Everything a marketer needs to know about machine learning
As consumer expectations grow for more personalized, relevant, and assistive experiences, machine learning is becoming an invaluable tool to help meet those demands. It’s helping marketers create smarter customer segmentations, deliver more relevant creative campaigns, and measure performance more effectively. In fact, 85% of executives believe AI will allow their companies to obtain or sustain a competitive advantage.1
We created this guide to help you optimize your machine learning marketing efforts — whether you’re just starting out or you want to discover more benefits of machine learning.
How machine learning works for you
A quick machine learning guide for marketers
At its core, machine learning is a way to quickly label and analyze huge data sets. People can do this on their own, but a machine helps do it faster and on an infinitely larger scale. In fact, 66% of marketing leaders agree automation and machine learning will enable their team to focus more on strategic marketing activities.2
But machines can’t learn on their own — they need the help of a human. Dive in to the interactive machine learning quiz below for a machine learning 101.
Getting started: It’s easier than you think
Why the first step is taking a step back
Now that you know what machine learning is, you might be asking yourself how to get started using it. We’ve seen many marketers dive head-first into building a machine learning program from the ground up. But that’s a tricky business. It requires a lot of upfront investment. And can take years to perfect.
Instead of jumping in too quickly, take a step back. Companies, including Google, are already doing the heavy lifting by integrating machine learning into existing and new marketing products, helping you gain deeper insights from your data without additional effort from your team. All you need to do is make sure your organization is set up to get the most value out of these products.
We’ve outlined three key considerations every marketer should make to prepare their organization for machine learning.
Define your machine learning marketing goal upfront.
Much like us, machines work best when they are given clearly defined goals. Your goal, or output, works as a framework. It helps a data scientist build your machine learning models and identify the right data to use when training your model. Make sure your goal is quantifiable, and measurable. Doing this upfront will help you define and measure the success of your model.
An algorithm is only as good as its data.
Here’s a golden rule to remember: a machine learning algorithm is only as good as the data it’s fed. So, to use machine learning effectively, you must have the right data for the problem you’re trying to solve. And not just a few data points. Machines need a lot of data to learn — think hundreds of thousands of data points. Your data will need to be formatted, cleaned, and organized for your algorithm, and you will need two datasets: one to train the model and one to evaluate its performance.
Assemble a diverse team with the right mindset.
Marketing teams can identify the best use cases for machine learning, but data scientists and analysts are critical when it comes to implementation. That’s why assembling a cross-functional team is essential to the success of any machine learning program. But to get the most out of machine learning at your organization, you need the right team and the right mindset. The latter requires a cultural shift that prioritizes and rewards experimentation, measurement, and testing throughout your organization.
How machine learning powers better marketing
A deep dive into key benefits and opportunities
There are countless ways that machine learning can help your business. Explore the marketing applications below to learn what products can help you optimize campaigns, and see how brands are already using machine learning to boost their marketing efforts.
Identify your most valuable customers.
How machine learning can help you reach the right app users
Let’s say you’re trying to market your app, and your primary goal is to get your app in the hands of long-term, paying users. But you’re finding that users aren’t opening the app that often after initial download. You wouldn’t be the first to encounter this issue. In fact, only 37% of app installs remain in use after seven days. So how can you find the right audience?
If you’re using siloed sources to identify your audience, you might be missing out. Machine learning can sort and analyze sources to help you learn which users are most valuable to your app, and help you get the most out of your budget by only showing ads to users who are most likely to download and use your app. That’s how the Google product Universal App Campaigns works — by helping marketers extend reach efficiently.
Maven, GM’s on demand car rental app, launched quickly in 2016. One of the biggest challenges the business encountered was identifying and connecting with the right users. Maven’s marketing team quickly realized that a download wasn’t enough. They needed high-value members that would continue to engage with their service.
It does all the hard work for us. With machine learning, it’s always optimizing so that we can focus our time, efforts, and energy on finding other ways we can continue to connect with our members.
Kristen Alexander, Marketing Manager, Maven
To discover and reach high-value customers, Maven’s marketing team integrated machine learning into their campaign strategy with Universal App Campaigns. As a result, Maven increased registrations by 51% while decreasing cost per registration by 74%, directing newly available resources toward more strategic initiatives.
Quickly serve the right message for every moment
Machine learning is helping marketers develop custom creative
Today’s consumers expect brands to deliver assistive, highly relevant experiences. And that goes for ads too. In fact, whether an ad is relevant or not has a huge impact on a user’s decision to buy. Our research has shown us that 91% of smartphone owners bought or plan to buy something after seeing an ad that they described as relevant.3
If you think the prospect of creating an ad for every one of your customers sounds like a huge challenge, don’t fret. Machine learning is helping marketers deliver unique and tailored creative to customers. Responsive search ads mix and match multiple headlines and descriptions to find the best possible combination for a user, simplifying the ad creation process and delivering stronger results.
When Apartments.com, a leading online resource for home and apartment rentals, wanted to optimize creative for its growing audience, it turned to Google responsive search ads for help.
On average, advertisers who use Google’s machine learning to test multiple creative campaigns see up to 15% more clicks.
Google Data, March 2018
Using responsive search ads, the brand created ads that were more relevant to users at key moments in their unique rental process. The initiative was hugely successful: Apartments.com saw a 10% lift in clicks. And ForRent.com, another brand in the Apartments.com family, saw a 16% lift in clicks with a similar strategy.
Find the right customers in key moments.
The right bid is in reach thanks to machine learning
People are searching more frequently and with more specificity. For marketers, this means that it’s more important than ever to land the right bid at search auctions. But it also means landing the right bid is harder, as a growing surplus of data creates more complexity for marketers to set bids based on each user’s content.
Fortunately there are products to help you automate this process. Smart Bidding uses machine learning to analyze millions of signals and make adjustments in real time. You choose a strategy designed to achieve your company’s specific goal. Then Smart Bidding factors a wide range of signals about the intent and context of every search.
When Nissan’s partner agency OMD wanted to boost qualified visits to the Nissan website, they turned to Google’s automated bidding products for help. OMD used these automated bidding algorithms alongside their own custom settings, like curated placement lists and private deals, and created a strategy that could reach key customer segments.
The algorithms act as extensions of the OMD analysts. They analyze the campaign performance at a granular level to ensure the optimizations are more accurate and deliver faster results based on our parameters.
Alex Kraft, Managing Partner, Omnicom Media Group
As a result of this initiative, OMD optimized Nissan ads in real time, provided control over where the ads were displayed, and reached more valuable customers. The agency increased conversion rate by 67%, while decreasing cost per qualified visit by 33% and cost per click by 14%. Nissan saw so much success with automated bidding that the brand is now using this strategy for more of its campaigns.
Unlock the true value of each step on the path to purchase.
How a data-driven measurement strategy can help you unlock consumer intent
Let’s say that before making a purchase on your site, a user decided to do some more searching, shopped around, or clicked on a few of your ads across platforms or devices. Typically, credit for a conversion is given to the last ad a customer clicked. But how can you be sure the last click is the most valuable? Today’s consumers are interacting with brands across a growing number of screens and channels, making it difficult to identify which parts of your marketing strategy are working.
Data-driven attribution uses machine learning algorithms to analyze the clicks across your Search ads. By comparing the click paths of consumers who purchased your product to those who didn’t, the data-driven attribution model identifies patterns among clicks leading to conversions and identifies the most valuable touchpoints across your consumer’s journey.
People take weeks or even months to research online and plan every detail of their trip, performing hundreds of interactions along the way. To understand which of these consumer touchpoints was actually driving long-term growth, HomeAway, a digital marketplace of vacation home rentals, knew it had to shift to a more data-driven measurement strategy.
One thing automation has helped us get right is who sees our ads. We’ve not only become better at finding the right customers, we’re now also really good at not spending money on the wrong customers.
David Baekholm, SVP of Growth Marketing at HomeAway
HomeAway started paying close attention to consumer behavior signals, determining which behaviors correlated with likelihood to convert. The company’s previous strategy only measured last-click, same-device conversions. Data-driven attribution is helping HomeAway’s marketing teams better understand the intent, interactions, and signals that are driving long-term growth.
1 The Boston Consulting Group, “Is Your Business Ready for Artificial Intelligence,” Sept. 2017.
2 Google/MIT Technology Review Insights, Global, ML Leaders and Laggards, Leaders (n=186) defined as >15% increase in revenue OR 15+ point market share increase, Laggards (n=176) defined as <0% growth in revenue OR <0 point market share, 2018.
3 Google / Purchased, April 2017.
Explore more articles on machine learning and marketing