👋 Hey, it’s Sundar! Welcome to experiMENTAL where I share B2C startup marketing frameworks, how-to guides, and stories from 5 years of early Marketing at Uber and 10+ years at B2Cs.

What is attribution

A customer’s purchase journey is non-linear, hard to track, and hard to measure with so many Marketing touch points.

Marketing attribution is the science of determining which marketing tactics are contributing to sales or conversions. In other words, which tactics are working and which are not. The goal is to help marketers do more of the “good” and less of the “bad”.

What is Last-Click attribution (LCA)

Something as complex as understanding which Marketing tactics lead to business impact is never going to be simple.

However, Last-Click attribution tries to do just that by simplifying things. As you can tell from the name, Last-Click attribution gives credit to the channel that the customer clicked on last. It became popular when Google added UTM tracking and used Last-Click as the default attribution methodology on Google Analytics.

Unfortunately, LCA, is often misguiding, lazy, and wrong. It’s not the fault of the attribution model, but how it’s used and the lack of understanding behind it. Let’s look at an example.

Susan sees the following:

• Day 1: An ad on Instagram for a book

• Day 2: An ad on Facebook for the same book

• Day 3: A billboard for the same book

• Day 4: She searches for the book on Google and buys it

LCA would give Google search the full credit.

It ignores that the customer:

• first saw an Instagram ad

• saw ads 3 days in a row

• searched for the book showing high intent

The marketer will then keep investing into Google Search and keep budget there which is great for Google, but not so great for the company. Why? Because it ignores all the other touch points which deserve credit and attributes all of the success to the last channel.

Imagine you’re back in high school and you have a group assignment. The 4 of you work on a project and one person hands it in to the teacher. The person that hands it in gets the credit.

“Um, teacher… we all did the work”

“Sorry, Timmy, I just saw that Sundar handed in the work so I have to give him full credit because I know he did that bit”

“WTF kind of weird rule is that?!”

“It’s called last click attribution and it’s how I always grade.”

If it it’s so bad, why are most companies still using it?

Why LCA won’t go away

Before we look at LCA, there’s a parallel example that I love using.

Microsoft Excel came out in 1985. 40 years later, it is still the #1 spreadsheeting tool. It’s also the #1 organization, productivity, and planning tool. No matter how much people try, it will never go away.

Just like Excel, LCA is easy to:

• 📚 Understand

• 🔧 Setup

• 🔨 Use

Easy to understand

Imagine going to a Finance leader and saying the following:

“Well, 20% of the credit goes to facebook. 20% to instagram. 20% to an email we sent. 20% to a billboard. 20% to google search”.

Um, what does that even mean? What would happen if we didn’t have Facebook? Would you distribute the rest of % to the rest of the channels? How should we invest our money?

LCA is easy to understand and therefore it becomes the default for attribution methodologies. You can try all you want to do more but simple always wins.

Easy to set up

To set up LCA, you’ll need UTM parameter tracking on all your ads and links that track to your website. There’s no additional software required and the logic is straight forward. Whatever the UTM parameter is associated with the action gets the credit. There’s no blackbox algorithm or if/else trees.

If you’re mainly a mobile platform, then last click attribution works too but you won’t be able to know on an individual basis who signed up given some of the privacy changes from Apple. You’ll only know at the campaign level, so LCA works but not nearly as well as on websites.

Easy to use

Last click is easy to understand and so it’s easy to report and discuss. It’s also easy to make decisions off of. Reminder though that ease of decision has no correlation to quality of decision.

When a startup is in early stages, you’re moving fast and need to make quick decisions. You don’t have the luxury of incrementality tests or the data to build more advanced models like Multi Touch Attribution (MTA) or a Media Mix Model (MMM). Because many companies start with LCA, it continues to stay the default for a long time. That inertia to not change is powerful.

Wrapping up

If someone is telling you LCA is trash and you need to improve, it’s for 1 of 2 reasons:

• They have more resources than you

• They are trying to sell you something

All of the world class Marketing teams have more rigorous approaches than LCA but they also have data scientists with PHDs in econometrics on the payroll. If you can afford that, then it’s time to upgrade LCA.

The other category is Martech vendors trying to sell you on their “new and improved” shiny package. If you have at least 2-3 years of reliable data on your ad spending and business outcomes across channels then it would be worth exploring them. If not, then just keep going with LCA.

The last thing on LCA is that we used it at Uber for longer than people might realize. We combined our Paid + Organic channel into one and made decisions off LCA even while I was there from 2016 - 2021.

LCA might be wrong, but it’s consistently wrong and you can use that to your advantage. Understand your channels and the context in which you’re seeing LCA attribute and you can make sound decisions. Don’t be shamed for using LCA.

Missed the last newsletter?

Check it out: Why companies shouldn't do Black Friday