What is signal-based marketing?
An explanation of the Amazon Ads approach to addressability
Advertising using signal-based marketing involves leveraging available signals and machine learning to deliver relevant messages without the need to rely on third-party cookies. Signals refers to a wide range of consumer events and behaviors, at particular moments in time, that can indicate interests and affinities
Today, almost 40% of web traffic and 37% of app traffic is unaddressable via traditional advertising methods, and that number is expected to climb to 95% across the board as third-party cookies deprecate.1 And yet, brands still seek to connect with consumers and measure their ads’ impact amid this seismic shift.
At Amazon Ads, we’re committed to helping brands navigate these challenges, and signal-based marketing is our approach to meeting their need for a durable way to reach their audiences.
Read on to learn more about how signal-based marketing can help brands overcome addressability challenges, why it’s different from other solutions across the industry, and how to get started leveraging this powerful approach.
What exactly is signal-based marketing?
Signal-based marketing is our approach to helping advertisers deliver relevant messages at relevant times without a reliance on third-party cookies. As Neal Richter, director of bidding science and engineering at Amazon Ads, recently wrote in AdExchanger2: “Cookies are … false precision. Signal-based marketing, by contrast, allows brands using Amazon DSP to engage consumers with relevant, useful advertising, as opposed to repeatedly showing them an ad for a product they had considered and not purchased.”
The term signals refers to a wide range of aggregated consumer events and behaviors, across the path to purchase, that help Amazon DSP machine-learning models predict interests and affinities. Signal-based marketing leverages these models along with contextual signals to help brands reach consumers without sacrificing reach, relevancy, or ad performance, and without relying on traditional advertising identifiers.
At Amazon, customer trust comes first. Continually earning and maintaining trust is one of the company’s guiding principles, and signal-based marketing was built with our customers in mind. With signal-based marketing, our predictive models result in audiences seeing more relevant messages in more locations.
By leveraging our machine-learning models and the breadth of signals available to us, signal-based marketing helps brands get their ads to the right places in the right moments, based on current context, rather than historical behavior. And so far, brands using Amazon DSP have seen a 20% to 30% increase in addressability on browsers like Safari and Firefox, as well as operating systems like iOS.3
Signals from our global retail and streaming properties help us understand how consumers discover, consider, and purchase products and services. These signals also include contextual signals that help us uniquely understand, based on our retail taxonomy, the interconnectivity between content and how consumers who engage with it go on to discover, consider, and purchase products. Together, these signals enable us to develop actionable insights that inform ad-serving decisions, both on Amazon properties and across the web.
These are inclusive of pseudonymous signals including website engagement and conversion signals on an advertiser’s owned site as well as advertiser audience signals stored across systems like data management platforms (DMPs), customer data platforms (CDPs), and customer relationship management (CRM) platforms . The combination of Amazon Ads signals and an advertiser’s own signals can feed into our models to enhance the brand’s campaign recommendations and optimizations.
We receive signals from Amazon Publisher Direct, which provides Amazon DSP with direct access to publisher inventory. These signals further inform our modeling, and because we’re minimizing intermediaries in the supply chain, they’re received more quickly and with higher fidelity.
Finally, we’re able to maximize the use of real-time bidding (RTB) signals including (but not limited to) device type, time of day, software version, and page URL, which helps our models better parse ad opportunities. This is possible by combining these signals with those listed above to make better ad-serving decisions at scale.
1 StatCounter, WW, 2022
2 AdExchanger, May 2023
3 Amazon internal data, US, 2022; 140K campaigns across verticals