Enhance Retargeting With a Simple Technique
Effective Retargeting Campaigns Using The GRID method
Almost all retargeting efforts, directly or indirectly, focused on one thing and one thing only, make users convert. Easy right? Not Really.
Bringing users to the platform is one thing and making them convert is another. It takes a village to move users across different steps of the funnel towards the final goal. They can drop at any step of the funnel. They land on the website and can bounce immediately, signup, and can drop off, view a product, and may do nothing. What a frustration.
A marketer has to do targeted campaigns for all types of users depending upon where they are on the funnel. They have to use all sorts of strategies to make users move to the next funnel step. And each step requires a different communication technique to educate users and make up their minds.
And don’t forget they have to do all this in a short period. The longer it takes to bring users back the lower their chance to convert gets. The lower their chance gets the more ads they have to be targeted with. Hence more cost.
And to top all of these, over time, it gets harder and harder to formalize, visualize, and monitor all the users and their status on the funnel.
In this article, I will explain a framework that can supercharge your retargeting efforts. It helps to segment your audience with retargeting lists. It is simple but very effective. And can be applied to any business type
Understanding the Concept
Before we dive into the method, let’s first understand the concepts used in the framework.
There are two things marketer has to consider.
First, the step of the funnel where the user is at
Second, how long since the last interaction with the business.
A funnel represents the journey of the personas, from the moment they know your brand until the time they become customers.
For example, an e-commerce business’s typical funnel would be:
landing on website > view a product > add to cart > checkout > purchase.
Purchase is what we call a conversion. These are call funnel steps that the user must take before the conversion.
Every business type will have a different funnel. Some businesses may also have more than one funnel as businesses may have multiple important actions that they want users to perform. Example: purchase funnel and rating funnel for Amazon.
Funnels are very important to understand where users are dropping off and what needs to be optimized to push more and more users towards the conversion.
For example, a website can have a high rate of users moving from landing page to product view but then 95% of people drop. This can indicate that the product view page needs to be optimized to make users add products to carts.
Here also comes the responsibility of the marketer to retarget user that has viewed a product. Users can be retargeted with similar products to make up their minds or provide coupons to make them convert.
And each funnel step requires a different set of approaches to move the user to the next step. Because you can’t send an add-to-cart intention email to a user who didn’t even view a product.
So they have to retarget for every step of the funnel and for all the users. Imagine business has a 10 step funnel and 10 million users. They would go nuts retargeting every one of them.
I have a question for you.
Say, two people, Jon and Max, both viewed the same phone on Google store but didn’t purchase. The only difference between them is that Jon viewed it yesterday and Max viewed its last quarter. My question to you is who has a higher chance to purchase the phone?
The more recent the user has interacted with the business the higher the chance of a conversion. Or we can say more time it takes to bring back the user the more efforts are needed for conversion.
An entirely different strategy is needed for a less recent user than a recent user.
And if we consider the steps of the funnel alongside the interaction time the challenge becomes 2x!
The GRID Method
So funnel and recency plays an important factor in designing the marketing strategies and the narrative to retarget users. The grid method combines both of them to provide a holistic view of users and allow to visualize the current standing of the business.
Let’s understand this with an e-commerce example where we have 5 steps funnel.
In the depiction above, 1000 represents the number of users who landed on the website but didn’t view a product. So each number shows who many users are at that step and didn’t move to the next step. Purchase step being that many users hadn’t re-purchased. We will discuss the re-purchase case later in the article.
So the first objective is to move the users from the top to the bottom of the funnel.
Let’s now add the time dimension in the above matrix.
For the sake of explanation, I have added two timeframes less than and greater than 30 days. Less than 30 days means users that performed given funnel step action within 30 days while more than 30 means action was performed more than a month ago.
If we put our discerning eye on above “GRID” we clearly see the user cohort broken down by their action and time. Easy right?
Marketer's whole and sole objective should be to move users from left to right and from top to bottom of the Grid.
Green being our desired outcome and red being the worst. We want as many users to be on the right side and on the lower end of this grid as possible.
Measure performance and visualize the results
Understanding how the numbers move across the GRID is imperative. Let’s discuss a case to grasp the formation before we move to visualization.
Let’s start with Red users in the above grid. These are churned users that didn’t even view a product. Say we bring all 500 of them back and all of them went to view a product. How the GRID will get updated in this scenario?
We can clearly see the change that our efforts brought, on the right. This much handy this grid system is.
The scenario that we discussed above, where churned landing users dropped from 500 to 0 and product view for recent users increased from 350 to 850, can be visualized easily in a dashboard.
For the sake of demonstration, I will use Google Sheets and Google Data studio but the same can be applied to any data source and dashboarding tool.
You can view the data in this google sheet.
Note: the numbers are fake.
The report above shows two things. One the funnel status for the given date, the current standing, the two rows at the top. Second how the numbers are changing over time for any selected funnel step, the trend line in the bottom.
Ideally, for the trend line, we would want the black line to keep decreasing and the red line to keep increasing. Meaning the re-targeting was able to bring more and more users towards the macro conversion.
Let me know if you need help in creating a similar dashboard here.
Employment in Analytics Platform
We know now which user we want to target and what we want as an outcome. But how do we use the cohorts?
I will go through some of the platforms in this article, how cohorts can be created using them, but if it doesn’t cover yours let me know and we can discuss.
In order to create cohorts in Mixpanel, you simply have to go to users>cohort and add the rules based on your grid. So for example for the cohort of users that viewed a product within 30 days but didn’t add it to the cart
you have to do this for all the cells of the GRID.
Once the cohort is created you can send emails, SMS, or push notifications to these users or import it in marketing platforms, FB or Google ads, and show advertisements to them.
Universal Google Analytics
In universal analytics, go to properties in admin > Audience Definition > Audiences.
From there create a new audience as shown in the image below.
One key thing to keep in mind is that we create the audience on user level instead of session-level. The reason being we want them to view the product regardless of the session.
Read here for more audience creation details in GA
Google Analytics 4
Goto audience from the left navigation panel to create audiences. The idea remains the same, include one event and exclude the other event for GA 4 as well.
You can read the following article for detailed instructions on audience creation in for Firebase analytics.
Audience Creation Guide for Firebase Analytics
A guide on how to create audiences in Firebase Analytics
So with data warehouses such as Bigquery, you will have to first calculate the cohorts using queries. The query would look something like this
Once you have the user cohort using SQL you can simply import them to the analytics platform of your liking. Here is how you can import data from BigQuery to Google Analytics
Enrich the GRID even Further
Our Grid already packs a tone of information about users but we can take it to another level by incorporating the frequency of an event performed by the user.
RFM analysis is a very effective technique for user segmentation. The main reason being is that it works on two important principles recency and frequency. How recent user has performed the desire action and how many times it was performed.
With the GRID method, we can also add frequency to further stratify our users. For simplicity, we will split users based on if the given action was performed 1 time or more than 1 time.
This gives an even more detailed picture of what our users are doing. Example The 500 users that only landed on our website and didn’t view any product. With frequency added to our Grid, we can now also see that of those 100 came multiple times hence were more interested but didn’t find any product to view.
Re - Purchases
With frequency, we can also discern that there are 5 users that made multiple purchases within less than 30 days hence our power users.
We can also note here is that we also had power users in more than 30 days time frame but they churned and we would really want to bring them back.
Other Ways to Employ the Technique
Since we have very detailed cohorts of users we can also use them for other purposes such as to create look-a-like audiences for the marketing campaigns. Or we can use the best cohorts for referral programs.
The sky is the limit.