In the second installment of our conversation with Interplay VC, we dive into:
The word “predictions” is frequently thrown around in marketing, but those are usually predictions based on static databases, like your CRM file.
If someone purchases eight times, it can be reasonably assumed that they're high value.But imagine being able to do that before someone has even purchased. Before Black Crow AI, we saw that understanding a customer's value while they're still just some anonymous internet user hadn't been solved.We saw a key piece unlocked as we actually figured out how we could bring these predictions to brands in a simple and scalable way.
To bring Machine Learning (ML) to a place where it was usable for small eCom companies, we started by figuring out where it has been done before – it was done by one company for the majority of users.The way enterprise machine learning has worked historically is if you're a Fortune 500 company and you decide, "I need predictions inside of my company," you'll go out and buy Databricks.Databricks is a privately held ML infrastructure company worth tens of billions of dollars. To implement Databricks required:
Even still, 10s of millions of dollars later, these projects fail somewhere between 50% and 80% of the time.
Knowing we could do real-time analysis and knowing that the market worked in this Fortune 500 model, we asked ourselves two key questions:
Fortune 500 will do things one way, but the rest of the market needs this to be democratized.
As with any new technology, it is important to be able to point to real impacts that it can drive. With our existing customers, we’ve seen both qualitative and quantitative improvements.
Our partners, including brands Mejuri, Cotopaxi, and Daily Harvest, have a really important direct relationship with their consumers.
While business models may differ, they're all very much focused on acquiring customers efficiently and ensuring that the relationship that they start is as valuable as it can be.
Our initial customers are the avant-garde or those folks who are a little more digital native. They get this immediately. They know that they're living and dying by the CAC/LTV equation.
Not every brand we’ve talked to has immediately seen the power of prediction. We see this not as whether we add value, but who has a fast sales cycle and is open to testing and learning.
1. Prioritizing Value Delivery
In order to deliver these outcomes more democratically, we’ve turned this three-year, $10 million Databricks project into a streamlined trial and onboarding process that starts with a one-click install. Then the Black Crow model builds itself out in 2 weeks, with no additional work required.
Everything we build delivers value in 30 days or less. That's one of our missions, as well as being able to do it for less than the cost of a data scientist.
When those two things prove out, it's a pretty compelling proposition, certainly for the avant-garde, but also for very large and medium-size, legacy multichannel retailers.
2. Filling an iOS Sized Gap
A lot of brands are sitting on a gold mine of first-party data that they’re not leveraging because in the past, they never had to.
You could build a brand and a business using the huge amounts of third party data that was easily accessible through platforms like Facebook. The sheer volume and specificity allowed you to target almost any type of audience with an incredible degree of accuracy.
The release of iOS 14.5 and ATT was kind of a watershed for brands and companies that grew with this basically infinite supply of data and cheap social media advertising, especially on Facebook and Instagram.
It was the CAC/LTV equation kind of warped. Customer acquisition costs rose through the pandemic, but iOS 14.5 was a real “off-the-cliff” moment for many brands. With so much turbulence in the post iOS 14.5 marketing landscape, the brands that will win are those who know how to use their own data to fuel their marketing.
3. How Data Impact Marketing
Most brands have to completely redesign their marketing and advertising efforts. Across platforms this will look like:
With first-party data and zero-party data, it's not going to be an option to be on autopilot. You need to take the understanding of your customers and make sense of it and activate it.
Activating it could be anywhere from creative optimization to better real-time decisions to many other pieces of the organization. But data is the key to understanding what's actually going on.
When we look at where we’ve come from and where we are headed, we think about two key reasons that AI and ML are the next frontier.
1. Technology Democratization
Thinking about the technological innovation over the last 1,000 years or 2,000 years, most technology tends to create differences between the rewards to capital and the rewards to labor.
On a farm, if you invest in a tractor, the returns that you can get in agricultural output versus a farmer, an individual farmer with a hoe, they get incrementally out of whack.
We see AI as one of the tools where the returns are going to be so incredible – they already are.
We have a belief that it's not a positive thing for the economy or the world if all of that value gets concentrated among a very, very small set of enterprises, meaning if Google or Apple owns AI.
A big focus of what we're doing here is democratizing AI, making machine learning accessible to people and companies where it wouldn't normally be accessible.
As much as we're innovating technically, we're also democratizing technically, which is we want to make sure this just doesn't all sit in a handful of global players.
2. Seeing the World as a Browser
The other reason is a little more abstract but it's what we're building for as a company.
If you think about what we're doing today, we work with a commerce company and we're predicting the future value of all their users in 15 milliseconds.
The key source of data is that brand's website. It's the browser. Today, browsers, browser data, and app data are the source of streaming real-time data, where tons of value can be generated.
It kind of feels like the world is becoming a browser. Not only in the sense of, "We're all living in the metaverse," but everything is starting to run on real-time data the same way a browser does.
Thinking about the internet of things and wearables and RFID tags and sensors and video monitors, everything is becoming a source of real-time digitized data.
The thing about data is that it's really cool to have it, but it needs to be made sense of. That's all machine learning is. It's making sense of vast quantities of data.
If the world is becoming a browser, generating all this data; the data needs to be made sense of; we want to be that company that's making sense of all this streaming real-time data.
