Attribution modelling is a hot (and complex) topic but at its core it’s just about giving credit where credit is due. Now brace yourselves for a more technical explanation. Attribution modeling is the process of identifying events along the journey from ad exposure to a sale and then assigning those events significance.

Why is this important?

Putting emphasis on specific events decides where future resources are spent. It’s all about those fancy ROI and ROAS acronyms so you can recommend the most cost effective and impactful strategy.

Marketers are doing more and more across different channels, and they would like to know what’s working.Shareen Pathak, Managing Editor at Digiday

Fun Fact: 28% of Marketers don’t have an attribution model and 100% of those marketers were panicking when asked if they had an attribution model.

Prominent Attribution Models

Now that we know why attribution is important, let’s talk about how we work it out. There are many different attribution models but for a basic understanding, you really only need to know the core four.

To help us explain these let’s take the example of Jim who is getting ready for a big job interview and is trying to pick out the right outfit.

Now let’s say Jim goes to the interview and gets a compliment on his outfit – who gets the credit? There were four touchpoints in Jim’s process of putting his outfit together – a salesperson; Jim’s friend; a yogurt accident; and Jim himself. Let’s take the four primary attribution models and see who would get the credit.

Last Touch

This model gives all the significance to the event that directly preceded the sale.

Last Touch Attribution Modeling

You’re right to give pause. Last touch ignores the impact that all other events have leading up to a sale. A bit cruel when you think about it. This will favour bottom funnel tactics like retargeting – but if you only retarget people how will you fill the top of the funnel to grow your audience? Clearly, this model is not actionable.

First Touch

Similar to last touch, first touch gives all significance to the first event.

First Interaction Attribution Modeling

You noticed something a bit off didn’t you? First touch privileges that initial exposure that leads to a sale. Making a purchase isn’t like love at first site, and who knows if that’s even a real phenomenon. I’m dead inside. The biggest problem with first touch attribution is that you don’t know to what extent every other event helped and how necessary it was. You can’t really change your strategy with that kind of data. 

Linear (multi-touch)

The linear model gives all events significance on the path to a sale. It is definitely the most socialist of attribution models.

Linear Digital Attribution Modeling

This is less wrong. That’s it. Just less wrong. Use it if you are shooting for that. Avinash Kaushik, Digital Marketing Evangelist at Google

It’s likely that certain events had a greater impact on the sale but it’s pretty difficult to distinguish which events were more impactful than others without mind reading nanotechnology. Creepy. This does let you see which pieces are working but it doesn’t let you grade the pieces – so how do you optimize?

Time Decay (multi-touch)

Time decay gives the most significance to the last event before the sale and decreasing significance to events going backward in time.

Time Decay Attribution Modeling

The first event gets the short end of the stick despite how important a first encounter can be. No model is perfect, not without that mind reading nanotechnology. Creepy again… This model is still pretty good though – definitely the best of the bunch. Why? Because it lets you take action. For example, if a certain channel has the lowest score you know it’s primarily a first touch point. Then you can work out a strategy to make that touch point more effective and earn a higher score.

If an attribution model doesn’t let you take action by optimizing your media spend and touchpoints it’s not worth the binary code it’s made up of.

Where do we go now?

So, now that you have a better understanding of attribution models, you may be curious as to what model marketers use most. Well, just feast your eyes on this all telling US based chart:

Primary Attribution Model Used to Measure Digital Marketing ROI According to US B2B Marketing, 2015 & 2016


The most used attribution model is… no model at all. Thankfully we’ve seen a drastic decline in this stat between 2015 and 2016. The only thing less actionable than a bad attribution model is no attribution model at all. Almost a third of marketers are still basing their marketing decisions on gut instinct (or maybe a Magic 8 Ball) over insights gleaned from data. Scary. Very scary.

Get a Free Copy of Our Little Book of Data & Analytics

Get a free copy of our Little Book of Data and Analytics eBook (co-presented with the IAB) when you subscribe to our newsletter to receive weekly updates including eBooks, research, tips and more.

First Name
Last Name
Job Title
Rate the movie Groundhog Day
Rate the movie Groundhog Day

Contobox A Crucial Interactive Company | 12 East 49th Street, 8th Floor, New York, New York, 10017, USA


Further Reading

This post is just the tip of the iceberg. We didn’t even touch on view-through conversions, viewability or omni-channel attribution. Here are some good reads to continue your journey into attribution.

Download our Audience and Engagement White Paper – Our ebook will introduce you to audience engagement and attribution. We’ve divided it into three parts. Part 1 deals with audience engagement, what it is and why it matters, part 2 covers attribution, and part 3 brings it all together with some real-life examples.

Multi-Channel Attribution Modeling: The Good, Bad and Ugly Models – The great Avinash Kaushik’s introduction to attribution modeling. There is a lot more detail in this post than ours and it is a great next step. While you’re there check out the rest of his attribution posts – you’ll thank us later.

Digital Attribution’s Ladder of Awesomeness: Nine Critical Steps – I know we told you to search for Avinash’s other attribution posts but this one is just too good not to link to directly. This post will let you work out where you are in the attribution journey and take the necessary next steps.

Multichannel Analytics- Tracking Online Impact Of Offline Campaigns – Omnichannel, offline-to-online conversion, online-to-offline conversion, cross-channel. All these terms are thrown around to bring a 360 (also a term being thrown around) picture of your marketing efforts. This (yet another) Avinash post will give you the basics (Our addiction to his blog is problematic).

Viewthroughs Gain Attention as Attribution Tactic – Learn a little more about this controversial but important aspect of attribution modeling.

‘I would call it fraud’: Why attribution has a viewability problem – Great Digiday article that explains the importance of viewability in viewthrough attribution.

And if you’re looking to gain a better understanding of more online marketing basics with fun comics like those listed above you can always:


Steve Patrick Adams is a writer, illustrator, and animator with a glowing love of all things Ad Tech. He's been a digital media producer since 2010 when he got his start working for Bite (Blue Ant Media) and has since worked for a variety of publishers including CBC Comedy, Mashable, and IAB Canada.