How to Use Customer Behavioural Data to Improve Retargeting on Facebook

Kostas Pardalis

Do you want to improve your retargeting on Facebook?

Of course, you do. How about enriching it with some Customer Behavioural Data from our product?

OK, let’s take it a step back. Creating a customer acquisition framework includes the definition of some funnels. They describe the different steps for a prospect to become a paying customer. These funnels are linear models of how a person might interact with our product or service.

Linear models are easy to reason about, and that’s why people love to work. But, they are rarely an accurate approximation of reality.

Linear models are easy to reason about, and that’s why people love to work. But, they are rarely an accurate approximation of reality. Only a small percentage of the people who interact with a brand will convert by following such a funnel. A commonly reported number might be as low as 2%.

That brings me to mind Seth Godin, marketing disruptor and author of 18 books. He said:

“You can’t buy mass anymore. It’s not for sale, and if it is, you can’t afford it.” Click to Tweet

Yeah. Hard right?

How to Use Customer Behavioral Data to Improve Retargeting on Facebook

But don’t worry, just because our reality is not linear it does not mean that you have to abandon your funnels.

The easiest way to deal with this non-linearity is by accepting that although a prospect might stop going through a funnel at some point, might return in the future. So either the lead will continue from the same step, start again from the beginning or just follow an entirely different funnel.

We need to make sure that this person will be aware of our existence, and this is where retargeting or remarketing ads come into play.

What Retargeting is

Retargeting ensures that someone who has come in contact with a brand will be continuously aware of it while wondering outside the pre-defined funnels that salespeople and marketers have constructed.

Retargeting helps to more accurately approximate the non-linear behavior of our potential customers using linear models like customer acquisition funnels. Click to Tweet

There are two main types of retargeting.

  • Pixel-based retargeting. We use it for traffic that lands to a digital touchpoint of our brand and bounces immediately. Visitor’s browser usually keeps a “cookie”, and whenever the prospect goes to another site — that supports retargeting ads — a message will pop up reminding the existence of our brand.
  • List-based retargeting. In this method, we curate a list of e-mails and provide it to the retargeting ads agency. This type requires the visitor/prospect to provide some contact details. For example, the person entered our funnel and abandoned it at a later stage for any reason.

In this article, we will work with List-based retargeting techniques. The reason is simple.

As the prospect enters the funnel, some interaction takes place, and thus we collect some Customer Behavioural Data. As a result, we can use insights from this data to create some ingenious retargeting campaigns.

Achieving the same result with Pixel-based retargeting is more difficult. In that case, the person has expressed little to no behavior at all towards our brand. So we rely mainly on the data the ad agency might have for segmentation and better targeting.

What we’ll need to do list based retargeting

Here’s a typical setup that we can use to perform a simple but powerful list-based retargeting campaign.

1) A mechanism that will track how people progress on the funnels that we have defined. To do that we usually need a way to capture the events a user generates on our product. For this post, we consider Intercom as the tool that collects the user-generated data.

2) A way to extract the data from Intercom and load it into a database. We will perform some basic SQL queries and analytics that will result in a list of emails to use for the retargeting campaign. For this part, we’ll use Blendo. Blendo simplifies the process of extracting the data from Intercom (and other tools) and loads it into a database.

3) We need the database as mentioned above of course. Any type will do! Blendo supports a broad range of different destinations, but select one that supports SQL and makes sense for the amount of data that you generate. * A retargeting provider. We’ll use Facebook.

Building a list for retargeting using behavioral data from Intercom

We’ll keep it simple! We’ll target all the people who have signed up for our product but didn’t convert into paying customers after a trial period.

It is a typical scenario for SaaS products and one that can act as the foundation for much more complex scenarios. In general, calculating the number of people who have churned in a given period is the inverse of calculating the retention. You may read in detail how to calculate the retention of our users using Intercom. We can build on these SQL queries to also calculate the churn.

Blendo will first pull our data from Intercom. Then it will create a table in our database with user-related data coming from Intercom. They are going to contain the following fields that we will need for our query:

  • signed_up_at: The date the customer registered to our platform.
  • last_request_at: The last time the customer was seen on the platform.
  • email: customer’s email.

Out of these three columns from the Intercom user table that Blendo creates, we can generate a list of emails who have churned in the first 2 weeks that the trial period lasts. We can do that with the following query:

select email, date_trunc('day',signed_up_at) as sign_up_date, date_trunc('day', last_request_at) as last_seen_date  
from intercom_users  
where date_trunc('day', last_request_at) < date_trunc('day',signed_up_at) + interval '2 weeks'


In the above query, we truncate the dates to a day precision instead of month that we were using for retention. Also, we are asking to get back the signed_up_at and last_seen_date just for validation reasons.

In the final version of the SQL query that we’ll use, we should remove these fields and keep only the emails.

This query will return all the customers that have churned in the first two weeks from the beginning of our product.

Of course, we can limit this behavior to any particular interval. As the modification to the query is quite simple, we will omit it for now, and in the next post, we’ll see how we can apply much more sophisticated segmentation techniques. Let’s see how to retarget your customers with Facebook.

Retarget your customers using Facebook

With Facebook, you can use both retargeting methods. To perform list-based retargeting, you need to do the following steps:

1) Create a list of the customers you’d like to retarget

Yep. It’s this post.

2) Export it

Save or copy/paste as a CSV or TXT the content of the previously created list.

3) Upload it to Facebook to create Custom Audiences

Create a custom audience on Facebook by uploading the data (or you can just copy/paste it).

Here is a nice video from Facebook about how to do it. It’s quite easy when you have a list of emails to be used.

After you have created your custom audience, you can use it for retargeting by setting up a Facebook ad for this audience. Information on how to build and run an ad campaign on Facebook is beyond the scope of this post, but you can find plenty of information on how to do on the documentation provided by Facebook.

After that, sit back and watch the hordes of customers coming back to reactivate their accounts.


To be continued…

By using the right tools like Intercom and Blendo, you can quickly start retargeting your customers and optimize your funnels in this highly non-linear world we live. The method we presented here is just the tip of the iceberg, but its purpose was to show how easy it is to start with retargeting.

In the next posts there will be some more advanced techniques on how to segment better the customers based on their behavior, to achieve better retargeting results.

Until then, start syncing your data using Blendo to understand better how your customers behave!

Happy experimenting! Let us know what questions you have in the comments section below.