Autobidding, Auctions, and Automatic Growth

Michael Ostrovsky, Co-Founder and Chief Scientist

July 29, 2022
2 min read
Monetization Tech

Going once, going twice, SOLD!

In the beginning, bidding in ad auctions was basic, done on a per-click basis, now, it has expanded to many other types and flavors: per-impression, per-action, per-purchase, per-like, and so on. Advertisers have freedom to vary their bids by time of day, geography, keywords, and many other dimensions - and these dimensions can be different on different platforms on which an advertiser may want to promote its goods and services. While natural, all of these variables and dimensions make bidding incredibly complicated.

Some large advertisers respond to this complexity by hiring armies of Ph.D. data scientists. Others end up outsourcing the entire bidding part to third-party firms. Many smaller advertisers get overwhelmed and give up advertising, or only focus their attention on just one or two ad platforms.

Game Changing Democratisation

We don’t believe successfully participating in advertising auctions should be that difficult. Why should brands and vendors feel like they need full training to be able to advertise their own products? Don’t brands have enough to do!? To help advertisers, some of the leading platforms have recently introduced a game changer—autobidding.

Autobidding allows advertisers to specify just a few of their high-level objectives and constraints (e.g., which products they want to promote and what their daily budget is), and the platforms’ internal bidding bots will do the rest, optimally determining an appropriate bid for each opportunity to show an ad to a user, and dynamically adjusting these bids in response to changes in the bidding environment.

Internally, these bidding bots can be very complex, utilizing large data sets and relying on a variety of ML/AI algorithms to optimize the performance of ad campaigns - but crucially, this complexity is hidden from the human bidders, who can enjoy a natural and intuitive user interface, and simply observe the fruits of the optimization.

Sounds a little technical? Well, in the words of the British science fiction writer Arthur C. Clarke, “Any sufficiently advanced technology is indistinguishable from magic.” And this is exactly what a well-designed and properly implemented autobidding system does for its users - it magically gives them strong results from advertising campaigns, without having to spend enormous amounts of time or resources on managing them.

How the giants are doing it:

Unsurprisingly, on the platforms where it is available, autobidding has become very popular among advertisers. According to Google, over 70% of their advertisers use autobidding.

On Pinterest, over 80% of cost-per-click revenue comes from automated campaigns. On Facebook, “optimized cost per mille” and other automated campaign types are widely used by advertisers. However, it took these platforms many years of trial and error and a lot of engineering resources to perfect these products.

How Topsort is doing it:

Topsort’s ad platform was built from the ground up with autobidding in mind, and incorporates it natively - so a marketplace that deploys sponsored listings based on Topsort’s infrastructure can enable autobidding instantly, making it easier for advertisers to start using the product and quickly get positive results, stimulating adoption and reducing attrition.

Read more like this