SPARC User Guide

SPARC, powered by Nativo Predictive Audiences, is Nativo’s proprietary, privacy-compliant solution that uses engagement data from our exclusive content offering to build an audience model based on behavior. Combining proprietary data, contextual signals, and machine learning, our solution finds and delivers new users most likely to engage with an ad.

Campaigns leveraging SPARC have achieved 2x greater click-through rate, 2x greater click-through conversions, and up to 50% lower cost-per-acquisition.

We offer several options when targeting with SPARC:
Off-the-Shelf Audiences - ready-to-use segments derived from activity across Nativo’s network
Custom Audiences - segments tailored to each advertiser using data derived solely from their campaign data
Custom-Built Audiences (coming soon) - a customizable group of segments unique to your organization using your first-party data

Audience Extension - Drive incremental revenue using hundreds Off-the-Shelf Audiences.

Direct Deals - Leverage SPARC with your direct deal campaigns to deliver better performance without cookies to advertiser partners.

Step 1:
Click into SPARC Audiences from the “Campaigns” tab on the left-hand menu. Next, click the “Create SPARC Audiences” button at the top-right corner.

To build the most comprehensive SPARC custom audience, it's recommended to maximize inventory availability by keeping targeting filters open as much as possible when setting up your campaign budget.

Step 2:
Type in the name of the SPARC Audience you’re creating, a description, and select the advertiser.

Step 3:
Select the desired signals for your seed audience from the drop down. The signal chosen acts as a grouping for building a seed audience.

If you use clicks, we will evaluate the behavior and identity signals for the users who registered a click. If engagement is selected, the seed audience generated is based on the users who had 15 seconds of engagement.

Choose between...

Step 4:
Select the campaign or budget you’d like to use. You can mix and match between budgets.

SPARC Custom Audiences can be established only upon meeting specific minimum signal thresholds: 2,500 for campaign and budgets and 8,000 for advertiser-based SPARC Custom Audiences. If a SPARC Custom Audience is created before reaching these thresholds, the audience will result in a 'failed' status.

Step 5:
Before saving, review your SPARC Custom Audience details in the right-hand box.

Step 6:
​

Click the “Save” button in the upper-right corner and the new audience will be generated. When the audience is ready, the status will change from "Processing" to "Ready."

Note: this may take a few hours

SPARC Custom Audience Status Definitions:

Ready: Audience has successfully been created and is ready to be applied to campaigns.

Pending: Audience is in queue to be processed.

Processing: The system is processing the audience, and this can take a few hours.

Failed: The audience creation has failed, and this likely can be due to minimum signal thresholds not being met.

Once changes are made to a SPARC Custom Audience, such as adding or editing sources and signals, and saved, the system will reprocess and refresh the audience segment. The system generally will take 24 hours to reprocess and refresh.

Step 1:
Click into “Campaigns” from the left-hand menu, and select the campaign you’d like to add SPARC targeting to. Once there, click into the budget settings.

Step 2:
Click “Targeting” from the tabs at the top, and then click “Audience” from the left-hand menu.

Step 3:
Within the audience library, click on the icon indicating SPARC Audiences.

Step 4:
Search for the SPARC Audience you’d like applied to your campaign. Both Off-the-Shelf and Custom Audiences (once ready) will appear here. You can add as many audiences as you’d like.

Step 5:
On the right, you can review all targeting, including SPARC Audiences, that have been added to the campaign. Below that is an inventory forecasting tool that will show you the estimated reach with the targeting criteria applied.

If you’d like to increase reach, you may need to remove some targeting filters.

Step 6:
Click the “Save” button in the upper-right corner before moving on.

How does SPARC work?
Our solution uses machine learning to find audiences similar to a seed audience known to be valuable to an advertiser. That audience can be clickers/converters from a given campaign, users visiting specific types of content, etc. The model looks at several attributes in the source audience such as demographics, browser and device type, and site context. We also tie in a wealth of vertical-specific data across our network.

The output of this model is a set of parameters our Ad Server can use to identify users similar to the seed audience. These parameters are packaged together into a “cookieless audience” targetable within the Nativo Ad Platform.

What is a seed audience?
A seed audience is simply a set of users that we know saw a brand’s ad and completed the brand’s KPI, e.g. a user who clicked on an ad from a campaign with a CTR KPI, a user that Engaged with a CLP for a Native Article campaign with an Engagement-based KPI, a user that filled out a quiz/lead form for a campaign optimizing towards leads/conversions.

This audience acts as the source for which we build predictive audiences based on

similar attributes.

What signals are collected to create SPARC Audiences?
Context-based: Placement category (IAB), Placement Location (Main well, MOAP, etc.)
Device-based: Device Type, Browser Family, Browser Version, Browser Language
Location/Demographics based: Zip (plus census data aggregated at the zip level), ISP
Time-based: Day-of-week, time-of-day

How is census data used?
We use all available signals in census data. Many variables are the amount of population in a zip code within a specific age range broken down by gender. Those values are aggregate/total values, and not averages.
We use medians. So for gross housing rent within a zip code, that value from census data is the median housing rent in that zip code, not the average.

Is SPARC fully privacy-compliant & free of personal identifiers if it uses census data?
The census data we use is not specific to an individual user - it is representative data at the zip-code level. We do have information on race, but we would never know if a specific user is actually a specific race. Instead we would know the percentage of people in a zip code of a given race.

How do we know if a SPARC Audience is performing?
​Go to “Reporting” as you normally would. Under the audience filter breakdown you’ll be able to view performance by audience.

If you think you'd like to utilize SPARC audiences and want to learn more, please contact your Account Representative.