SEM bid modifiers: Optimizing Google, Bing Ads bids with predictive geo-targeting

Google and Microsoft both allow for flexible geographic targeting up to a point, which means we can use AI to bundle groups of individuals, find the commonalities, and make a recommendation about how much a marketer should be willing to spend to engage with them.

SEM bid modifiers: Optimizing Google, Bing Ads bids with predictive geo-targeting

This article is part of Faraday's Out of the Lab series, which highlights initiatives our Data Science team undertakes and challenges they solve.


What is search engine marketing?

Search engine marketing is the long-established workhorse of digital consumer outreach. Nearly as old as the concept of search engines themselves, SEM is the shared-interest umbrella under which vendors can meet potential customers.

Over the years, the search incumbents (let’s be real here: we’re talking about Google and Microsoft, who combine to hold almost the entirety of search traffic in the United States) have made tools available for marketers to fine-tune their outreach. These can lead to a focus on specific search terms, audience characteristics, dates, times, or even — spoiler alert — geographies.

SEM is now such a mature approach that an entire economic landscape has sprung up around it, based on good ol’ supply and demand. If you want an ad impression included with a popular search term, then you’ll have to outspend your competitors to get it. Conversely, if you’ve got a niche audience, you may find outreach to be pretty cheap — at first.

How predictive geo-targeting can enhance your SEM bidding strategy

Faraday's bid modifiers are based on the spatial aggregation of individual consumer predictions. Google and Microsoft both allow for flexible geographic targeting up to a point, which lets us bundle groups of individuals, find the commonalities, and make a recommendation about how much spend advertisers should allocate to each geography.

A map of the mid-Atlantic region of the US, color-coded to show examples of cities where Faraday might suggest increased bid-modifiers by zip-code
Bid modifiers by zip-code for a mid-Atlantic SEM campaign

Using a combination of market size and propensity range, we can divide the landscape into optimal units, each with a recommended modification of the “standard bid,” or what a marketer would pay for each interaction in an information vacuum. Clients who use these bid modifiers typically see increases in ROAS and conversion rates.

Keep in mind that geographic SEM isn’t always the right approach! The toolset continues to expand in the era of AI, and “smart” campaigns are frequently available to auto-tune the parameters of an outreach effort. However, this can cede a lot of decisions to opaque algorithms, and at Faraday we prefer to be transparent — about the factors that define a predictive model, and about the geographic patterns that emerge from those.