Big Data, Big Language, and Big Learnings: A History of AI for Marketers

The concept of artificial intelligence existed before the computer did. But some of the biggest paradigm shifts in tech have happened in the past 10 to 15 years with the precipitous increase in a computer’s ability to process information. This has made huge ripple effects in our space, starting with this term that you’re probably very familiar with—big data. 

The ability to process and store data has increased tremendously over the past decade and the big question remains: how can we leverage these learnings? For many of us, this ability has arrived in the form of predictive analytics a.k.a the ability to run machine learning on top of these massive datasets. The companies that did well were using a lot of structured data (i.e. numbers and values that are easier to harness) and then connecting predictive models to learn against it. 

More recently, when we talk about AI for marketers, we’re having conversations around generative—Chat GPT and other new models. Right now many martech solutions are using it as a wrapper or quick message generator within their own product. 

 If you still consider yourself an AI novice, then buckle up. This is your crash course.

Predictive analytics uses machine learning to predict a numerical value. Sounds fancy, right? It's really just an applied statistical model that, over time, gets fed more and more data and continues to learn.

A lot of examples in the martech space are ones many marketers have seen, and perhaps still use today. For example, being able to predict the purchase propensity based on previous purchase data, click data, or behavioral data on a website or mobile app, or other similar services. This helped marketers gauge how many messages they could send somebody within a period of time before they would start to disengage. This was helpful in maximizing a customer’s engagement or purchase intent.

Now remember, predictive analytics has been around for many years and, while valuable in certain contexts, has yet to deliver on the promise of harnessing data for personalization that results in significant revenue lift.

More recently, the dawn of generative AI has been one of the more exciting technology advancements for marketers to harness.

The most well known product utilizing generative AI is ChatGPT. GPT is an LLM, or a large language model, which has the ability to process a lot of unstructured data. Unstructured data is exactly how it sounds: due to its size and the fact that it can come from video, audio or long form files, it doesn’t fit neatly into a table of data as structured data can. 

If you’ve ever been curious about how this actually works in the background, GPT consumes information that is publicly available, and tokenizes that information, meaning it will segregate the text into segments following certain rules, and it looks for patterns in language. It's able to identify the most appropriate answer based on all the patterns across the internet. A key benefit from generative AI is that it’s consciously learning and optimizing, constantly iterating and improving on end results. Plus, the end user has the opportunity to provide input to get closer to a desired outcome. 

As marketers, we've been thinking about having this holy grail of creating the perfect message, with the perfect product paired with the perfect offer. And in the realm of when predictive analytics was popular, you could see a small improvement. One thing you couldn’t do? Personalize the text. 

All that to say—we’re getting closer to bringing marketers exactly what they’ve been waiting for. 

Predictive models alone didn’t really have the results marketers hoped for. Brands typically see a single digit performance lift with predictive analytics, but nothing massive. Now with GPT, however, we're seeing a double digit lift on messages with optimized text. No complaints here!

The marriage of predictive and generative tech will ultimately transform martech and deliver next level revenue results for ecommerce businesses. It's the combination of the two that will enable brands to achieve SMS personalization that drives significant revenue lift...and it's coming soon!

- Big data led to advancements in predictive analytics about a decade ago, primarily focusing on structured data to generate numerical insights.

- Generative AI (e.g., GPT) has emerged in the last few years, operating on unstructured data to generate text content.

- Combining predictive analytics (numerical insights) and generative AI (text generation) has the potential to unlock advanced personalization that drives meaningful revenue lift when applied to ecommerce marketing.

When we say the future of AI is at Postscript, we’re serious. Sign up for our upcoming webinar to stay in the know.