DSP can optimize your performance-focused campaigns for placements with the lowest cost per acquisition (CPA) or the highest return on ad spend (ROAS).
See the following best practices for performance campaigns:
It’s important to understand whether the goal of the campaign is to achieve the highest possible ROAS or the lowest possible CPA. For each package in the campaign, you’ll specify the objective goal accordingly as either Highest ROAS - Custom Goal or Lowest CPA - Custom Goal.
You also need to determine the success event(s) that will lead to the overall goal and create custom goals accordingly. For each package, you’ll specify a custom goal to be used with the overall optimization goal for reporting and algorithmic optimization using Adobe Sensei. For more information about creating custom goals, see the Best Practices for Building a Custom Goal.
Upper funnel packages include placements with very broad targeting to reach net new consumers.
Find new audiences that are likely to convert using the following tactics:
Use run of network (RON) targeting: It’s important to include a run of network placement without audience targeting and with broad inventory targeting. This allows the Adobe Sensei algorithm to find valuable users who may have newer cookies that haven’t yet been categorized into an audience.
Lower funnel packages include placements that target users who have already been to the advertiser’s webpage or for whom the advertiser has CRM data.
If the advertiser is an Adobe Analytics or Adobe Audience Manager customer, then you can build first-party segments (such as homepage visitors, product pages, or cart abandoners) and use them as placement targets in DSP.
Avoid assigning too much budget to an audience-targeted placement. As a general rule, budget $30 per 1,000 users per month.
The best practice is to create separate packages for upper funnel prospecting and for lower funnel retargeting. Optimization occurs at the package level, so that performance data from all placements within a package is pooled. Therefore, group placements into packages with similar expected performance.
Also, use the following settings.
Pacing & Capping: To select a CPA or ROAS optimization goal, the package must use package-level pacing. This ensures that all placements within the package are optimized to distribute spend based on performance and scale to the selected goals.
Flight Dates: (Prospecting packages) When your campaign runs for longer than 25 days, use the Activate Custom Flighting feature. First, set a custom flight for the first 10 days at approximately 75% of the necessary daily budget to reduce spending during the learning phase. Then set a second custom flight for the remainder of the budget.
For example, if you have $100,000 to spend in 30 days, then set the budget for Flight 1 (Days 1-10) to $25,000 (75% x $100,000/30 days = $2,500 per day). Use the remaining budget of $75,000 for Flight 2 (Days 11–30).
Budget: DSP will always try to allocate 100% of the package budget evenly between all placements in a package. If a placement has low spend or no spend, we recommend budget capping the placement to allow more of the budget to allocate to placements with scale. Allow 24-48 hours for budget changes to calibrate.
Optimization Goals: Use one of the two performance optimization goals, Highest ROAS or Lowest CPA, depending on the package goal. These goals auto-optimize the package towards the Highest ROAS or Lowest CPA placements, respectively.
Flight Pacing and Intraday Pacing: For both types of pacing, select Even to maximize your performance goals by pacing uniformly throughout each day and throughout the entire flight.
Use FrontLoad and Aggressive Front Load for flight pacing and ASAP pacing for intraday pacing only when you are fully prioritizing delivery and spend over performance optimization because those strategies can negatively impact your desired performance KPIs.
Less is more. If you can set up fewer than six placements per package, then the available budget can dynamically shift to the best performing placements most easily.
Also, make sure to add each placement to a package with a package goal type of Prospecting or Retargeting, as appropriate.
The following are the recommended placement settings for performance campaigns.
You’ll configure CPA or ROAS optimization at the package level (see Step 3 - Create Packages), but you can add additional placement-level settings.
Pre-bid Filters: Minimize, or ideally avoid, setting aggressive pre-bid filters, which prevent the placement from achieving scale. Best practices include the following:
Use one (1) pre-bid filter per placement. Multiple pre-bid filters will require that both are met, which reduces scale.
Consider setting less strict pre-bid filters in cases where additional targeting (such as audience, geo, and site targeting) is applied.
See descriptions of when to use each pre-bid filter at Placement-level Pre-Bid Filters and How to Use Them.
To maximize scale, use Public (Open Exchange) and On Demand inventory.
Your ads will perform best if a user can be reached by only one placement. Significant overlap in users across placements can cause competition, which yields a cycle of continually increasing bids, driving up the cost per user. Therefore, if you include multiple audiences, make sure they don’t consist of overlapping users/audience members.
You can avoid overlapping audiences by creating your audiences in tiers so that you can suppress the higher, more inclusive tiers from placements as needed.
Using contextual filtering, pre-bid fraud blocking, and/or Ads.txt filtering will limit the scale of your placements, but use them if needed.