How to keep opted-out people out of every campaign with one dataset

Send your do-not-contact list to Faraday as its own dataset, set its privacy option, and Faraday leaves those people out of every pipeline and deployment in your account.

How to keep opted-out people out of every campaign with one dataset
Andy Rossmeissl
Andy Rossmeissl
on
3 min read

To keep people who opted out of your marketing out of every Faraday campaign, upload your opt-out list as its own dataset and set its privacy option to Suppress. Faraday then leaves those people out of every pipeline and deployment in your account. You don't have to remember to exclude them each time you build a new audience.

This post covers why an account-wide list beats per-campaign exclusions, the difference between suppressing and deleting, and how to set one up.

Why opt-outs belong in one place

Most brands collect opt-outs in several places: unsubscribes from the email platform, do-not-call requests from the call center, do-not-mail requests from customer service, and opt-out-of-sale requests under state privacy laws. Each one is a promise to a person that you won't contact them again.

The usual way to keep that promise in a marketing tool is an exclusion on each campaign. That works until someone builds a new audience and forgets the exclusion, or excludes last quarter's copy of the list. A single list that applies to the whole account removes that step. Every pipeline you build, now or later, leaves those people out.

Suppress or delete?

A dataset's privacy option has two settings, and they answer different requests.

SettingUsed for modeling?Delivered in pipelines and deployments?Use it for
SuppressYesNoDo-not-contact requests: unsubscribes, do-not-mail, do-not-call, opt-outs of sale.
DeleteNoNoDeletion requests: people whose data you may no longer use at all.

Suppress is the right choice for most opt-out lists. Someone who unsubscribed from your emails is still a customer, and what they bought still helps your models learn who your best customers are. Faraday keeps them in cohorts and model training, and leaves them out of anything it delivers for you to contact.

Delete is for people who asked you to stop using their data entirely. Faraday removes them from your other datasets and cohorts as well, so they never reach a model.

How to set it up

  1. Collect your opt-outs in one place. Include every identifier you have for each person: email, phone, name and postal address. Hashed emails and phone numbers work too. A CSV file is fine to start with. To keep the list current without uploading files by hand, have your systems write opt-outs to a cloud storage bucket or a warehouse or database table, and connect Faraday to it.
  2. Create a dataset from that file or connection and map its identifiers, the same way you would for a customer list.
  3. Expand Advanced options and choose Suppress or Delete under Privacy.

Screenshot of the privacy option in a dataset's advanced options, set to Suppress

Once the dataset finishes processing, pipelines and deployments that run afterward leave out everyone Faraday matched in it. A dataset on a connection picks up new opt-outs each time it refreshes. If you're still uploading files, add new opt-outs to the same dataset. Either way, each update rebuilds the list.

If you work through the API, set privacy to suppress or delete in the create dataset or update dataset request.

A few things to know

  • Faraday can only keep out people it can match. An opt-out list matches the same way any other dataset does, so a row with an email, a name and a postal address matches more often than an email alone.
  • Don't set the privacy option on a customer or order dataset. Everyone in it would be kept out. Opt-outs need a dataset of their own.
  • This is not a pipeline exclusion. A pipeline's exclusion settings still have their place: leaving existing customers out of an acquisition campaign, or skipping addresses you mailed last month. Those are choices about one campaign. An opt-out list records what a person asked for, so it applies everywhere.
  • Managing sub-accounts? If you want one do-not-contact list to cover all of them, contact support.

For the full reference, including an example API request, see opt-out and do-not-contact datasets in the docs.

Andy Rossmeissl

Andy Rossmeissl

Andy Rossmeissl is Faraday’s CEO and leads the product team in building the world’s leading context platform. An expert in the application of data analysis and machine learning to difficult business challenges, Andy has been running technology startups for almost 20 years. He attended Middlebury College and lives with his wife in Vermont where he lifts weights, makes music, and plays Magic: the Gathering.

Ready for easy AI?

Skip the ML struggle and focus on your downstream application. We have built-in demographic data so you can get started with just your PII.