2022-05-31 | Article | Insights
One of the key benefits of programmatic buying is the ability to buy media based on data in a user-centric way and optimize campaigns across different publishers, deal types, and ad slots in near real-time, without committing to a fixed spend allocated on a publisher-by-publisher basis. The establishment of optimization algorithms in demand-side platforms in recent years has led to a further leap in the quality and efficiency of campaign execution in this area, using machine learning to enable greater compute power and more sophisticated analysis. However, the standard algorithms optimize standard KPIs based on the data available in the DSP, which doesn't necessarily relate to customer profitability. The feature Custom Bidding in Display & Video 360 (DV360) is here to help.
Custom Bidding lets you determine how much an impression is worth for your campaign goal. DV360 then uses your insights to create an algorithm that optimizes your bids for the highest performance based on that goal. There are two ways to utilize Custom Bidding - either by setting up a script based on simple Python constructs or by using the Goal builder within the user interface.
Measuring margins to determine customer profitability is critical to continued business success because it tells you whether certain customers are costing you money instead of making you money. A high-level glance at your business focusing on revenue might be deceptive, leading you to believe that some customer groups are more attractive than they actually are. Armed with the revenue data alone, you may shift your strategy, budgets, and acquisition efforts to cater more to Segment B in the following example, although including cost data shows that Customer Segment A is far more profitable.
Once you have a framework in place to measure the margin, it becomes easy to use it for campaign optimization utilizing the Custom Bidding feature.
You can build the optimization model based on your margin leveraging Google Cloud BigQuery and Google Analytics and optimize your campaigns in DV360 based on margin clusters instead of the revenue values. The first step to start with is to ingest your data from your CRM/ e-commerce solution and your GMP accounts in Google Cloud BigQuery and process it using Dataproc. You can connect the data sources based on the order id so you can calculate the margin via BigQuery Jobs or with BigQuery View. Then you can visualize the results with Looker Studio. By using user data import and the Management API you can enrich your transactions in Google Analytics by storing the calculated margin values as a custom dimension and adding the values to the hits that have already been processed to then create audiences representing different profitability clusters. Please note that this process is for Universal Analytics only.
After mapping your margin data to your transactions in GA, you can build audience clusters in GA based on the margin amount such as audiences of users with top 10%, top 20% etc. highest-margin transactions based on products purchased/ service enquiries made considering all costs including discounts granted within the acquisition process.
After defining and building those audiences, you can set up a unique event to be fired with each specific audience cluster. Then you can define a score for the events and use them in your Custom Bidding algorithm. The higher the calculated margin of the transactions of the customers within the given audience, the higher the score assigned with the fired event associated with that same audience.
Striving to improve customer profitability is an ongoing endeavor. Once you have determined the data to use and built a structure, it should be easy to continually update that data as your business evolves. By taking the time to evaluate different profitability clusters and use them for campaign optimization purposes with Custom Bidding, you can achieve a meaningful impact on your business. With an individual approach to every potential impression that is bid on, the margin can be optimized and significantly increased. Custom Bidding lets you define and set up your individual approach to optimization by creating an algorithm to fit exactly your strategic goals and is informed by your company’s unique data.
The feature “Experiments & Lift” lets you set up a testing framework to verify the Custom Bidding algorithm is working as intended and helps you evaluate how the introduction of the Custom Bidding strategy impacts the campaign’s performance compared to standard bidding strategies.