Beeswax Ups Its Optimization Game

While most DSPs let marketers increase or decrease bids based on factors like time of day, Beeswax is upgrading that idea with a product that lets buyers optimize bids based on up to 40 factors from the bid requests.

Bid Models, which graduated from beta Wednesday, uses multivariate bidding algorithms to help marketers find “pockets” of the most effective inventory, said Beeswax CEO Ari Paparo.

For instance, instead of optimizing their bid toward a news publisher that performs well, marketers can optimize toward someone reading the finance section of that same newspaper on Chrome during the workday.

Bid Models also shows which exchange provides the most efficient way to reach that potential customer.

These upgrades can dramatically improve efficiency. One financial advertiser using Beeswax’s Bid Models saw its cost per install decline from $150 to $50, Paparo said.

Beeswax’s granular optimization capabilities also help marketers hold onto their data, and avoid sharing it with others, where it can be used to help competitors optimize.

“If you are a client in an incredibly competitive industry, and your campaigns are being grouped with other campaigns to create shared learnings, you are giving your competition momentum,” said Brian Tomasette, head of programmatic at Camelot Strategic Marketing & Media, an independent agency.

Camelot’s engineering team builds bid models, then uploads them into its targeting database to change bids and optimize audience targeting in near real time, Tomasette said. Beeswax then handles the execution.

The multivariate bid models also allow marketers to compete with black box, machine learning models where they must cede control of optimization and transparency in order to improve campaigns.

“Machine learning works best when there is a ton of data, which means sharing data across companies,” Paparo said. “That’s not that appealing to marketers from a transparency or data ownership point of view.”

Beeswax also offers more information so marketers can sharpen their bid models. Most DSPs only share win logs. But Beeswax lets marketers know all the bids won and lost.

So marketers might realize they have more opportunities to win on a publisher’s site and lower their bids. Or they might find out the opposite is true, and increase bids for a small pocket of inventory.

“It’s not about having unique, magical data. It’s about having more access, control, APIs and robustness,” Tomasette said. Marketers need to be able to pull back the curtain and make decisions based on the big picture, which means looking at log-level data.

By offering more powerful optimization with fewer resources, Beeswax plans to win more clients, many of whom don’t use Beeswax because they don’t have the in-house talent to enable build-your-own algorithms.

Also, the decreasing cost of cloud computing makes it easier for a client’s data scientist to work directly with Beeswax, minimizing the need for specialized talent to plug into the tech.

“We are in the early days of democratizing data science,” Tomasette said.

And of Beeswax’s 75 clients, a half-dozen are already using Bid Models, having joined from the beta test.

“We think this is going to be an accelerator for our business,” Paparo said – helpful for a business that raised Series B funding this January. “We are giving people full power, but reducing the skills and requirements to execute.”

 

 

Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>