How DSP Optimizes Your Campaigns

This page outlines how the Advertising Cloud DSP optimization engine, which is powered by Adobe Sensei, optimizes the packages in your campaigns. For tips and tricks on how to manually optimize your campaigns, contact your Adobe account manager.

Package optimization goals operate at two levels:

  • For each package: DSP allocates budget to each placement within the package based on the placement’s performance against the selected KPI.

  • For each placement/auction in the package: DSP calculates the real-time economic KPI value for each auction per placement, and then uses this value to determine the bid.

    NOTE

    The economic value can be heavily weighted based on how well a placement is spending. If a placement is behind its spend goal, then it will be allowed to buy lower-quality auctions. If a placement is easily meeting its spend goal, then it will focus on higher quality auctions.

Package Optimization

DSP can optimize your delivery in two fundamental ways, with 20 variations available to align with your specific performance goal. You can choose to:

  • Prioritize the performance rate

  • Prioritize balancing cost efficiency with performance rate

See Optimization Goals and How to Use Them to determine which optimization goal will help you achieve your KPI(s).

Packages That Prioritize the Performance Rate

For optimization goals that prioritize the performance rate, DSP predicts the performance of each auction and always bids at the Max Bid. Examples of applicable optimization goals include Highest Viewability Rate, Highest Clickthrough Rate, and so on.

This optimization mode works well if:

  • You already know the effective/acceptable CPM level (for example, a historical benchmark).

  • You’re willing to manually adjust Max Bids if you experience challenges with scaling.

  • You’re prioritizing scale over efficiency.

Pacing logic

  • If spending is on pace, bidding will become more selective, so that you’ll only bid on auctions predicted to have high performance rates.

  • If spending is behind pace, bidding will become less selective, so that you’ll bid on auctions predicted to have lower performance rates in order to catch up to the pacing goal.

Clearing Price/Bid Shading

After it executes the pacing logic, DSP runs the proposed bid through a Clearing Price prediction model. If the prediction indicates that the bid can be lowered with minimal decrease to the win rate, then the bid is decremented per the prediction.

Packages That Prioritize Balancing Cost Efficiency with Performance Rate

For some optimization goals, DSP predicts the performance of each auction and adjusts bid prices automatically, never exceeding the Max Bid. Examples of applicable optimization goals include Lowest CPM, Lowest CPA, Lowest Cost per View, Lowest Cost per Click, and so on.

Pacing Logic

  • If spending is on pace, the system will become more price sensitive - bidding lower amounts to tradeoff win rate with the pacing plan

  • If a performance metric is also being balanced (all goals except Lowest CPM), the predicted KPI will be blended into the amount bid such that we bid higher for auctions predicted to be more performant on a “cost per” basis.

  • If spending is behind pace, then DSP will become less price sensitive - bidding higher amounts, up to the set Max Bid, to tradeoff win rate with the pacing plan.

  • If a performance metric is also being balanced (all goals except Lowest CPM), the predicted KPI will be blended into the amount bid such that we bid higher for auctions predicted to be more performant on a “cost per” basis.

Clearing Price/Bid Shading

After it executes the pacing logic, DSP runs the proposed bid through a Clearing Price prediction model. If the prediction indicates that the bid can be lowered with minimal decrease to the win rate, then the bid is decremented per the prediction.

Placement Optimization

Placement pre-bid filters are the strictest way to ensure strong performance. DSP uses pre-bid filters strategically across different ad types to achieve performance goals across placements within each package. You can use pre-bid filters concurrently with package-level optimization or independently.

NOTE

The available pre-bids filters vary by ad type. For example, for a standard display placement, you can filter by click-through rate and viewability but not by completion rate.

See Placement-level Pre-Bid Filters and How to Use Them to determine which pre-bid filter will help you achieve your KPI(s).

On this page

Adobe Summit Banner

A virtual event April 27-28.

Expand your skills and get inspired.

Register for free
Adobe Summit Banner

A virtual event April 27-28.

Expand your skills and get inspired.

Register for free
Adobe Maker Awards Banner

Time to shine!

Apply now for the 2021 Adobe Experience Maker Awards.

Apply now
Adobe Maker Awards Banner

Time to shine!

Apply now for the 2021 Adobe Experience Maker Awards.

Apply now