clicking a search ad, watching a YouTube video, seeing a display ad or engaging with a remarketing campaign.
In the last two years, attribution modelling in PPC has undergone a major transformation. In June 2023, Google Ads retired several long-standing models, including first-click, linear, time-decay and position-based attribution. This shift leaves advertisers with only two viable options: last-click attribution and data-driven attribution (DDA).
While this simplification streamlines Google’s approach, it also forces marketers to rethink how they measure campaign success. For industries with long and complex buying cycles, understanding the mechanics of these models is critical to optimising ad spend and accurately assessing campaign performance.
Last click attribution
Last-click attribution follows a deterministic model that assigns 100 percent of the conversion credit to the final ad interaction before a purchase or lead submission. This approach makes tracking straightforward because it eliminates ambiguity – if an ad was the last touchpoint before conversion, it gets full credit.
Last-click attribution is event-based, operating within Google’s click-tracking infrastructure. Each ad interaction generates a GCLID (Google Click Identifier) that is stored for attribution. When a conversion occurs, Google’s system checks the most recent GCLID associated with the user’s session and attributes the conversion accordingly.
The simplicity of this method makes it particularly effective in environments where conversion paths are short. For example, if a prospect searches for an ‘AC servo motor replacement’ and clicks an ad before completing a purchase in a single session, last-click attribution provides an accurate reflection of the user journey.
However, in multi-touch environments, where customers interact with multiple ads, this model creates attribution bias, disproportionately favouring lower-funnel activities such as branded search campaigns.
Data-driven attribution
Data-driven attribution takes a probabilistic approach by leveraging machine learning to assign conversion credit based on observed patterns. Unlike last-click attribution, which uses static rules, DDA analyses large volumes of historical data to determine which interactions increase the probability of conversion.
The model examines sequences of user interactions and uses a value-based algorithm to assess the marginal impact of each touchpoint.
Google’s DDA model processes user journeys through a framework that compares paths that led to conversions with those that did not. This approach accounts for multi-channel interactions, including display, search, remarketing and video ads.
The system assigns conversion weight dynamically, with the algorithm continuously refining its credit allocation based on new data. Unlike traditional rule-based models, which assume that every customer journey follows a linear progression, DDA adjusts its attribution weighting in real time, reflecting actual user behaviour.
The advantages of DDA become clear when analysing customer journeys in industries where decision-making cycles are lengthy and involve multiple stakeholders. In these sectors, it is common for prospects to engage with early-stage content, such as whitepapers, industry reports, data sheets or technical webinars before taking action.
A last-click model would disregard these upper-funnel touchpoints, attributing full credit to the final ad interaction. DDA, however, can identify the statistical impact of these early engagements, distributing credit accordingly and ensuring that top-of-funnel efforts are properly valued.
However, despite its strengths, DDA is not without limitations. The model requires a high volume of conversion data to generate reliable insights, making it less effective for smaller advertisers or campaigns with infrequent conversions.
Additionally, its black-box nature means that marketers have limited visibility into how credit is distributed across touchpoints.
Unlike rule-based models, which provide deterministic and repeatable results, DDA continuously evolves, making direct comparisons between different time periods more challenging. This can complicate performance analysis, particularly for teams that rely on historical benchmarking.
Selecting the right attribution model
The decision between last-click and data-driven attribution depends on the structure of a PPC strategy, as well as the complexity of the customer journey. For businesses with short conversion cycles, such as industrial e-commerce companies selling components, last-click attribution remains a viable option due to its clarity and ease of implementation.
However, for B2B and STEM companies with multi-touch sales funnels, DDA is the superior model as it accounts for the influence of upper-funnel interactions and provides a more accurate reflection of campaign impact.
Google’s deprecation of rule-based attribution signals a broader trend towards automation and machine learning in digital advertising. As AI-driven models become the standard, marketing teams must adapt by ensuring that their PPC strategies align with these new methodologies.
Regular auditing of attribution data, A/B testing of campaign structures and the integration of offline conversion tracking will be essential for maximising the effectiveness of DDA in complex buying environments.
For marketing managers in technical industries, adopting a data-driven approach is no longer optional. Understanding how attribution modelling works at a granular level provides the foundation for smarter budget allocation and campaign optimisation. As Google continues to refine its machine learning models, leveraging probabilistic attribution will be key to gaining an edge in increasingly competitive PPC landscapes.
Understanding which campaigns drive results is only half the battle after all – knowing why they succeed is what turns data into strategy. By choosing the right attribution model, you can move beyond surface-level insights and make decisions that truly reflect the impact of your PPC efforts.
To keep up to date with the latest theory and practice in technical PR, subscribe to Stone Junction’s podcast, The Junction Box, on Apple Music, Spotify or wherever you prefer to get your podcasts today.
Contact Details and Archive...