What used to cost millions in 2017, can now be solved with machine learning for $499/month
Companies spend tens of millions of dollars each, every year, on trying to improve the likelihood their customers will engage with something, or buy something, by targeting the customer dynamically. Content personalization, ranking and recommendation are critical to so much of the digital economy, from Netflix recommendations to picking the nearest Lyft driver.
According to a survey put together by Evergage, Inc.:
Respondents continue to report substantial improvement or “lift” from their real-time personalization efforts, with 88% stating they see measurable improvements due to personalization, and 23% reporting an improvement above 20%. This success leads to increased investment, as 46% of respondents expect their budgets for personalization to increase in the coming year.
And furthermore, according to McKinsey.com:
Personalization can reduce acquisition costs by as much as 50 percent, lift revenues by 5 to 15 percent, and increase the efficiency of marketing spend by 10 to 30 percent.
Its actually a tough problem to solve, as there are multiple layers of complexity as it relates to understanding the customer, picking from potentially millions of things to show, to ranking items by most-likely-to-engage.
This diagram demonstrates a basic outline of the architecture behind these concepts.
How do you know which video, which product, which ad, which news post to show to the customer first?
Part of the answer comes from understanding your customers. But that problem has been pretty well solved. Most E-Commerce, video streaming, and other types of sites have a good grasp of who a given user is, where they came from, and what they’ve engaged with in the past.
But companies pay a lot of money to cross reference that data, with past behavior, and use it to somehow generate a ranking of things that user is more likely to engage with.
It turns out, machine learning is a great way to solve this problem.
Learning customer’s behavior, then applying that learning to some predictions, is pretty much machine learning 101.
This is why Machine Box made Suggestionbox. We wanted companies to have access to these powerful, machine learning-enabled capabilities, running in their own environment, on their own data, without having to go through the usual expense and time-drain of figuring all of this out themselves.
You can read more about what Suggestionbox is and how it works in this blog post:
But the key takeaways for any enterprise or startup trying to solve this problem for their millions of users, is that you can now increase user engagement by huge factors, for a trivial monthly fee of $499.
That’s a huge shift in the ROI mechanics of applied machine learning, as it no longer constrains use by charging per API call or putting limits on the numbers of requests.
Companies can run this tech in their own infrastructure, behind firewalls, on their own data. Most importantly, they can innovate, experiment, and try new things without running up a huge cloud bill.
This is going to be game-changing.
What is Machine Box?
Machine Box puts state of the art machine learning capabilities into Docker containers so developers like you can easily incorporate natural language processing, facial detection, object recognition, etc. into your own apps very quickly.
The boxes are built for scale, so when your app really takes off just add more boxes horizontally, to infinity and beyond. Oh, and it’s way cheaper than any of the cloud services (and they might be better)… and your data doesn’t leave your infrastructure.
Have a play and let us know what you think.
