Introduction

Wattle Labs has created a "Large Behaviour Model" (LBM) of the human population. The LBM, an analogue of LLM's, which seeks to understand human behaviour, rather than language.

This video is an introduction to our LBM and how it can be applied to sales and marketing

#datascience #AI #digitalmarketing #ecommerce #behaviourchange #BehaviouralScience #psychometrics #nudging #boosts

https://www.youtube.com/watch?v=xI4mqfN7Nas

Transcript

Wattle Labs is an AI and machine learning startup with a focus on sales and marketing, and we have created a large behaviour model. Everyone's already aware of what a large language model is — well, a large language model is to speech and language what a large behaviour model is to behaviour. We can do amazing things with it.

For anyone in sales, marketing, loyalty, and so forth, I'd like to start here and say that the purpose of all sales and marketing is really to change behaviour. That's what you're trying to do. You're trying to change the behaviour of a person — buy more Coke, come into a brand, try a new education programme. It works offline and online — our technology works in both contexts — but you're really, fundamentally, trying to change behaviour.

And I'll put it to everyone in sales and marketing that the most effective way to change someone's behaviour is by knowing who they are. Think about it — if you tried to get a random stranger off the street to try a new beauty product, you've got almost no chance. Maybe 10%, maybe they happen to hate their current brand. It's knowing who the person is that enables you to change their behaviour, to upsell, cross-sell, and get them into a brand. The more you know about a person, the better and easier it is to change their behaviour.

Think about it this way — if you're trying to change the behaviour of a sibling, a parent, or a partner, and you say, "Hey, you should try this new type of lolly," you're quite successful in doing it. Not withstanding the fact that there's a trust relationship there, you also know the person. You know what they want — and what they want is in fact the product. You know the product to sell them, and you also know the message by which to sell it. I'd posit that that's the basis of sales and marketing from an engineering perspective.

And here's the problem. The way current data science works — everyone's data science — is it takes a whole bunch of transactions and creates a propensity model: a model of ones and zeros and conditional probability to predict behaviour. Everyone uses it. This model — directed graphs, conditional probability models — is what your data science people are doing every day. They take a bunch of transactions, build a propensity model, and for each person say, "You should put this ad in front of this person." This is 1990s technology. It comes from Amazon, the earliest parts of the internet, and Google. And it's of course flawed, because this is not how human sales and influence actually works.

With human sales and influence, you change behaviour based on what you know about the person. It really starts with the person first — and this is what our AI and machine learning does. It asks the question: who is the person, and what do they want? That's really what our large behaviour model seeks to answer using machine learning and AI.

To do this, we need to know certain socio-economic factors, and we need to know their psychology and behaviour. We can keep building upon this profile — understanding their wants, dreams, and desires. This is what we would consider the behavioural model of the person. We have profiled all 28 million people in Australia across all 16 million addresses and created a profile of each and every single person. We've started this equation by fully understanding who people are — and we've done it without stalking their emails, listening to their personal phone calls, watching their browser behaviour, or recording everything they do in apps. We've done it through brute force mathematics. We've created a very deep profile of every single person, and currently the model works on Australians.

So we know who the person is — and only then can we understand what they want. By what they want, I obviously mean products. But here's the really cool bit: by understanding what someone truly wants and who they truly are, we can predict the products they really want — not just the products they're currently using. We can look at what product is most aligned with their psychological and psychometric profile. What product should they really want, but they're not currently using?

The second thing we can do is construct the message. The best message that resonates with a person is one that's built around their psychometric profile, their behaviour, and their psychology. So we can identify what product they should really want, and then determine the optimal message to deliver it — knowing everything we know about that person. Together, those two things allow you to do truly optimal sales and marketing. It's really quite unfair. If you know who a person is before you knock on the door to sell them a vacuum cleaner, you've got a really, really unfair advantage.