Documentation        Log in

Advanced Techniques to Capture Engaged Users with Facebook Ads | Fixel

In this post we discuss Advanced Techniques to Capture Engaged Users using Facebook Ads.

These tactics were extracted from an expert session given by Etgar Shpivak, CEO of Fixel at a professional webinar co-hosted by Fixel and Zalster. You can view the full recording of the webinar here.

I like to start with this quote by one of the leading marketing pioneers in the 19th century, John Wanamaker. Since, then everything has changed, and yet, nothing has changed. We will try to understand how this applies to your retargeting and localized campaigns, and how you can dramatically improve your results in a simple manner.

The challenge that we face with our retargeting campaigns is focused on understanding the intent of users. Because of rising media prices, understanding the way visitors engage with your content has become critical issue, when we look to build successful and scalable data-driven campaigns.

But what is this intent? How can we know if someone who visits our site is relevant or isn’t?

The simplest way is by understanding what he did in the site. We know that when someone has completed a purchase, and even has added to their cart or even began the checkout process.

In a B2B landscape, we can observe other meaningful interactions, such as viewing the pricing page, as an indicator to a user with a high intent to make a purchase. It’s usually quite easy for us as marketers to identify these URLs or actions, and highlight these relevant visitors.

But the reality is that 10% or less of the people engage with such explicit actions. It means that if we have 100,000 people on our site, less than 10,000 of them will actually take an explicit action.

So, what do we do with the remaining 90%? How can we as marketers focus on them effectively?

If we focus our remarketing efforts strictly on down-funnel who expressed this explicit interest, we’re losing out on scale and leaving “money on the table”.

If we aim too wide, we will probably target users with a low probability to convert, which will most likely result in poor campaign results.

(There can also be an issue with driving incremental sales, aka lift, but let’s park that for now)

To approach this, we’ll try to go back to Marketing 101 and the AIDA model:

  1. Awareness
  2. Interest
  3. Desire
  4. Action

We want to be able to distinguish between visitors in different stages of the funnel. Ones that are up funnel will have a lower conversion probability than ones that are down funnel. That’s pretty simple.

What we do at Fixel, and the tools I’ll show in a bit, are made to expose exactly that.

Matching User Engagement with Intent

There are several implicit user engagement signals we can use to gauge the user’s relative engagement and derive what their intent is.

Time on Site/Page

This tactic is one of the simplest to execute and yet has been proven to be one of the most efficient.

In many cases, there’s a simple correlation between the time spent on your site and the user’s intent. One way of segmenting your audience by the time spent is using Facebook’s built in time segments. These can be segmented directly in Facebook when you’re creating a new Custom Audience from Website Visitors. You can segment the users by Top 5%, 10% or 25% percentile of users by the time spent on your site. You can also cross that with other behavior such as visiting a specific page.

Another tactic is to observe the time spent on a specific page. This way we can identify users who engaged with a specific product on our site or measure engagement for a landing page. This can easily be tracked using Google Tag Manager (see our awesome guide for Time on Page tracking here).

I recommend using a threshold of over 30 seconds, but you can set a stricter or wider threshold depending on your specific needs.

Pages Viewed

The second tactic that is commonly used is the number of Pages Viewed by the user. This assumes that the more pages the user viewed, the more likely he is to convert.

From my experience, this tactic is best used in conjunction with the page the user landed on, i.e. a user that landed on a Category page and then viewed over 3 pages. Users that entered through a Product page usually have a higher intent.

This too can be implemented using Google Tag Manager (see our guide on Page Depth tracking)

Scroll Depth

The third tactic, which I really like, is measuring Scroll Depth. This tactic is highly efficient for measuring the engagement of mobile users but can apply to desktop users too.

By measuring the depth of scroll on a page, you can assume how engaged that users is. I would recommend using this on specific product pages that have a depth of over 2000px. Usually, a threshold of 50% is sufficient, but I would first gather some data to see where users drop off on the page to make the decision. For shorter pages, I advise using the Time on Page metric instead.

You can use our guide for implementing Scroll Depth tracking in Google Tag Manager to track this behavior.

Bringing it all together

After all the technical setup is done, we need to select two or three factors to categorize user engagement, e.g. Users who spent more than 30 seconds, Users that scrolled over 75% and Users that viewed over 4 pages.

You can then check the conversion rate of users that qualify for each category, so we can understand how to approach each group. This is why I advise that the events are sent to both Facebook and Google Analytics in parallel. I also recommend creating Google Analytics segments for each category to make the analysis stage simpler.

Advertisers that use the Fixel audiences, can get these broken down in Google Analytics so they can easily decide on which users to focus on.

Understanding the potential scale of each category
Analyzing the conversion rate of each category

In this example, we can see a site with almost 36k users. These are broken down by the Fixel engagement categories: Basic, Medium, and High, based on their likelihood to convert.

We see that 14k users met the Basic engagement threshold. Of these, 5k users met the High engagement threshold. When observing the delta between the two categories, we can see that 9k users have driven only 37k USD in Revenue. This means that our remarketing will likely be targeted only to the Highly engaged audience, as they are the ones likely to convert. At most, I would expand to the Medium engagement level for additional scale.

How much do you waste?

50% of all media budgets are wasted. Find out how much of your budget isn’t delivering the results you’re paying for.

Use cases for Engagement Based Audiences

Top of the Funnel (TOFU) Retargeting

The first and most obvious use case for these audiences is using these for retargeting campaigns. This is especially useful in removing low-quality users from our campaigns. As advertisers working on Facebook, we can confidently say that the CPM we’re seeing is constantly rising. With the holiday season around the corner, Black Friday, Cyber Monday & Christmas sales, media costs are expected to rise even further.

In this prime season, CPM can reach up to $100-200 for high-quality visitors. That’s a lot, and every impression counts. In an eCommerce scenario, what I would advise doing, is to focus the campaign beyond the default Dynamic Product Ads (DPA) audience of all users who have seen a product.

You can do this by clicking on the Advanced Setting button in the DPA setup, and narrow the targeting by selecting only users that belong to a certain audience. You can use the engagement thresholds you set up to create such audiences, e.g. Users who spent over 30 seconds or viewed over 5 pages. This will generally be a more permissive targeting that will still allow scale.

Statistically speaking, if you remove even 50% of users, which have a lower likelihood to convert, you can improve your campaign’s results by over 30%. This is crucial in the holiday season when campaigns run in short bursts and the platform doesn’t have enough any time to optimize. In this case, keeping the targeting at a high scale is critical for driving a mass of users to you site.

Bottom of the Funnel (BOFU) Retargeting

Similar to the previous tactic, our recommendation for advertisers that are sensitive to their Return on Ad Spend (ROAS), so focus only on Cart Abandonment audiences, is scale using engagement signals.

While focusing only on the Add to Cart users drives great ROAS, it usually lacks scale. So even if your ROAS hits 10, 20 or 50X, if there’s no scale it won’t drive substantial revenue to your store.

In this case, my advice would be to understand what your target ROAS is, and slowly scale towards it. Start with a high bar, i.e. users with a high engagement score that have also viewed a product, and create a DPA campaign to target these. If you see that these drive a ROAS that is higher than your target, expand your campaign to a wider audience until the ROAS aligns with your target.

Driving Scale in Retargeting

For advertisers that want to take a more aggressive approach, I recommend using the engagement signals to tier their remarketing campaigns.

The tiers can use both a bid cap and lookback window. In this case, users with a higher intent will receive a higher bid for a longer period of time, while low intent users receive a lower bid for a shorter period of time.

You can use multiple engagement signals to determine this breakdown (or Fixel’s engagement scoring) and assign each group the appropriate bid and lookback window.


In a previous session, our co-hosts from Zalster expanded on the importance of creating lifetime value (LTV) audiences in Facebook. I truly believe that this is an amazing feature that Facebook offers and in most cases have found this to be a high quality audience for my clients. One addition I would make to their tip is to cross the LTV with the category the user purchased from.

For example, if you’re selling sports shoes, the intent of someone who’s buying a pair of running shoes for himself is a completely different user than someone buying shoes for his kids.

These might be completely different people. The way they engage with your content is probably different. So distinguishing between these can lead to more accurate seed for your lookalike audience, assuming this refinement results in an audience that’s still large enough to create a seed from.

Facebook recommends at least 1000 users for a good seed, but can technically create such a seed off of 100 users.

If you lack such scale, I recommend creating the lookalike seed from highly engaged users crossed with the specific category they engaged with (e.g. Running Shoes). These can be created by selecting users of high engagement, such as Users with over 5 page views or Fixel High engagement.