Google earlier this week unveiled Google Attribution, a tool that uses machine learning to measure the effectiveness of online marketing campaigns across a variety of devices and channels.
Part of a series of new measurement tools introduced at Google Marketing Next, Google Attribution aims to help marketers determine what is driving consumers to make their online purchasing decisions.
Google is at an “inflection point,” noted Sridhar Ramaswamy, senior vice president of ads and commerce, when it comes to the use of technology to measure online marketing effectiveness.
“What’s exciting is that things that were really hard we’re going to be able to make easy, because technology is doing the heavy lifting,” he told conference attendees.
Page-loading time is an important factor in terms of how long advertisers have to hook a consumer online, Ramaswamy said.
As load time increased from one to seven seconds, the probability of a consumer bouncing more than doubled, found a recent study that analyzed 900,000 mobile landing pages in more than 100 countries. That translates to conversions falling by up to 20 percent for every one second delay in page-load time.
Most existing attribution measurement tools have three main shortcomings, according to Ramaswamy. First, they lose track of the customer journey when they move between tools; second, they are hard to set up; and third, they are not integrated with ad tools.
Because of those shortcomings, marketers typically are stuck with using last-click attribution, which misses the impact of most marketing touchpoints.
The new Google Attribution tool incorporates data from AdWords, Google Analytics and DoubleClick Search, which provides a more comprehensive picture of data from all marketing channels, according to Google.
Google Attribution also makes it easy to switch over to data-driven attribution, which uses ads, keywords and campaigns that have the greatest impact on a company’s goals, Google said. Using the data-driven attribution model typically requires a minimum set of data, which is 15,000 clicks, and at least 600 conversions within a 30-day period.
Google will offer an enterprise version, Google Attribution 360, to meet the needs of larger advertisers, noted Bill Kee, group product manager at the company.
Google Attribution currently is in beta and will roll out to additional customers over the next few months.
Not Quite Ready
Google Attribution is an “aspirational sell” for advertisers, particularly for small and medium-sized businesses, suggested Paul Teich, principal analyst at Tirias Research.
“Google intends to make it free, so it will see a lot of uptake, but many users will have zero comparative experience to judge how much better they are doing with Attribution,” he told the E-Commerce Times.
The major benefit will be providing more-accessible feedback for designing online user experiences, Teich said, noting that it’s not clear how artificial intelligence can make a contribution — at least until Attribution has been able to analyze millions of Web pages.
Google is ahead of the curve compared to other Internet companies, which are only average at measuring the efficacy of online marketing, observed Michael Jude, a program manager at Stratecast/Frost & Sullivan.
“Machine learning does give it an edge in making sense of the relationship between key clicks and online purchases,” he told the E-Commerce Times.
“While simple algorithms can provide some insight, especially as applied to last click models, machine learning can understand such things as click back, where the last key click is not the purchase, but rather an artifact of comparison shopping,” Jude explained.
Google made several other announcements related to providing additional marketing data.
It launched a new AdWords beta to let advertisers use fast-loading AMP pages as landing pages for search ads and to speed up ads across the Google Display Network. This follows Google’s introduction about 18 months ago of the Accelerated Mobile Pages project to speed up the Web with faster browsing.
Google also announced a new tool to bring in-market audiences to Search, in order to help advertisers reach users who are ready to buy their products. Using machine learning, the tool analyzes trillions of search queries and millions of websites to determine when searchers are close to making a buying decision on a product.
The company unveiled plans to roll out store-sales measurement at the device and campaign levels, which will allow advertisers to measure in-store revenue and store visits delivered by search and shopping ads.
If advertisers collect email information at the point of sale through loyalty programs, Google said, store transactions can be imported directly into Adwords by the advertiser or through a third-party data partner.
Good article. I am quite eager to learn as much as possible about how to be an early adopter of machine learning to propel our ecommerce sales. Keep them coming!