Use machine learning to manage and forecast inventory more effectively

Accurately forecasting and managing advertising inventory is critical to successfully monetizing content. Not only do you need to ensure that your Google Ad Manager account is set up for success, you also need to make sure that your sell-through rate — the total number of page or video views that are actually sold — is as high as possible for every ad unit across your sites or apps. Using machine learning, Ad Manager can help you develop an effective network strategy, detect trends, and uncover insights so you can better manage and make the most of your inventory.

To manage your inventory with Ad Manager, you first configure a network based on ad units, key values, and placements. Once you have designed and set up your network, you can use forecasting to estimate and plan for how much inventory you’re likely to have available, so you can make more impactful decisions about how to package and sell it.

Below we’ll review how to approach managing and forecasting on your inventory within Ad Manager.

When you set up your network, you have flexibility within Ad Manager to design and manage your inventory structure to support your unique sales and business needs. By customizing your network setup, you can ensure that your unique sales scenarios, audience profiles, and reporting requirements can be met.

Define the structure that makes sense for your inventory.

We recommend planning your inventory structure before creating ad units, placements, or key-values. The inventory you create will define all the spaces available for ads across your site, so the structure should mirror the organization of your site’s content. Once you’ve planned your inventory structure, you can create ad units, placements and key-values.

Think of Inventory in Ad Manager as the space on your website or app where you display ads. Each ad unit is a component of your inventory, and you also group ad units together and key-values using placements, which provide more targeting granularity within an ad unit. When you define your networks' inventory structure in the Inventory section of your Ad Manager network, it makes it easier to plan and manage your inventory’s performance. To understand how to best set up your ad inventory, start by viewing our article to create your inventory structure.

When checking available inventory, Ad Manager analyzes the line item you’re forecasting against competing booked line items. The platform looks at the last 24 months or more of unsampled historical traffic volume data as a basis to build and predict future traffic volumes.

Using these predictions, Ad Manager runs a simulation for the line item to be forecasted. The output of the simulation estimates how many impressions are available for booking and how many matched the criteria of the line item.

Here are some additional benefits Ad Manager’s forecasting capabilities offers:

• Real-time forecasting that takes into account over 24 months of aggregate data to deliver better forecasting accuracy and trend detection.
• Visually appealing user interface that provides you with visibility into how we predict future traffic and tools to further drill down into specific segments for more granular insights.
• Adjustment tools to fine-tune forecasts based on insights and trends Ad Manager is not able to predict.
• Continuous improvements to our algorithms and machine learning models such as Seasonality — our forecasting engine can automatically detect seasonal trends in your traffic and apply them to future traffic.