Marketers are on a constant journey to optimize the efficiency of paid search advertising. In B2B in particular, many organizations are eager to train bidding algorithms based not on generic website actions, but on unique, first-party data signals that are more directly correlated with pipeline generation. By doing so, search campaigns can be optimized against measures that truly move the needle for the business, instead of against data that’s simply easy to collect.
Merkle | Cardinal Path recently helped a SaaS client do just this, by integrating a combination of different Google technologies. The results have been impressive, most notably an estimated 20% improvement in ROAS. In this post, we’re sharing the details of how various Google tools can be connected to build a data pipeline that allows B2B marketing organizations to deploy bidding strategies that are optimized based on first-party data. For example, no longer do you need to optimize based on one-size-fits-all “form submit” events on your website. Instead, you can blend lead-gen data and offline conversion data to bid against a predictive lead score that’s been generated based on your proprietary data. Let’s see how it comes together.
Not surprisingly, one critical piece of this solution is Google Analytics (GA). When users click through on ads and reach your landing pages, they sometimes end up becoming leads (e.g. by submitting a form). This lead data — including the Google Click ID (gclid) — is collected by GA, and then exported to BigQuery (BQ) via the native GA-BQ integration.
Later, some of those leads will ultimately convert into customers. It’s important to understand which leads convert, and which leads don’t, in order to assign a predictive “lead score” to any current lead. To make this happen, we push a daily import of leads data into BQ, enabling us to understand which of the leads we’ve tracked in the past have now converted into customers.
There’s one more important piece of the puzzle to highlight: the Floodlight tag. Floodlights allow the Advertiser ID, which is critical to this solution, to be pushed into BQ along with behavioral data from GA and conversion data from your back-end systems. In addition, the Floodlight is the “vehicle” by which Search Ads 360 (SA) is able to bid against a lead score, instead of against a more generic conversion event. Critically, “Tightlock” — an open-source Google tool for onboarding first-party data to Google via pre-existing Google APIs — allows us to push the Predictive Lead Score from BQ into CM using a Floodlight tag. From CM, the Predictive Lead Score can then be shared with SA for bidding.
All in all, after connecting the various pieces of the Google puzzle, you’re left with the ability to bid against a fully customized, first-party, ML-driven lead score, which learns over time to become even more effective. Here’s what we’re seeing so far.
- Approximately 90% of offline conversions from paid search can be matched back to a GCLID, meaning that the process to attribute offline conversions back to Google ad spend is highly performant.
- Approximately 35% of paid search campaigns have successfully been migrated to a Predictive Lead Score-based bidding strategy
- Based on early results, we’re projecting approximately a 20% improvement in ROAS.
If you’re currently optimizing SEM campaigns using generic website actions, and you’re looking to optimize performance through the power of machine learning and first-party data, get in touch with us to talk to an expert.
To learn more about Tightlock, check out Google’s documentation here.
