Google is giving search advertisers the ability to match in-store credit-card transactions with its online ads to prove that spending on search really works.
As with most Google products these days, it's turned to machine learning to improve the intelligence it can provide advertisers about consumer behavior online and offline.
Its latest effort will keep a tab on purchases at the store register and correlate them with actions on the phone, such as searching for the product or an alternative location.
"In the coming months, we'll be rolling out store sales measurement at the device and campaign levels. This will allow you to measure in-store revenue in addition to the store visits delivered by your Search and Shopping ads," Sridhar Ramaswamy, Google's senior vice president of ads and commerce, said.
Advertisers can access this information via a new product called Google Attribution, which is rolling out now in beta to help them adopt "data-driven attribution", allowing them to see which clicks and keywords had the biggest impact on a consumer's decision to buy.
Google promises that matching transactions with Google ad interactions will be done in a "secure and privacy-safe way, and only report on aggregated and anonymized store sales to protect your customer data".
The ad technology builds on the cross-device marketing analytics tools Google unveiled in 2014 to improve store-visit measurements, which were useful but didn't offer anything about conversions to sales.
Google says the new in-store measurement service can exploit its relationships with firms that track about 70 percent of all credit- and debit-card transactions in the US.
The other way of beefing up transaction data is by firms or their marketing partners importing store transactions to AdWords when the store collects email information at the point of sale for loyalty programs.
Google also boasts that its new deep-learning models, combined with 'nearby places' searches in Google Maps, can help advertisers bridge the gap between what customers do on their phones and the physical world.
The company said advertisers had measured some five billion store visits through AdWords in the past three years. Deep learning is helping it predict store visits in historically difficult to measure densely packed places like multi-story malls.
It's rolling out in-store visit measurements to YouTube video ads soon, and it's already available in Search, Shopping, and Display campaigns.