Marketing Attribution Models Explained in Plain English

Attribution is the act of assigning credit for a conversion to the ads, clicks and other touchpoints along a customer's path. A conversion is the action you want, such as a purchase or a signed form. An attribution model is the rule, or set of rules, that splits that credit.

Without a model, many teams give all credit to the last ad someone clicked. Google's own help pages ask the obvious question: was it only that ad that made the person decide? The other touchpoints on the path may have done real work.

Sources: Google Analytics Help, "About attribution and attribution modeling"

These two models are the simplest. They give 100% of the credit to one touchpoint.

• Last touch. All credit goes to the final touchpoint before the conversion. Adobe calls it the most basic and common model. It fits short buying cycles. Google says last interaction may suit a business that is mostly transactional, with no long consideration phase.
• First touch. All credit goes to the first touchpoint the customer had with you. Adobe notes it is useful for channels meant to build awareness or win new customers. Google says it fits when your campaigns create first awareness, such as when your brand is not well known.

Here is where each one misleads. Last touch ignores everything that built interest earlier, so it tends to make bottom of the funnel channels look great and top of the funnel channels look useless. First touch does the reverse. It ignores the emails, demos and follow ups that closed the sale. If you run a long B2B sales cycle, trusting either one alone can push you to cut the wrong channel.

Sources: Adobe Analytics documentation, "Attribution models and lookback windows"; Google Analytics Help, "About the default MCF attribution models"

Multi touch models spread credit over several touchpoints.

• Linear. Every touchpoint gets equal credit. Google says this suits campaigns meant to keep contact with the customer through the whole sales cycle. Adobe says it helps with longer consideration cycles.
• Time decay. Touchpoints closer to the sale get more credit. In Google's Universal Analytics, the default half-life is 7 days. A touchpoint 7 days before the sale gets half the credit of one on the day of the sale. One 14 days before gets a quarter. Adobe uses the same idea, with a default half-life of 7 days.

Linear can mislead because it treats a casual ad view and a pricing page visit as equal. Time decay can mislead when an early event did the heavy lifting. A trade show meeting six weeks ago may have started the deal, yet time decay gives it almost nothing. Time decay works best for short promotions, such as a one or two day sale.

Sources: Google Analytics Help, "About the default MCF attribution models"; Adobe Analytics documentation, "Attribution models and lookback windows"

A position based model gives most of the credit to the first and last touchpoints and a smaller share to the ones in between. Google describes a common setup: 40% to the first interaction, 40% to the last, and 20% shared by the middle ones.

Adobe offers this as the "U-shaped" model. It also offers "J-shaped" (60% to the last touch, 20% to the first) and "Inverse J" (60% to the first, 20% to the last). Adobe adds a "participation" model that gives full credit to every unique touchpoint, which adds up to more conversions than you actually had.

The weak spot is that the split is a guess. Nothing in your data says the first and last touches deserve exactly 40% each. It is a fair compromise, but it is still a rule you chose.

Sources: Google Analytics Help, "About the default MCF attribution models"; Adobe Analytics documentation, "Attribution models and lookback windows"

Data driven attribution does not use a fixed rule. Google says it uses machine learning to study both converting and non converting paths, then works out how much each touchpoint contributed. It looks at factors such as time from conversion, device type, number of ad interactions and the order of ad exposure. Each model is specific to the advertiser and the conversion event.

Adobe has a similar option called the algorithmic model, which uses statistical methods to set the credit. The benefit is less guesswork. The catch is that it is harder to explain to a boss or client than "first click wins."

Sources: Google Analytics Help, "[GA4] About attribution and attribution modeling"; Adobe Analytics documentation, "Attribution models and lookback windows"

If you use Google Analytics, check which version you are in. Universal Analytics offers the models above in its Multi-Channel Funnels tools. Google Analytics 4 offers fewer. Its help page lists three types: paid and organic rules based models, a Google paid channels model, and data driven attribution. A note on that page says the first click, linear, time decay and position based models go away starting in May 2023.

Google also says GA4 attribution models leave out direct visits, unless the whole path was direct. That is worth knowing before you compare numbers across the two tools. They will not match.

Sources: Google Analytics Help, "[GA4] About attribution and attribution modeling"

1. Match the model to your sales cycle. Short cycle, last touch or time decay. Long cycle, a multi touch model.
2. Compare two models side by side. If a channel looks strong under last touch and weak under first touch, it is probably a closer, not a finder. Both jobs matter.
3. Use the same model every time you report. Switching models between months makes trends meaningless.
4. Remember the limits. Every model sees only the clicks and visits your tools can track. Word of mouth, calls and offline meetings are missing.

No model shows you the truth. Each one is a lens. The goal is to pick a lens that fits how your customers buy and to know what it hides.

1. Google Analytics Help, "About the default MCF attribution models." support.google.com/analytics/answer/1665189
2. Google Analytics Help, "[GA4] About attribution and attribution modeling." support.google.com/analytics/answer/10596866
3. Adobe Analytics documentation, "Attribution models and lookback windows." experienceleague.adobe.com