Third-party attribution is a necessary evil 😈
Our marketing agency keeps complaining about attribution tracking. It annoys me. Attribution is now black-hole ML models, thanks to Apple's privacy changes. Let's break it down to find alternatives.
👋 Hola! Welcome to Out of Singapore. I am Shantanu and I write weekly about the businesses I am building. I want to build 5 brands that do at least 1 Mn annual revenue in the next 5 years.
The first brand is almost there. I am working on the second one. These are not owned by me. I would like to own at least 2 out of the 5 brands I build. That’s another goal! 🫡
What are your 5 year goals? Please write to me. I am keen to listen.
Today, no polls. We jump straight to the topic. 🎯
Isn’t marketing = content
Few weeks back, I argued that marketing is essentially content distribution. How do you track if a specific marketing channel is working better than others? There comes the devil - attribution.
Advertising is expensive. The richest companies in the world are pseudo advertising companies (read: Google, Meta). Advertising is like gambling. The house always wins. However, you can make advertising work with proper attribution.
Here is the game plan for today
What is attribution?
How do you attribute? Enter pixels.
What is third party attribution?
Attribution tricks that are free
Let’s get started 💨
#1 What is attribution?
Attribution tells us which ad platform is helping us get a purchase, a lead or awareness. This is how it looks in real life -
My Facebook ad about magnesium glycinate on sleep health benefit got me these 10 purchases today.
There is no straight marketing funnel and hence attribution can never be clean. This same user purchase will be claimed by Google Ads, Klaviyo or other marketing tool that you might be using.
Look at the customer journey below for reference.
Customer interactions are multi-touch. Our customers have 4.5 times touchpoints on average before they make a purchase. It take them 6 days since the first interaction to final purchase. I do not have industry benchmarks to validate if these are good numbers or bad. However, this is what I learnt
The customer needs convincing. This means same customer should see a “flow” of ads designed to convince incrementally in each interaction.
Purchases are not impulsive. A customer takes almost a week to make up their mind.
Facebook is the source of acquisition. Consumers discover us on Facebook, not Instagram or Google.
Google is cheaper than Meta. However, the volume is small.
Attribution is powerful. It helps you take efficient business calls. However, it is not straight-forward or clean.
#1a Complicated attribution logics
Here are the three types of attribution methods commonly used.
First click attribution - First interaction point gets all the weight. I never use this.
Last click attribution - Last interaction point gets all the weight. I use it often.
Linear Attribution - All interaction point gets equal weight. I use this sometimes.
Attribution methods that that are variations of linear attribution -
Time decay attribution - The most recent interaction point gets the highest weight. It decreases as you go further behind in time. I like the concept, but I haven’t used it.
U-shaped attribution - First and last touch point get higher weight. Everything else in between gets a low weightage. Assume, first point gets 40%, last point gets 40% and all other points in between get 20% in total. I have used it for a while.
W-shaped attribution - This gives high weightage to three points, namely, first, mid, and last touch point.
Most DTC brands won’t have the cash to use or set up all of these models. The easiest one that is set up on Google Analytics - Last click and Data Driven - are the ones used by brand.
Before you can apply attribution model, you need to track user behavior. Let’s get to the second question for today.
#2 How do you track user behaviour?
To attribute, you need something to track the user behaviour. User behaviour for advertising simply refers to the information on interaction touch points.
The most common way of tracking is saving a cookie file in every users system. That’s how Meta pixel works. This is called browser based tracking.
With iOS 14.5 update, this no longer works for Apple users. There is second type of tracking - server side tracking. Simply put, it is not dependent on the user’s browser. Rather it leverages the website’s server. It avoids ad blockers, gives accurate data. Beyond this, it is a technical concept and I choose to stay away.
Here are the three common methods of tracking.
Install a pixel for browser based tracking - Meta Pixel, Tiktok Pixel, Pinterest Pixel, Twitter Pixel. Essentially, every ad platform that you use needs their own pixel to be installed.
Install tag for server side tracking - Conversion API for Meta, Tiktok Event API, Amazon conversion API. These user a tag manager like Google Tag Manager (GTM) or Adobe Launch. I have only used GTM till now.
In-platform tracking by Shopee and Amazon - I would say this is the easier implementation since users have their profile saved on the respective apps. However, what would I know about complexity?
#3 What is third-party attribution?
Well, first thing - attribution is not in our hands. It is all controlled by the tags or pixels. It is critical that they are installed properly.
There are two types of attribution methods
Platform attribution - Meta pixel, Google tag, Tiktok pixel. They don’t care about other touch points. If any user interacted with their and purchased within 7-days (depends on your definition), they claim the purchase as their own.
Third party attribution - GA4 using Google tag, Triple Whale, North Beam, Polar Analytics. They act as independent players helping you identify the contribution of each channel.
Third party attribution tools track the journey of user across all platforms. The most famous tool on Shopify is Triple Whale. Unfortunately my bigger brand is on BigCommerce. The alternative is North Beam, again a great solution but prohibitively expensive.
Google Analytics is a great tool to start tracking customer journeys across different marketing channels. However, it is not customisable. I also face so many errors that using it frustrating.
Third party tracking tools allow you to take independent decision on your ad spends. It tells you the most common touch points and the ones that lead to final purchase. Life is so sad without such a tool at brand marketing team.
#4 Attribution tricks that are free
Here are the 4 things that don’t cost you any money to mimic third party tracking.
Google Analytics - GA4 is your friend. Entirely free to use.
UTM for every link you put out - This ensure that you can track user behaviour even if GA4 says something else.
Coupon code - This is a neat hack. Send out different coupon code for each platform and track the one being used to make purchase. This is manual but the safest. Of course you won’t know if it was last click or not.
Post-purchase survey - Most straightforward, just ask the customer how did they discover you. It is possibly the truest response. The survey should be given right after the purchase is complete on the thank you order page.
That’s it for today. I wanted to complain about tracking breaking every few weeks. I was thinking about sharing tools that improve data quality and save customer info for targeting later. But for now, let it be. Too much technicality and they aren’t needed for daily operations of a brand.
I have some great insights to share about our first Black Friday Sale. I had been looking forward to this week for 2 months. It has proved to be massive success (okay not entirely true). Wait for my newsletter next week to read about BFCM sale performance. 😃
If you have read till here, can you please
Like the post if this was interesting
Send me reply - “Can be improve” or “boring” - if you didn’t like it.
Share with a friend who might like this.
My dear friend Jisoo likes you so much! Thank you again!
Have a great week ahead. Cheers 🥂