All Eyes on DV360: Building Your Own Personal Algorithm

If you’re familiar with DV360 you’ll know that it has a selection of automated bid strategies which allow programmatic traders to go beyond applying a fixed bid strategy. Traders can use the system to bid on impressions based on its likelihood to achieve a chosen KPI. But what if a trader wants to actively factor in a variable that isn’t considered in Google’s automated algorithms, such as variable conversion values, as opposed to all conversions being considered equal? The answer? DV360’s Custom Bidding.

Nickesh Patel

What is Custom Bidding?

Custom bidding is DV360’s outcome-based buying feature which goes beyond the standard automated bidding strategies, wherein traders assign values to impression data points. DV360 then looks at that advertiser’s historical data and uses machine learning to predict the value of each incoming impression in order to bid accordingly.

The most common use case is the variable weighting of conversions and for that DV360 has a feature to streamline this process: Goal Builder.

What is Goal Builder?

Goal Builder is an easy-to-use module within the interface which allows a trader to select Floodlight activities and assign a value to them.

Assigning a level of ‘importance’ to data points numerically is a key concept of understanding Custom Bidding. E.g Assigning Floodlight Activity ‘X’ a weighting of ‘5’ and Floodlight Activity ‘Y’ a weighting of ‘10’ will cause DV360 to treat Activity ‘Y’ as twice more valuable than Activity ‘X’. Therefore if an incoming impression is more likely to lead to an Activity ‘Y’ conversion, DV360 will adjust its bid accordingly.

Currently, traders can only assign weightings to Floodlight activities and GA Goals in Goal Builder. Once a trader becomes more comfortable with this form of optimizing, they may want to start scoring impressions based on other parameters, such as geo location or device type. DV360’s has a feature to enable this: Custom Script.

What is Custom Script?

Before trying to write a script from scratch, it is useful to understand how the syntax is structured, here is a simplified example of how to layout a script.

In this sample script, if criteria_x and criteria_y are both true, a score of 5 is returned and if criteria_z is true a score of 10 is returned. Given this falls under a sum_aggregate function, if all criteria are true the total score will be 15.

To help develop a better understanding of how certain syntax can be applied, Google has created several sample scripts that suit the requirements of certain real world scenarios. For example, there are sample scripts for:

  • A car-rental chain that wants to optimize for the number of vehicle reservations. The reservation confirmation activity is tracked in a Floodlight and the number of nights booked is in a Custom Floodlight variable.
  • An e-commerce advertiser that wants to run a campaign to launch a new product. The campaign’s KPIs focus on maximising brand, so they want to optimize toward metrics that measure brand exposure to their users.
  • A car manufacturer is tracking conversions on three different product pages: minivan, SUV, and hatchback.  Each range is assigned different values to allow optimisations towards the most popular model

Once traders understand how each sample script satisfies the requirements of the scenarios they can move on to modifying the samples to suit the specifics of their own real world campaign objectives where eventually it will become more convenient to make the transition towards writing their own scripts from scratch.

When it comes to bidding strategies there are many steps that take you from fixed bidding to writing a custom bidding script with no assistance. If you need help in navigating this journey, we at Incubeta are here to help. Whatever stage you’re at, get in touch and let Incubeta take you to the next level.


Read Next

Find out how we can help you

With offices around the world, we can build a team perfect for your needs.

Keep up to date with our monthly insight digest