AI isn't just changing creative tools, but it is also changing the way we use data in marketing.

Most of us are familiar with anonymised data, but a new frontier is emerging, and that is synthetic data powered by AI.

In this episode of The Marketing Guy, I break down:

(a) What is synthetic data, actually, and what does it mean for marketing

(b) How it compares to anonymised data

(c) And why it becomes the backbone of targeted, simpler, and privacy-friendly campaigns

Notes from the podcast

(1) MostlyAI (www . mostly . ai) specialises in synthetic data. Their platform helps brands create these realistic data sets, mirroring your real data, that you can play around with. All that data is safe to use and can unlock insights without exposing the real customer data you have.

(2) Podcast on cookiesLook for "Goodbye Cookies, Marketing's Biggest Shift Explained" from my channel. Link in comments as well.

(3) What models does synthetic data use?

(i) GAE stands for Generative Adversarial Network, and it is not something that LLMs use. There are essentially two neural networks in it. (a) Generator. This makes fake data, and (b) Discriminator. This checks if the data is real or fake. With enough training, the generator creates data that is almost indistinguishable from your real data. It's used heavily in deep fakes.

(ii) VAE stands for Variational Autoencoder, and it has two parts as well. (a) Encoder. This essentially compresses the data you feed. (b) Decoder. This then rebuilds that compressed data into something new, slightly different outputs.