Delivery Models, a New Way to Optimize

At Beeswax we’re always looking to push the envelope on techniques for optimizing performance and workflow for our customers. When we first launched we were the only buyside platform that allowed customers to “BYOA” – Bring Your Own Algorithm. Then last Spring we launched Bid Models, a data-driven approach for customers to finely control their bid prices. Now, we’re happy to announce the availability of Delivery Models, the next step in controlling and optimizing your programmatic buying.

Delivery Models allow you to finely control the distribution of your programmatic ads separately from the prices you bid for those ads. This capability dynamically shifts budget according to your instructions without jeopardizing overall delivery. For brand campaigns, Delivery Models lets you flow your spend intelligently across deals and tiers of inventory to both satisfy business requirements and enhance secondary KPIs. For performance campaigns you can reduce waste on larger sites that eat up budget and move more spend to where performance is optimal.

you can think of the process of bidding on an auction as consisting of three inputs: Targeting, Optimization, and Delivery.

Optimization gets all the attention, with millions of hours of data science going into creating algorithms to bid the right amount on every auction. Delivery, meanwhile, has been seen as an afterthought with pacing, frequency capping, and budgeting relegated to “commodity” status in the bidder. However, delivery factors have big impacts on results, especially for branding campaigns where there is less consideration of optimization around bid price.

How it Works

Consider a practical example. You want to deliver half your budget on a set of Deal-IDs and half on the open auction, but in no case do you want to under-deliver or under-pace.

In a typical DSP set-up the way to fulfill this order is to create two line items, each with their own budget:

The problem with this approach is obvious – if you don’t have enough deals inventory to deliver against the sub-budgets the whole campaign will under-deliver. The typical solution is for the ad operations folks to constantly adjust budget and to build in buffer (waste) to make sure everything delivers in full.

With Delivery Models, you create one Line Item, then you create a separate object to define relative budget weight:

Because the budget weights are relative and not fixed you don’t have to use guesswork to estimate the availability of inventory. And our algorithm will re-adjust against actual delivery in real-time so you don’t need to worry about pacing or under-delivery. You can set up to 100 levels of weight and can use multiple criteria in combinations, so there’s a lot of power to this feature.

Further Reading

Please read our documentation:

• About Delivery Models
• API documentation

For more on some of the programmatic advertising terms referenced in this article and other programmatic advertising, bookmark our glossary of programmatic advertising terms here!