Facebook FAQ

Utilizing Faraday for your Facebook campaigns is key to finding new audiences and optimizing outreach

Overview

Leveraging Faraday's predictive models, we build custom audiences tailored to your company and business goals. With our direct connection to Facebook, you can easily create and deliver right from the Faraday app, sending custom audiences to your Facebook account.

How is a Faraday audience different than creating a lookalike right on Facebook?

Our audiences are developed using your first party and our third-party data to build a custom, predictive model focused on your business needs. Each model is reviewed by a member of our Data Science team and a full report detailing the model findings is shared, allowing an upfront, transparent process. In addition to this, Faraday knows every individual that is in the custom audience that is passed to Facebook which allows for a much richer attribution report, following leveraging the audience.

As a note, Faraday audiences are generated regardless of how active folks are on Facebook, meaning we are channel-agnostic. Due to this, it can sometimes take Facebook longer to find folks in our audiences as we are not just identifying those that are highly active on Facebook, but rather those that are the most likely to achieve the desired outcome.

How long should I run my Faraday audience? What if I am seeing high cost-per-result in the first couple of days?

Faraday audiences are channel-agnostic. It can sometimes take Facebook time to find the folks in the Faraday audiences, rather than advertising to the same active groups Facebook had previously found using their lookalike algorithms. It’s imperative that our audiences get the chance to make it out of the learning phase and ideally are live for a month. This will allow Facebook to begin recognizing the audience and those within it, as well as provide our Data Science and Analyst teams time to review the performance, resulting in reports and improvements to the predictive model.

A few key points to optimize performance:

• Make sure you’re setting the audiences up for success. Don’t share the budget in a campaign with competing adsets, as Facebook will always favor the historical example of strong performance, not allowing the Faraday audience a chance.

• Don’t add or apply targeting to our audiences, as this will reduce the size of the audience. Our models already take factors like demographics and location into account when creating the audience.

Are there exclusions or additional filters that I need to apply to a Faraday audience on Facebook?

Nope! Any exclusions or filters that you would select on Facebook should be selected in the Faraday system, as that way we provide the best audience before it gets to Facebook, and won’t reduce the size of the custom audience. As a note, our predictive models take our 300+ attributes into account when generating the custom audience.

What is the difference between a Faraday custom audience, and a Faraday audience for a seed in a Facebook lookalike?

A custom audience from Faraday is built leveraging our predictive models that are built specifically for you and your business needs. These custom audiences factor in your first-party data, our third-party data, and predictive modeling technology to produce high propensity audiences for optimal targeting. Any audience built using a predictive model should generally not be used as a "seed" to then create a lookalike on Facebook.

A Faraday audience seed is a list built using the Faraday system based upon segmentation and or personas. In these instances, a predictive model is not being leveraged but rather segments selected by you are used. When it comes to personas, depending on the size of your customer base, typically it makes sense to pass over a list of your customers broken out by which persona they’re in, and allow Facebook to develop the lookalike. Small, niche audiences on Facebook tend to have trouble performing, so allowing Facebook to use their algorithm once Faraday has built personas is the best of both worlds!

How do I use the Faraday audience as a seed for a Facebook lookalike?

If you’ve determined that you’ll be using the seed method, either focused on segmentation or personas from Faraday, turning that audience into a Facebook lookalike is easy. Simply create the delivery you want in the Faraday app, send it to your Facebook account, and let your Account Manager know. From there, the Account Manager will build the lookalike right on your Facebook account.

When should I use a Faraday audience seed for a Facebook lookalike (and how many should I include)?

The size can vary depending on how the seed is being built, i.e. what segments or personas you’re seeking. It’s best to consult with your Account Manager if you’re unsure.

How do I know when a Faraday audience is fatigued?

After a Faraday audience has been running for a number of weeks, sometimes you’ll notice the performance starts to slow -- this is what we call an audience being fatigued. A few keys signs of this include:

• The audience has been running the audience for well over 4 weeks

• The estimated reach is nearing the size of the total audience

• The frequency has increased over the last few days

How do I refresh a Faraday audience after it’s fatigued?

The best option is to set up a recurring cadence with your Account Manager to update the predictive model(s) you are leveraging, usually around every 4-6 weeks.

Alert your Account Manager that the audience appears to be fatiguing, and the Account Manager will work with our Data Science team to update the predictive model to ensure we’re optimizing based on the most recent customers and results. From there, you’ll be able to create a new delivery from the Faraday app and push it right back into Facebook.

Can I use multiple Faraday audiences at a time on Facebook?

You can, but make sure you’re thinking about how the audiences you’re using are similar or different in terms of overlap. Be sure to implement proper exclusions and when in doubt, check with your Account Manager to further strategize.