Last-click attribution is not just imprecise. It is systematically biased in ways that lead to predictable, repeatable budget misallocation. It over-rewards the final touchpoint in a conversion journey and under-rewards everything that created the conditions for that conversion to happen. For paid media programmes with any meaningful investment in top-of-funnel or mid-funnel channels, this bias compounds over time into significant strategic error.
The solution is not to declare attribution a solved problem. It is not. But there are practical steps available to most marketing teams that produce a materially more accurate picture of how ad spend connects to revenue, and they do not require an expensive enterprise platform to implement.
Consider a typical B2B buyer journey: a prospect sees a Meta prospecting ad, clicks through and reads a blog post, does not convert. A week later they see a display retargeting ad, return to the site, spend time on the pricing page. Two weeks after that, they search your brand name on Google, click a branded search ad, and submit a lead form.
Under last-click attribution, 100 percent of the credit for that lead goes to the branded search campaign. The Meta prospecting ad receives nothing. The display ad receives nothing. If you are optimising budget allocation based on last-click data, you will systematically defund the channels that initiated the journey and over-invest in the channels that harvested it.
This creates a self-fulfilling dynamic: top-of-funnel channels appear to underperform, budgets are cut, the pipeline of mid-funnel opportunities shrinks, and branded search volume declines as a result. The attribution model has not just misreported performance; it has actively degraded it.
Understanding the available models is a prerequisite for choosing the right one, or more accurately, the right combination.
First-click: assigns 100 percent of credit to the first touchpoint. Valuable for understanding which channels are most effective at initiating journeys, but blind to everything that follows.
Linear: distributes credit equally across all touchpoints. Simple, democratic, and often produces a more balanced picture than first or last click, but treats a brief ad impression the same as a detailed product page visit.
Time-decay: assigns more credit to touchpoints closer to the conversion event. Logical for short purchase cycles where recency genuinely correlates with influence, but systematically penalises top-of-funnel channels in longer consideration journeys.
Data-driven attribution: uses machine-learning to calculate each touchpoint’s marginal contribution to conversion probability. Currently the most sophisticated model available within GA4, and the default for accounts with sufficient data.
GA4's data-driven attribution model uses a counterfactual approach: it compares the conversion paths of users who saw a particular touchpoint against those who did not, then attributes credit based on the measured difference in conversion probability. Google’s GA4 attribution documentation describes the model as incorporating time from conversion, device type, number of ad interactions, order of exposure, and creative format into its calculations.
The limitations are practical rather than conceptual. The model requires sufficient data to produce reliable outputs: low-volume accounts or accounts with few conversion events will see the model default to last-click behaviour in practice. The model only accounts for touchpoints that GA4 can observe, which means it misses paid social view-throughs, offline conversions, and any channel not properly tagged with UTM parameters.
Critically, GA4's data-driven model operates within Google's ecosystem by default. It is better at valuing Google channels than it is at valuing Meta, programmatic display, or email, because those signals are less native to the platform. This is not a design flaw, exactly; it is a structural limitation that matters when your media mix extends across multiple platforms.
View-through conversions are conversions where a user saw an ad but did not click, then converted through a different path within the attribution window. Meta reports them by default in its platform reporting. Google Display and YouTube campaigns offer them as an option.
They are genuinely contested territory. The argument for including them: in channels where ad formats are designed to build awareness rather than drive immediate clicks (video, display, social), ignoring view-throughs systematically understates the channel's contribution. The argument against: view-through windows can be set generously enough to capture conversions that would have happened regardless of the ad exposure.
A pragmatic approach: report view-throughs separately from click-through conversions rather than adding them together. Use view-through data to understand reach and awareness impact, not to justify performance against ROAS targets. When Meta reports a seven-day click plus one-day view conversion, be explicit about which component is click-through and which is view-through when presenting results to stakeholders.
You do not need a dedicated attribution platform to implement meaningful multi-touch analysis. A Google Sheets or Looker Studio build using UTM data from GA4 can produce a workable position-based or linear model for most paid media programmes.
The prerequisites are consistent UTM taxonomy and a unified conversion event structure. Without these, the underlying data is too fragmented to model accurately.
UTM structure: every paid channel must use a consistent naming convention for source, medium, campaign, and content parameters. Inconsistency at this level, mixing "cpc" and "paid" as medium values, for example, or omitting UTMs from paid social link posts, makes multi-touch modelling impossible regardless of the tool you use.
Conversion events: define the conversion events that matter for your business and ensure they fire consistently across all channels. A lead that enters via Meta should trigger the same GA4 key event as a lead that enters via Google Ads. If your conversion tracking is platform-specific and not unified at the GA4 level, you cannot do cross-channel attribution.
Once these foundations are in place, you can export GA4 path data and apply custom attribution weights in a spreadsheet. It requires manual work but produces a more accurate picture than any single-platform report.
Media Mix Modelling (MMM) approaches attribution from the opposite direction to click-level tracking. Rather than following individual user paths, it uses regression analysis to identify the statistical relationship between marketing spend across channels and aggregate business outcomes over time.
Northbeam’s guide to MMM describes it as a top-down method that works from aggregated data to identify broad patterns and long-term channel effectiveness, in contrast to multi-touch attribution's granular individual-journey focus.
The practical threshold for MMM is broadly around £500k to £1 million in annual paid media spend. Below that, the statistical signal-to-noise ratio makes the model outputs unreliable. Above it, MMM provides something that click-level attribution cannot: an estimate of each channel's incremental contribution that accounts for offline effects, seasonality, and the long-term impact of awareness-building activity.
Tools such as Northbeam and Triple Whale offer third-party attribution that sits between platform-native reporting and full MMM, providing a cleaner cross-channel view than GA4 alone for DTC and e-commerce brands.
Regardless of budget size, these three steps will improve attribution accuracy immediately:
First, audit your UTM taxonomy. Pull a GA4 source/medium report and identify every variant that should be the same channel but is labelled differently. Standardise the naming convention and enforce it across all paid activity.
Second, define primary and secondary conversion actions across all platforms. In Google Ads, ensure only high-quality conversion events are set as primary actions. In Meta, check that your pixel is firing on the events that matter and that Custom Conversions are aligned with your GA4 key events.
Third, compare platform-reported conversions against GA4 on a weekly basis. The gap between platform and GA4 reporting is normal and expected. Tracking that gap over time, and understanding whether it is widening or narrowing, tells you more about the health of your attribution infrastructure than any single number does.
Attribution is not a problem you solve once. It is an ongoing calibration between the data you have, the models you apply, and the business decisions those models inform. The goal is not perfect accuracy; it is reducing the systematic biases that lead to predictable budget misallocation.
The Viaduct Generation Optimisation service covers performance measurement frameworks alongside channel-level optimisation. For teams building a reporting infrastructure that connects paid media spend to revenue outcomes, the Outcomes page outlines how we approach attribution and measurement across growth programmes. To discuss your specific attribution setup, get in touch.
