How 5 models split credit across touchpoints in a customer path.
Inputs
3 to 10 typical
Revenue from the conversion
Credit allocation by model
Touch
First-Touch
Last-Touch
Linear
Position (40-20-40)
Time Decay
When each model is most useful
Model
Best For
Bias
First-Touch
Top-of-funnel discovery channel ROI
Overcredits awareness, undercredits conversion
Last-Touch
Last-mile / conversion-optimized channels
Overcredits retargeting and branded search
Linear
Default fair-split when path matters but no preference
Underweights truly important touches
Position (40-20-40)
When discovery + closing both matter, middle is filler
Arbitrary 40/20/40 split
Time Decay
Short consideration windows; recent signal matters most
Underweights early-funnel awareness
About this calculator
Attribution is one of the most consequential and least-understood decisions in performance marketing. The model you choose determines which channels look profitable, which campaigns get scaled, and which get cut. Two operators looking at identical raw data can reach opposite conclusions about Meta vs Google ROI based purely on attribution choice.
The most common failure mode is using last-click attribution by default and never questioning it. Last-click systematically overcredits whatever sits closest to the conversion — usually branded search and retargeting — while undercrediting the awareness channels that put the customer in market in the first place. Brands that audit this often discover they have been starving the channels that drive demand and overfunding the channels that capture demand.
The opposite failure is treating first-touch as the "real" attribution. First-touch overcredits whatever introduced the brand, regardless of whether that channel was efficient or whether the customer would have found you anyway through other channels. Without controlling for organic baseline, first-touch can make poorly-targeted top-of-funnel campaigns look like discovery wins.
The mature approach is multi-model triangulation. Look at first-touch, last-touch, and position-based for the same campaign. If all three agree the campaign is profitable, it probably is. If they disagree dramatically, dig into the conversion path — usually the disagreement is signal that the campaign is doing something specific (driving discovery but not closing, or closing only customers who would have converted anyway). Pair this calculator with the Tracking Gap Estimator and Conversion Lag Calculator for a complete attribution audit.
Frequently asked questions
What is attribution and why does the model matter?
Attribution is how you assign conversion credit to marketing touchpoints. A customer might see a TikTok ad, search Google twice, click a Meta retargeting ad, then buy from an email link. The model decides how much credit each of those touches gets — and that allocation directly drives budget decisions. Different models can change reported channel performance by 200%+ on the same data.
How do the 5 standard models differ?
First-touch gives 100% credit to the first interaction (good for measuring discovery channels). Last-touch gives 100% to the last interaction (good for measuring conversion drivers). Linear splits credit evenly across all touches. Position-based (40-20-40) gives 40% to first and last, 20% spread across middle touches. Time-decay weights credit toward later touches using exponential decay (typically 7-day half-life).
Which model is right for my business?
There is no single right answer — the question is what decision the model is informing. For top-of-funnel investment decisions, first-touch or position-based are more useful. For conversion optimization and last-mile efficiency, last-touch or time-decay are more useful. Most mature operators look at multiple models and triangulate, since any single model overweights some channels and underweights others.
Why do platform reports (Meta, Google) disagree with my own attribution?
Because every platform reports last-click within their own attribution window — which double-counts conversions across platforms. Meta claims a sale, Google claims the same sale, your platform aggregates and shows higher numbers than actual revenue. The fix is single-source attribution from your store data (Shopify, etc.) rather than summing platform reports.
What about data-driven attribution (DDA) and MMM?
Data-driven attribution uses machine learning on your conversion paths to derive credit weights — only useful if you have thousands of conversions and reasonably-tracked path data, which most DTC brands do not. Marketing Mix Modeling (MMM) is a regression-based approach that does not require user-level tracking — gaining popularity post-iOS14 because it works with aggregate data. Both are more sophisticated than the 5 models here but require more data infrastructure.