Nearly fifteen years ago, in September of 2009, I attended ad:tech London. Now almost forgotten, the ad:tech series of conferences in San Francisco, New York, London (and more) used to be a magnet for ad technologists: every new trend would set the floor abuzz and every innovation was on full display (mostly on paper. Most were faking it and hadn't made it yet).
I remember that 2009 conference for two reasons: (1) I had to share a room with my boss because that's how CEOs of publicly traded companies demonstrated strategic vision and earned the big bucks during the Great Recession. (2) Third-party cookies were the bomb and "behavioral targeting" was the buzzword of the moment.
The rise of AI and synthetic data
Fast forward to late 2023. The buzz is now around how AI and synthetic data will reshape marketing practices. I had never even heard of this until about a year ago, but the rush to move past third-party cookies is accelerating the adoption of synthetic data that is generated through advanced algorithms and machine learning models. By creating realistic yet artificial datasets, marketers can maintain privacy compliance while gaining valuable insights into consumer behavior. This approach also allows marketers to optimize campaigns, personalize content, and enhance customer experiences without third-party data.
Synthetic data is now being used by brands for probabilistic CRM enrichment (this in turn impacts retargeting strategies, UX personalization, etc.) and by ad networks to train advertising algorithms and refine targeting sans third-party cookies.
What is synthetic data?
Synthetic data, in the marketing context, refers to artificially generated datasets designed to replicate the characteristics and behaviors of a target audience without compromising individual privacy. This innovation enables marketers to glean insights, optimize campaigns, and personalize strategies without relying on traditional, often privacy-intrusive, methods.
How Artificial Intelligence generates synthetic data
Advanced algorithms and machine learning models create artificial datasets that mimic the characteristics of real-world data. The process starts by analyzing the structure, patterns, and characteristics of an original (real) data set. This could include customer behavior, demographics, purchase history, or any other relevant information.
Once trained, the generative models can produce synthetic samples that resemble the original data. These synthetic samples are generated in a way that statistically matches the patterns observed in the training data. You can think of it as a ChatGPT-like model that – once trained on a large, anonymized set of information, generates audiences instead of text.
Examples of synthetic data in marketing:
To better explain how synthetic data is being used in marketing, I'll cite a few examples.
CRM Enrichment and adaptable personalization.
Traditionally, brands often purchase data to gain additional information about their existing customers: an e-commerce company might have someone's email, physical address, and purchase history. To better re-engage that customer with personalized communication, they would enrich that information by purchasing from a data broker additional personal information (such as age, gender, purchase habits on other sites, etc.) known to be associated to that user.
Synthetic data can replace real personal information with probabilistic figures, avoiding the actual trade of potentially sensitive and identifiable data. This way synthetic data allows marketers to experiment with various personalization strategies without risking the privacy of real customers.
Synthetic data empowers marketers to create hyper-targeted campaigns by generating artificial datasets that mirror the demographics and preferences of their actual audience. This approach enhances precision targeting without relying on a transfer of sensitive personal information.
Privacy-Compliant Audience Insights.
Marketers can utilize synthetic data to extract valuable insights about their target audience without compromising individual privacy. This approach facilitates the creation of more effective marketing strategies while adhering to evolving privacy regulations.
Dynamic A/B Testing.
A/B testing is fundamental in refining marketing strategies. Synthetic data provides a versatile and privacy-conscious environment for conducting A/B tests, allowing marketers to optimize elements of their campaigns without using real user data.
Innovative Product Development.
Understanding customer preferences is key to product development. Synthetic data generated by AI models trained on large, anonymized datasets enables marketers to simulate customer feedback and preferences, guiding the development of innovative products and services that align with market demands.
The future of marketing with synthetic data
As privacy regulations become more stringent, and consumers demand more control over their data, synthetic data becomes a massive opportunity for marketers. The ability to navigate the complexities of targeted marketing without compromising individual privacy positions synthetic data at the forefront of the future marketing toolkit.
What excites me most about the rise of synthetic data is the potential democratization of digital marketing. By supercharging first-party data, it erodes the advantage of incumbent players that sit on overwhelming amounts of data to optimize their marketing, and lowers the barrier to entry for smaller players and innovators.