Best Practices: Targeting: CPA and ROAS

The Quantcast Platform is built to drive campaign performance with minimal need for manual optimizations. We have developed this guide to provide some best practices when it comes to setting up and editing campaigns with cost per acquisition (CPA) and return on ad spend (ROAS) objectives. 

Conversion Events

It is always a best practice to tag and define your conversion event as the user action closest resembling a conversion.  For example:

• If your goal is to drive purchases, tag the purchase confirmation page

• If your goal is a lead form, tag the form complete page or the form submit button

If your campaign conversion has not collected the minimum amount of data required to populate in event dropdown list when building your campaign, we advise using an event slightly up the funnel that has more traffic. 

Full Performance

Full Performance (both prospecting and retargeting) will generally drive your best CPA/ROAS performance in the platform. This tactic allows Quantcast’s proprietary algorithms to adjust the balance of tactics in real time between prospecting new customers and retargeting visitors to achieve the best overall performance possible.

Lookalike Events

The ideal setup for the best prospecting performance is for the lookalike event to match the conversion event. This allows the system to model off of your converters. The only time we recommend using additional events on top of your conversion event is as follows:

• If your conversion event gets <10 conversions on average per day, consider testing adding the page next highest in the conversion funnel. (i.e. A retail company driving purchases including the shopping cart page as a secondary event if there are < 10 purchases per day).

• If you have an event on an even deeper funnel page than your conversion event (i.e. an insurance company driving to quotes may want to model off people who go one step further and complete a policy)

Retargeting

It is best practice to tag as much of your website as possible and then create rules to break out key pages in the conversion funnel. The article here walks through the process of creating rule events. 

• The Quantcast Platform's retargeting models will look at the behavior of visitors across your website based on the labels you have defined in Q Pixel. Retargeting will dynamically bid more on visitors of label patterns more likely to convert and bid less on label patterns less likely to convert.

Geography

Target the widest area possible at all times to see the best CPA or ROAS. Restricting geography on a campaign almost always lowers performance. Try expanding your geo to allow the targeting models to dynamically find the most likely converters. Experiment with the following changes that loosen geographic restrictions:

• Zip code → City

• City → Metro Area

• Metro Area → State

• State → Country

Demographics + Interest Audience

If CPA or ROAS is your ultimate goal, avoid layering on demographics and interest audiences when first running your campaign. Demographic optimization will push delivery toward the attributes you select while trying to balance the likelihood of conversion. Interest audiences filter delivery down to the people who are interested in that content.

Quantcast’s default prospecting (lookalike) and retargeting models will look for users across the web who have similar behaviors, interests, and demographics to the converters on your website as defined by the conversion event. There is no need to layer on demographics or interest audiences when starting out.

• Do test running a second ad set layering on additional demographics or interest audiences. The new ad set may not have as strong of a CPA or ROAS as your ad set without demographics or interests, but it may expand the reach of your campaign to a new audience. (e.g. if your product is typically purchased by younger females, run a second campaign with a demographic overlay of older males -- Don’t forget to have a creative that will also resonate with that new audience!)

• Use the Audiences tool of the platform to understand how your website audience overlaps with a particular demographic or interest audience. This tool can be leveraged to see the targetable audience size of pixel events combined with a particular Interest Audience or demographic filter. 

To maximize CPA/ROAS avoid limiting targeting restrictions

Target all devices - finding the right person who is ready to convert means being able to target them wherever they are and on whatever device they are using. By limiting targeting to specific devices, you may impact the ability to reach all people who may convert on your campaign.

• Viewability - keep your viewability target realistic.  Increasing the viewability goal lowers the number of bid opportunities available, which in turn may have a negative impact on performance.

• Site lists -  keep your site lists broad.  Limiting allowlists to just a few sites or blocking too many sites could negatively impact performance. Our models are looking to reach the right person at the right time, so blocking inventory may limit the ability to reach the right internet users wherever they are across the internet.

• Frequency Caps - Avoid restricting delivery wherever appropriate. Ad sets with a strict frequency cap can prevent an ad from being shown to a user at the right time to get their attention when they are ready to convert. 

Experiment

Use the experiments feature to test a hypothesis. This feature splits the targetable audience into two mutually exclusive groups, so you can test a different targeting tactic simply and effectively. Test, learn, and apply those learnings to your whole campaign moving forward. Here are a few ideas of tests you can run:

• Two different sets of creatives with different messages. 

• Two different site lists. On one experimental ad set, try the default allowlist. On the other ad set, try a blacklist you have uploaded. 

• Try running a lookalike model off your conversion event vs. a page deeper in your funnel or adding in a secondary event.

• Run two different interests audiences against each other to see which performs better.