Every campaign manager working on an ecommerce site strives to take a data driven approach in his campaigns. This is a common saying all of us will agree on. But what does that mean?
The average advertiser keeps valuable two tools in his toolbox: Remarketing and Lookalike audiences. Using these audiences, it’s relatively simple to generate a positive ROI over time. The above campaigner will set a threshold for the users to which he remarkets or creates lookalikes from.
This threshold can be a low or high one:
- A low threshold would exclude unengaged users, e.g., bounced users.
- A high threshold would focus on the highly engaged ones, e.g., cart abandonments.
Matching these behaviors to the products the user engaged with (product views) yields good results.
But the harsh truth is that this just doesn’t cut it any more.
Cleaning your lists
Most users coming to your site are irrelevant. Some studies show numbers as high as 60% of users being irrelevant to you. Bounced users are a good example for this, with most sites having a Bounce rate of 60-70%. But what if that user scrolled down the product page, viewed several images and shipping costs. Can they still be counted as irrelevant?
Tracking all the interactions described above is simple – simply add matching events to fire into your ad platforms. These, in turn, can be used to create various segments based on user actions.
But real-life interactions are complex. They use a measure across multiple different user actions, user attributes (device, language, browser, etc.) and their endless mixtures. So even you find the right “mix” of actions and attributes, these are usually hard to manage and tend to change over time.
Crèmè de la crèmè
But cleaning out the unengaged users is only half of the equation. Finding at your highest engaged users is just as important and uses a similar process. The classic target audience is users who added a product to their cart but didn’t complete a purchase (AKA Cart Abandonments). The downside to this tactic is that it overlooks a great number of engaged users, leaving you with only a fraction of your top users.
From action-based to intent-based targeting
Working daily with high spend campaigns we’ve experienced these exact pains. Instead of ranting, we decided to develop Fixel, a fully automated segmentation tool that ranks your website audience based on their level of engagement.
Using a powerful machine learning algorithm, we take out the guess work and manual labor from segmentation.
The result is a straightforward user engagement score within the ad platform: Unengaged, Basic, Medium or High. This additional layer of insight enables campaign managers to target their top audiences with the right message accurately.
The user score is set directly on the ad platforms (Google Adwords, Facebook, Twitter and more), and can be used in conjunction with other segmentations you already use.
Use case 1 – Remarketing Audiences
Most ecommerce sites use an “always-on” campaign targeting users with abandoned carts, serving them with items from their cart using Facebook’s Dynamic Product Ads (DPA) or Google’s Dynamic Remarketing. This usually is the best converting audience, which you can expect to drive the best Return on Ad Spend (ROAS).
But when looking at global benchmarks, we can see that the size of such an audience varies between 3-7% of users only. So even if the campaign to this audience works amazing, it’s still too small an audience to move the needle.
Using Fixel for remarketing
Working with many ecommerce sites, we have observed that switching the focus from Cart Abandonments to users that reached the High score, yields a similar ROAS while scaling the audience size as much as 10X.
Our suggestion: Create two separate ad sets, one for cart abandonments and another targeting users ranked “High”.
Use case 2 – Scaling to Mid-Funnel users
In some cases, scaling to users beyond the top percentile can still drive a positive ROI. Even on these broader campaigns, a minimal threshold of engagement can be useful to avoid careless ad spend on uninterested users.
Using Fixel for Mid-Funnel marketing
In this case, we suggest expanding your targeting to either all the visitors (broader) or all the product viewers (slightly narrower). These should be crossed with users ranking “Basic” and up.
This will allow you to reach a larger scale in your campaign while excluding low quality users.
Use case 3 – Lookalike Audiences
Creating Lookalike audiences is an excellent way to scale your user base with a relevant audience. But to create a good audience, a good seed is needed.
A good seed needs to answer three criteria’s:
- This seed should span over 500 users (Facebook recommends 1000 for best results at not more than 10,000).
- It should include users who expressed interest in a specific product. Users browsing through men’s shoes are different than users in your lingerie section.
- Finally, it’s recommended to target users as close as possible to the actual conversion.
Ideally, this would be a list of actual users that already purchased your product, by tracking their conversions or a manual upload. But not all advertisers have access to this data, nor do they have a sufficient purchase volume to segment their audience in this way.
Using Fixel for Lookalike seed
Using Fixel, it is simple to cross highly engaged users with specific products. This drives the volume needed while keeping the required purchase intent.
When creating the audiences for your lookalike seed you should match users ranking “High” with the URL of your product (notice to choose the “meet ALL of the following criteria.)
Using Fixel to improve your Facebook ecommcerce campaigns
Setting up Fixel audiences:
- Sign up to Fixel (it’s free for 30 days/100K sessions!)
- Create your Fixel Segment and install it on your site (learn more)
- Create Custom Audiences in Facebook based on the Fixel events. The audiences and event reflect the user’s engagement ranking.
Important note: These groups are inclusive by nature. This means that any user ranked High will also be part of the Basic group. So when running your campaigns use exclusions to narrow down a specific ranking group (e.g. Basic exclude Medium, will give all Basic users only)
Working with Fixel audiences in Dynamic Product Ads (DPA)
Fixel and Facebook Dynamic Product Ads work great together! All you have to do to use your Fixel audiences in these campaigns, is to click on the “Show Advanced Options” menu in the audience’s selection.
Then you can choose the audiences you created. Use Fixel “High” for best ROI and Fixel Basic/Medium to scale your remarketing audiences.