In App Advertising

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.

Our AI/ML driven psychological and behavioural profiling can greatly enhance digital marketing/advertising across all channels

#Cookieless #CookielessMarketing #DigiatlMarketing #CustomerProfiling #BehaviouralModelling #ContextualAdvertising #PsychologicalTargeting #DataScience #BehaviouralScience #Psychometrics #Nudging #Boosts

https://www.youtube.com/watch?v=IFP1VqhVImw

Transcript

Hello everyone in Australia who wants to do in-app advertising. My name is Aaron and this is Wattle Labs. Today I want to show you how you can increase revenue by putting better ads in front of people using our technology — and we're going to do that without employing very expensive data scientists, without huge lead times, integration times, or testing times. And in the end, we're going to do it better than any contemporary data scientist in Australia can, for the fundamental reason that current data science relies on 1990s technology. And it is wrong. Totally wrong. I'm going to prove it to you right now.

So in this model, you have people — customers — who come onto your app and buy something. Through that user journey, after they've clicked purchase or added something to cart, the journey continues and you're going to have one or more opportunities to put an ad in front of them. It could be in the checkout process, it could be ads as they're browsing your e-commerce store. And finally they check out and disappear. Your problem is this: for every single customer, what ad should I display? It's not a trivial question to answer.

The way most businesses tackle this today is to talk to a data scientist or engage a company that has data scientists. They come in, take all your sales data and transactions, and say: "We found correlations between Red Bull and pies and noodles. If someone purchases a Red Bull, there's a 50% chance they'll purchase a pie and a 50% chance they'll purchase noodles." The maximum performance of this type of approach using modern data science techniques is a 50/50 chance of putting the right ad in front of the right person. The person purchased a Red Bull — you've got two options, and it's a coin flip. Half the time you'll put the pie ad in front of them and you'll be right. But the other half of the time you put the pie ad in front of someone who was actually more interested in noodles, they didn't think to purchase noodles at the time, and you've lost the revenue from that upsell or cross-sell.

And that's fundamentally because this model is wrong. It's really nonsensical to be correlating a product to another product. I'll show you why.

With Wattle Labs, we start with the fact that someone purchased the Red Bull. Wattle Labs takes that specific customer and profiles them to create a persona. In this example, there are only two types of persona — tradies and students. Our system can psychologically, behaviourally, demographically, and socio-economically profile each and every one of your customers with 92% accuracy. The customer purchased the Red Bull — we then profile them to determine whether they're a tradie or a student. And then the really important bit: ask anyone in Australia what tradies eat, and they'll say pies with 100% certainty. Ask what students eat — noodles, with 100% certainty. It's really trivial, but we've just doubled your revenue.

Current data science takes a bunch of sales, extracts all the transactions, and creates a propensity model linking product A to product B, product A to product C. And as I said, this is wrong. It's not the fact that the person purchased the Red Bull that is correlated to pies and noodles. What's actually happening is that the person who purchased the Red Bull is correlated to tradies and students, who are themselves correlated to pies and noodles. There is a missing element in current data science. The models are wrong.

What Wattle Labs does is map the product to customer personas — essentially taking the persona of the product. Who purchases the Red Bull? What type of person are they? Then map that persona to the secondary product. Rather than mapping a product to a product, we're mapping a product to a person to a product. And more importantly, it increases the conditional probability — the chance of getting it right — to 100%. In this simple example, we have increased your upsell and cross-sell revenue by 100%.

That's the power of our behavioural and psychological modelling. That's the power of our product modelling, our customer modelling, and our recommendation engines. We can provide you with better recommendations than any propensity model can, because we fundamentally model humans and human behaviour — and we can do so without the huge data science cost, integration cost, technology design and delivery cost, scrums, and project management. It's just a nightmare — I don't know how anyone's still in IT anymore. Less cost, more revenue.

Finally, I'll offer an open challenge to any company in Australia. Let's have a competition — an A/B test between Wattle Labs and your entire BAU. Run a thousand people through your current BAU technology, tally up the dollars, and we'll do the same using behavioural and psychological modelling of your products and customers. Is there more money here? Is it less cost? We'll run it as a skunkworks. We can do two-way NDAs — you don't have to tell your customers that you're using behavioural and psychological profiling. It's like we were never even here. We'll just run a skunkworks competition quietly to see whether behavioural and psychological techniques can outperform your BAU quicker, faster, and with better results. Otherwise we'll just disappear and it'll be like it never happened, and you can't prove anything.

Anyway, I hope you have a really great day. That's all I wanted to say — have a good weekend. Thank you. Bye.