Custom bidding overview and limitations

With a basic knowledge of Python and some help from Display & Video 360, you can build a custom bidding algorithm that will maximize impression values for you and your goals.

Custom Bidding lets you write a custom script that defines how much an impression is worth. Display & Video 360 then uses that script to build an algorithm that optimizes your bids for the highest performance based on your goal.

Use Cases 

Because Display & Video 360 also has automated bid strategies, custom bidding is best for optimization to metrics other than standard automated bidding goals. These include:

  • Custom Floodlight variables such as product SKU, basket size, etc. 
  • Floodlight Sales revenue from the revenue parameter being tracked by a Floodlight Sales tag.
  • Weighted conversions with different Floodlights tracking activities that carry different values to you (such as different activities on different product pages).
  • Conversion activities that occur infrequently. You may want drive volume by adding more weight to the conversion event and less weight to an upper-funnel event.
  • Key performance indicators for your brand such as viewability and video completions.
  • Linked Google Analytics 360 account goals. You may want to use goals created in a linked Google Analytics 360 account to apply different weights to events on the advertisers site.
    Note: You will need to first link the Google Analytics 360 account with the Display & Video 360 advertiser, then create goals within the Google Analytics 360 account. When linking a Google Analytics 360 account with Display & Video 360 for custom bidding, you are sharing conversion data, which will only be available in Display & Video 360 custom bidding metrics through reporting. Make sure the linked accounts are sharing data and have goals in place, scripts not using Google Analytics 360 goals will not score impressions.

Requirements and limitations

Data requirements

  • At least 1,000 positively scored impressions for an advertiser.
  • At least 50 positively scored impressions for an individual line item.
  • Individual impression values must be greater than zero, and in the range of 0.000001 and 1,000,000.
  • Google Analytics goals require clicks. Make sure there's a sufficient amount of data to generate a model.

Limitations

  • Custom bidding doesn't apply to YouTube, Gmail, Twitter, Audio, Tag Guaranteed, or Programmatic Guaranteed inventory.
  • Custom bidding can only have 10 active algorithms at a time (per Advertiser ID). Inactive algorithms can be archived.
  • Custom bidding requires prior scored impressions (within the advertiser) to train a model. You may have to wait several days before using Custom Bidding (as your bid strategy) if there are no past scored impressions to train the model.
  • Custom bidding models require a 30 day window of data. If an advertiser that uses custom bidding is paused for more than 30 days, the model will need to gather data again (the same as a new algorithm).
  • Custom bidding metrics can take up to 2 days to process. These metrics (including impressions with custom values, impressions with positive custom values and other derived metrics) may not be available for the last 2 days in your reports or in the Experiments tab.

Suspended algorithms

Algorithms will be suspended when all of the following happen:

  • The algorithm's script hasn't been updated or edited for 21 days.
  • There has been no spend associated with the algorithm for 21 days.
  • The algorithm isn't associated with any insertion orders or line items that have future budgets or flights.
Was this helpful?
How can we improve it?

Need more help?

Sign in for additional support options to quickly solve your issue