Imagine you have several marketing channels—like Facebook Ads, Google Ads, Organic Search, and Email/SMS—that help drive sales (or conversions). A multi-touch attribution model tries to figure out how much each channel contributed to the final sale when they all played a role along a customer’s journey.
Two common ways to do this are:
Full Impact (Shapley-based)
Linear
Below is an explanation of each.
1. Full Impact Model
What are Shapley values (in simple terms)?
Think of each marketing channel as a “player” on a sports team. When the team wins a game (i.e., a sale occurs), you want to figure out how much each player contributed.
Shapley values do this by looking at all possible ways that players (channels) could combine to win. It calculates each channel’s fair share of the victory by seeing how often having that channel in the mix makes a difference compared to not having it.
How does this work for attribution?
For each sale, the model simulates different “what if” scenarios:
What if we had Facebook Ads + Google Ads + Email, but not Organic Search?
What if we had Google Ads + Email, but not Facebook Ads?
What if we had Organic Search + Facebook Ads, but not Email?
It compares how likely a conversion would have been with and without each channel in various combinations. This shows the marginal contribution of each channel.
Why is it called “Full Impact”?
Because it tries to measure the complete, unique contribution that each channel brings to the table, rather than just splitting credit evenly or looking only at the first or last touch.
Why group channels like Organic Search or Email/SMS?
Calculating Shapley values can get very computationally heavy if you have many unique channels. Grouping similar channels together (e.g., “Organic Search” combining Google, Bing, etc., or “Email/SMS” for all your Klaviyo sends and texts) makes the math more manageable while still giving a fair view of how each grouped channel impacts conversions.
In a nutshell:
A Shapley-based “Full Impact” model is like evaluating each channel’s true added value by seeing how the outcome changes if you include or exclude that channel in different possible scenarios.
2. Linear Model
How does it work?
In a linear model, if someone clicks three channels before buying—say they come through Organic Search, then Facebook Ads, then an Email—each channel simply gets equal credit for the sale.
So if a purchase is worth 100 points (or $100), and there were 3 channels, each gets 33.33 points.
Advantages and disadvantages
Easy to understand and implement. You just distribute credit evenly across all touches.
Doesn’t consider the unique impact of any channel. If a particular channel was crucial in “nudging” the customer to buy, linear won’t capture that extra importance—it just splits the credit equally.
In a nutshell:
A linear model is straightforward: every channel in the path to conversion gets the same slice of the pie. However, it may oversimplify the real influence each channel has.
Comparing the Two Approaches
Full Impact
Pros: More accurately represents the unique contribution of each channel. Provides deeper insight into the true impact of, say, Facebook Ads vs. Google Ads vs. Organic Search.
Cons: Computationally intensive. Requires grouping or limiting the number of channels to keep performance high.
Linear
Pros: Simple and quick to set up. Easy for stakeholders to understand at a glance.
Cons: Might misrepresent each channel’s real effect on conversions because it splits credit equally.
Simple Example
Let’s say a customer’s journey is:
Organic Search (they looked you up on Google, found your site)
Facebook Ads (they saw a retargeting ad)
Email (they finally clicked an email from Klaviyo and made a purchase)
Linear Model:
The total conversion is 100 points.
Each channel (Organic Search, Facebook Ads, Email) gets 100 / 3 = 33.33 points.
Full Impact Model:
The model looks at different scenarios (with or without Organic, Facebook, or Email) to see how each changes the chance of purchase.
If Email is generally the final push, it might get more than 33.33 points. If Facebook is critical in the middle for retargeting, it might get a fair share that’s higher than a simple 1/3 split. If Organic doesn’t change outcomes as much, it might get a lower share.
The final number each channel gets will reflect how crucial it was in closing the sale, on average.
Key Takeaway
Linear is a quick-fix method: everyone in the journey shares the credit equally, but you might be under- or over-valuing certain channels.
Full Impact (Shapley) digs deeper, looking at how much each channel truly moves the needle for conversions, giving a more fair distribution of credit.
In practice, many companies use grouped channels (like combining all search platforms into “Organic” or combining Email and SMS under “Email/SMS”) to keep the Shapley-based approach computationally feasible while still getting a more accurate read on each channel’s marginal contribution than with a purely linear model.