Facebook Guidelines

A guide to leveraging Faraday deliveries as custom Facebook audiences

In order to ensure the best performance and clear reporting, Faraday encourages clients to follow and refer to these guidelines when using Faraday audiences on Facebook. Ask your Account Strategist for the full guide, otherwise the overall best practices can be found in this article.

Overall best practices

  1. Audience Naming: the part in bold must remain the same, the part in italics can be changed.

    1. fdy.clients name.abcd1234.possible_persona

    2. We need “fdy” , the client name, and delivery ID to be able to do attribution.

  2. Setup structure: Faraday audiences tend to perform best when placed in their own campaign. Often times if a Faraday audience is in the same campaign as say a broad or a LAL that has been running for awhile, Facebook will put spend towards those audiences, without giving the Faraday one a chance to get out of the learning phase / succeed.

  3. Optimization: Faraday recommends that the “tracking event” of an ad set correspond to the conversion event used to build the Faraday model in use.

    1. For example: Traffic/brand awareness: Optimizing for impressions, unique reach, etc. Conversions: Optimizing for completed forms, add-to-carts, purchases, etc.

  4. Exclusions + Restrictions within Facebook

    1. Generally there is no need to add on additional exclusions or filters such as applying age, gender or even interest targeting.

      1. Please talk to your AM before adding on as restrictions to the model are normally best done at the delivery level within Faraday

  5. Stacking or combining audiences

    1. Faraday deliveries shouldn’t be stacked (e.g. 2+ Faraday custom audiences per ad set). This inhibits valuable performance reporting of specific Faraday audiences. Please consult your AM when thinking of stacking audiences

  6. Evaluate audience overlap

    1. If you’re using Faraday to for the pruposes of audience expansion via a custom audience, before launching any spend against this audience, confirm that the overlap between your Faraday audience and your existing Facebook audience(s) is low.

  7. Timeframe

    1. Audiences should have around two weeks without changes being made to the ad set and/or have gotten out of the learning phase.

  8. Deleting audiences

    1. Faraday ad sets should not be deleted. This destroys data and inhibits effective reporting

  9. Set up testing

    1. Facebook A/B testing suite is useful for evaluating the performance of your existing audiences vs. those optimized with Faraday predictions