Guide · Marketing
Marketing Attribution: First Click vs Last Click Models
Marketing attribution is the rule that decides which ad, click or channel gets credit for a sale. Change the rule and the same sale lands in a different place. This guide explains the common models, shows one example, and says why two tools often report different numbers for the same month.
What marketing attribution means
Most buyers do not act on the first ad they see. They may click a search ad, read a blog post, come back from an email and then buy. Each of those steps is a touchpoint.
An attribution model is a set of rules for sharing credit for the sale across those touchpoints. Google describes it as a way to choose how much credit each ad interaction gets for your conversions.
The model does not change what happened. The customer still bought. It only changes which campaign, keyword or channel looks like the hero in your report.
The common models, in plain words
Google Ads lists six models. Google Analytics, in its Multi-Channel Funnels reports, has a similar set. Here is how each one shares credit.
Model How it shares credit Often used when Last click All credit to the final click Buyers decide fast First click All credit to the first click You want to build awareness Linear Equal credit to every click You keep in touch over a long sale Time decay More credit to clicks near the sale, with a 7-day half-life Short promotions Position-based 40% to the first click, 40% to the last, 20% to the middle Both the introduction and the close matter Data-driven Credit set from your own past conversion data You have enough conversions for it to learn A half-life of 7 days means a click 8 days before a sale earns half the credit of a click 1 day before it.
Sources: Google Ads Help, About attribution models; Analytics Help, About the default MCF attribution models
First click vs last click: one example
Google gives a simple case. A diner searches four times and clicks an ad each time. The searches move from broad to specific. At the end, she books a table.
Say the four searches are:
- "restaurant tuscany"
- "restaurant florence"
- "3 star restaurant florence"
- "3 star restaurant abigaille florence"
Here is how each model shares the credit:
- Last click gives 100% to the fourth search.
- First click gives 100% to the first search.
- Linear gives 25% to each of the four.
- Position-based gives 40% to the first, 40% to the fourth and 10% to each of the two in the middle.
Time decay gives the most credit to the fourth search and the least to the first. Data-driven splits the credit by how much each search helped, based on the account's past data.
Notice the last search names the restaurant. That person already knew the place. Last click rewards that search and ignores the broad search that first brought her in.
Why your numbers disagree across tools
Many teams see one sales total in their ad account and a different one in their analytics tool. Models are a big reason.
Google says most advertisers are used to judging ads on a last click basis. It also says that view ignores the other ad interactions a customer had along the way. Google Ads lets you pick another model for each conversion action, so two accounts can follow different rules.
Google Analytics works from a wider view. Its default report model, Last Non-Direct Click, skips direct visits and gives all credit to the last channel the visitor clicked through from. Its Multi-Channel Funnels reports can also be switched to First Interaction, Linear, Time Decay and Position Based.
So the same sale can be counted for a paid search ad in one tool and for an email in another. Neither tool is broken. They follow different rules.
Other things can also move the totals, such as the look-back window and which conversions you count. Check these settings before you blame a model.
Sources: Google Ads Help, About attribution models; Analytics Help, About the default MCF attribution models
How to compare models before you change anything
You do not have to guess. Both tools let you see the same data under different rules.
In Google Ads, click the tools icon, then under Measurement choose Attribution, then Model comparison. Pick a dimension, such as campaigns or keywords. Then use the Compare and With menus to put two models side by side.
Try last click against first click to see which keywords start customers on the path. Then try last click against linear or data-driven. A keyword that looks weak under last click may look strong under first click. That can mean it brings people in and others close the sale.
If you change a conversion action's model, Google Ads also offers "current model" columns. They show how your past conversions, cost per conversion and conversion value would have looked under the new model.
Analytics has a similar Model Comparison Tool for its Multi-Channel Funnels data.
Sources: Google Ads Help, About attribution models; Analytics Help, Overview of Attribution modeling in MCF
Choosing a model and what to watch for
There is no perfect model. Pick the one that fits how your customers buy.
- If buyers decide in a day, last click is often fine.
- If you sell something people think about for weeks, first click, linear or position-based show the early steps.
- If you have enough conversions, data-driven is worth a test. Google lists it as the default model for most conversion actions in Google Ads.
Google advises you to test a new non-last-click model and see how it affects return on investment before you rely on it. The model also feeds your bids if you use automated bidding, so a change can shift your spend.
When you read a report from an agency, ask which model it used. A strong result under one model can look average under another. This is a fair question for any agency to answer.
Sources