Strategies for Optimizing Custom Audiences

This document provides you with in-depth technical information on strategies for working with custom audiences.

A custom audience is defined as a group of users with similar web browsing or buying habits.

Once you create an audience, you can include that audience in any new campaigns you create. Likewise, you can also exclude that audience from any new campaigns.

Overview

Using Verizon Media’s Custom Audience API, you can better target your ads by creating audiences from your website, from your mobile app or from your customer or email lists.

The API lets you create, read and update tag endpoints using standard POST, GET and PUT calls. Audiences can be defined using Apple’s Advertising Identifier (IDFA), or Google’s Advertising ID (GPADVID).

Audiences by Dot Tags

Dot provides you with a snippet of JavaScript code that when placed on a website, sends event data to Verizon Media for use with your advertising campaigns. The event data may be either through standard or custom implementation of DOT tagging.

In the standard Dot implementation, you place the tag in the <head> of every page on your website to transmit standard HTTP header information for each page a user visits: most importantly, the URL of the page.

The more advanced option lets you send custom event parameters that transmit more complex event data. You can then use the event data collected through your Dot tag to define specific conversion rules or retargeting segments.

Important

With the exception of assigning a DOT tag to a campaign, the Native Ad Platform API provides you with all of the functionality available in the Native Ad Platform UI.

Pros & Cons

Following are some of the pros and cons of implementing DOT tags for custom audiences.

Pros

Cons

Once the DOT tag is deployed, it’s easy to create custom rules to isolate audiences that visited specific pages or took specific actions all in one place.

DOT tags rely on cookies, which are not a stable identifier. For activities over 30 days in the past, DOT tags tend to lose scale quite significantly.

Data is activated in near real-time, so there is no latency between when users visit the site and when they are added to segments created from DOT tags.

In any environment where cookies are blocked (for example, on Safari or FireFox browsers), DOT tags f won’t be effective.

For advanced use cases (for example, tracking custom events, dynamic values and so on), DOT tags require custom configurations of the tag, which may take extra time.

Important

You can only build website audiences based on your own DOT tags. You can’t build audiences based on the DOT tags that belong to your competitors, for example.


For more information, refer to How to utilize Dot tags and Verizon Media Dot tags.

Audiences from Email Lists

Audiences from your email lists allow you to define and target audience segments of people who are on your mailing lists. Once created, these segments will be available for targeting, using the TargetingAttribute service.

Pros & Cons

Following are some of the pros and cons of using custom audiences by email lists:

Pros

Cons

Email addresses are powerful because they are one of the strongest and most long-lived identifiers. Users rarely change their email address, so emails can easily be used to target or exclude users based on activities performed several months in the past, if needed.

While Native Ad Platform has very good match rates, a drop has to be expected. Depending on the advertiser and country, match rates may vary between 30% and 50%.

Yahoo mail, Verizon Media’s email app, provides you with multiple options. With Yahoo mail, there are over different 1B email addresses you can match against, and even beyond Yahoo.com, you can use the Yahoo mail app with non-Yahoo addresses. Focusing on commerce for email, for example, you have a higher chance that the emails, even if they might not be the primary addresses of some users, are still very likely being used for purchases, subscriptions, and so on.

While Ad Platform offers API integrations, advertisers that don’t want to integrate with the Native API might find the process to keep audiences up to date manually-intensive, especially when compared to DOT tag-based audiences.

Lead generation ads that run on mail enable you to run ad campaigns to collect emails, and then deploy them as custom audiences.

Because Verizon Media doesn’t store the email address files in any form, old email custom audiences might not benefit from Native Ad Platform’s ability to match more email addresses from new users who signed up for your products, unless advertisers re-upload their files again.

For more information, refer to Audiences from email lists.

Audiences by Device ID

Custom audiences by device ID let you target your ads to a specific set of people with whom you have already established a relationship.

Audiences can be defined using Apple’s Advertising Identifier (IDFA), or Google’s advertising ID (GPADVID). Once created, these segments will be available for targeting using the TargetingAttribute service. A user will remain in an audience for 30 days from the moment their device was added, unless removed earlier using the API.

Pros & Cons

Following are some of the pros and cons of using custom audiences by device ID:

Pros

Cons

Both Apple and Google device IDs are universal, which means there is no drop when these files are onboarded in Native Ad Platform, and the match-rate should be considered 100%.

Device IDs are also more stable and long-lived than cookies (but less than email). While users have the ability to churn or reset them on their device, very few do. So in practice, device IDs change only when users change their phone, which is a fairly rare event.

Similar to emails, while Native Ad Platform does offer APIs to automatically onboard new device IDs, advertisers who don’t want to go down this route might find it manually-intensive to keep them updated.

Easy to collect for all users. Each device has a device ID that can be collected and used for targeting (with proper consent).

For more information, refer to Audiences from customer lists.

Best Practices

Following are some of the best practices recommended by the Verizon Media team:

On Excluding Converters

  1. Use ALL types. Most advertisers have at least two (for example, e-commerce will have DOT tag and email, while app developers will have device IDs and emails). Given their different pros and cons, they are quite complementary.

  2. For app install advertisers, once the integration with the measurement partners for conversion tracking is completed, all subsequent installers are automatically excluded by Native Ad Platform without the need to use custom audiences. It’s recommended to use custom audiences to exclude any user that installed the app before the integration was completed.

On Retargeting

The “waterfall” strategy is very effective for many advertisers. This consists in breaking down the whole retargeting pool into (ideally) mutually exclusive, and jointly exhaustive cohorts of users, and targeting them on different ad-sets with different creatives, bids, and so on.

For example, an e-commerce advertiser could build this type of waterfall:

  1. Include all past converters with only one purchase.

  2. Include users who started the checkout process, and exclude all converters.

  3. Include users who added at least one product to their basket, and exclude the ones who started the checkout process already.

  4. Include users who visited at least one product page, but exclude the ones who added at least one product to their basket.

  5. Include all users who visited the site, but didn’t visit the product pages.


Each bucket will represent a very unique but homogeneous cohort of users, which can and should be treated differently. For example, an advertiser may want to use tailored creatives for the lower funnel audiences, like types (2) and (3) and promote a coupon or a discount to incentivize the conversion.

On the other hand, focusing more on the brand and the product for upper funnel audiences, like types (4) and (5), is highly recommended.

On Prospecting

  1. Keep in mind that the DOT tag only has information on users from the moment it gets deployed, and for app install campaigns. Native Ad Platform receives (and automatically excludes) installers only from that moment onwards. Advertisers setting up new campaigns for the first time should always try to exclude existing users by leveraging all custom audiences, especially the ones that look back in time (for example, device IDs and emails). In this way, you can minimize the chances that an existing user will see these ads.

  2. Lookalike audiences allow you to find more people similar to your existing customers. A lookalike audience uses an existing custom audience segment as a “seed” and expands that seed using Verizon Media data.

Lookalikes are the best prospecting audience strategy, because they are created ad hoc for a specific use case. We recommend always using the most valuable seed, if available at scale. For example, an app developer should use as a seed not all their app users, but only the most valuable based on their KPIs (for example, engagement, lifetime value, and so on). For web and e-commerce advertisers, it may be easier to do this from emails, given that qualifying users into a high quality seed may take several months.

For more information, refer to Lookalike Audiences.

  1. If the seed doesn’t have sufficient scale, we recommend that you pick an event that is upper funnel versus the conversion, but still highly correlated to it. For example, an e-commerce advertiser with few conversions might use the add to cart event or a visit to the beginning of the check-out as the seed to expand on. We highly recommend to explore this strategy for smaller advertisers, advertisers operating in smaller markets, or brand new advertisers with limited historical data available to them.