Attribution Models Explained

What is attribution in marketing?

Attribution is the process of assigning credit for a conversion to specific channels, campaigns, creatives and touchpoints. It connects marketing activity to business outcomes so you can see what actually influenced results.

The four things being credited are channels (such as paid search, social, email and display), campaigns, creatives and touchpoints, which are the individual interactions between your brand and a prospect.

It matters because every channel claims credit for the same conversion, and channel owners tend to overestimate their contribution. Without attribution, budgets are misallocated, underperforming activity keeps funding and what truly works is underinvested.

What are the main attribution models?

The right model depends on your goals, sales cycle length and the complexity of the customer journey.

When should you use last click or first click?

Last click is the default in most native platforms, including Google Analytics and Facebook Ads. It is easy to explain and works for simple journeys, short sales cycles, impulse purchases and direct response campaigns. Its weakness is that it ignores earlier touchpoints and can lead to overinvestment in bottom-funnel tactics at the expense of brand building.

First click credits the touchpoint that introduced the customer to your brand. It helps measure awareness and brand building. However, it can misattribute conversions on long decision paths, and it tends to overvalue display and social while undervaluing search and email. Use it alongside other models.

In practice, organisations running several channels should supplement last click with at least one alternative model, rather than relying on a single view of performance.

How do linear, time decay and position-based models differ?

Linear gives every channel equal credit. It is useful for understanding overall channel effectiveness in multi-touch campaigns, but it does not separate high-impact from low-impact interactions.

Time decay gives more credit to recent touchpoints. It suits shorter purchase cycles and reflects recency bias in consumer behaviour. Both models help surface undervalued mid-funnel interactions, and both beat single-touch methods for complex omni-channel campaigns.

Position-based, also called U-shaped, values discovery and decision moments. It suits longer B2C journeys such as education, health and insurance. Its gap is that it does not reveal the influence of mid-funnel assets, which can undervalue nurturing touchpoints.

What is data-driven attribution?

Data-driven attribution (DDA) uses platform data and machine learning to calculate each touchpoint's real contribution. It is available in Google Ads, GA4, Meta and other major platforms. It is tailored to each account, learns over time and reduces reliance on last click.

It needs enough data to activate, typically 3,000 or more ad interactions and 300 or more conversions in supported conversion categories. To implement it, activate it in platform settings, allow one to two weeks for data collection, monitor performance when switching models and adjust bidding based on what you learn. Its limits are that the calculation is a black box, it needs consistent conversion tracking and it may differ from third-party platforms.

Multi-touch attribution (MTA) goes further by mapping the full path across channels and devices. It needs full tracking, significant data volume and consistent user identification, so it is most common in enterprise settings.

Why do Google Ads, GA4 and Meta report different numbers?

Each platform uses its own method. Google Ads reports the last Google Ads click before conversion, prioritises its own touchpoints and uses a default 30-day click window. GA4 supports multi-channel models, includes all tracked channels and now defaults to data-driven attribution. Google Ads typically shows higher conversion counts, and the two will never match exactly.

To reconcile them, document the expected discrepancies, set a primary source of truth and use consistent lookback windows.

Meta offers 1-day click, 7-day click, 7-day click with 1-day view and 28-day click windows. The default 7-day click may not suit longer consideration periods, so compare settings and pair Meta data with GA4 or Looker Studio.

For GA4 set-up, define key events, enable Google Signals for cross-device tracking, use conversion path reports and set the primary model to data-driven where available.

How do you handle cross-platform attribution challenges?

Meta, Google and programmatic can all claim the same conversion, which inflates totals. View-throughs add complexity, as platforms weight them differently and use windows from 1 to 28 days. Privacy changes such as iOS 14.5 and cookie deprecation add blind spots.

The answer is a unified source-of-truth dashboard, for example in Looker Studio, with consistent models, incrementality testing and blended metrics. For view-through, set lookback windows of 7 to 30 days and use blended view and click data to avoid undervaluing upper-funnel activity.

A New Zealand health campaign illustrates the payoff. Credit split 25% YouTube awareness, 25% native content and 50% Google Search. Data-driven analysis showed conversions dropped 30% without early YouTube exposure. Mid-funnel investment rose 40%, which brought a 27% lift in conversion rate and a 22% lower cost per acquisition within three months.

How should attribution shape budget planning?

• Compare first-touch and multi-touch reports and look for patterns in high-performing journeys.
• Do not plan on last click alone.
• Shift investment towards channels that start customer journeys, and balance awareness, consideration and conversion.
• Validate findings with incrementality tests.
• Review models quarterly.

Avoid common mistakes. Platform-reported ROAS can be inflated by 20 to 30% because platforms claim conversions that would have happened anyway. Not all conversions are equal, as new customers are typically worth three to five times a repeat purchase. Branded search can hide the effect of ads elsewhere. For many businesses, 40 to 60% of the journey happens offline.

Frequently asked questions

Written by the KBR team, an independent media and digital agency in Auckland and Christchurch.