It’s Not Magic. It’s Data – AI for Business – We feel like we need hardcore experts to harness these strange powers of AI. Don’t despair all the buzzwords and chatter, are actually buzzwords and chatter The tools for business AI are already out there. There’s no need to build them yourself. All they need are data that you already have. AI is just a complex set of algorithms with no magic involved.
Hey everybody! Today I’m gonna be handing the floor over to Alina who’s been using AI and machine learning. She’s been using machine learning and AI for marketing and growth. She’s been teaching it and applying it. Alina is gonna help us take the first steps of demystifying what AI is and how you can apply it to machine learning and growth. So you have data. Are you using it effectively? Probably not…enter Artificial Intelligence and machine learning.
Yeah and what’s really interesting here is that as a marketing person or as a business person you can actually start using AI right away for your marketing and growth without prior coding knowledge. And you must think we’re crazy…bear with us. You’ve probably heard some pretty crazy and scary stories recently. Headlines read that Facebook developed an AI that spoke in a language that no one could understand.
Self-driving trucks are taking over our jobs Or this bot that helps scammers write emails… This list goes on and might seem pretty overwhelming. We feel like we need hardcore experts to harness these strange powers of AI. Don’t despair…all the buzzwords and chatter are actually buzzwords and chatter the tools for business AI are already out there. There’s no need to build them yourself, all they need are data that you already have.
AI is just a complex set of algorithms with no magic involved Yes, I know that algorithm might be another scary word for many of us in marketing but it shouldn’t be. An algorithm is just a set of rules that a computer follows to make a calculation so…
So let’s throw some questions out there that AI and machine learning can actually answer for you. Now if any of the answers to these questions are strategically important for your business you should start thinking about taking first steps with machine learning. For how long will your customers use your product or service. How likely are these customers to stop using your product or service and when?
How much will your customers buy through their whole customer lifetime? How do we match the right message, the right content, with the right person at the right time? Which leads our customers should we spend more time nurturing. Which leads or customers are just a lost cause? Well, in fact, you can run a model to answer some of these questions with just 15 strokes of a keyboard. We do this every day now and we all learned within a few hours. We are not all data scientists and you don’t have to be either.
All you need are some glorified tools and no coding. Run a model and see AI in action for yourself. Now, imagine you’re head of a start-up that does people’s laundry. You have access to some information about your existing users: Age, gender, relationship status, favorite activities the number of times they do their laundry, their job function and years of experience.
And other similar features. And this is cool because Alina run into an almost identical use case a couple of weeks ago and what’s great and reassuring here is that Alina didn’t have big big data. She had like medium data, she had 800 customers that had been collected over nine months. When you thought about your
business persona before you assume that “Age” and “Relationship status” are the most important features to differentiate your customer.
You run a cluster analysis: this is used by data scientists to find patterns in large sets of data, and by marketeers it’s used to make segments. So you’ve done this and now you realize that “Age” and “Relationship status” are not actually the primary indicators of your users behavior. It’s “Sports activity” and “Years of job experience”. Imagine your marketing department had never thought of that. That’s kind of a big deal.
Maybe you’ve been segmenting your customers based on the wrong criteria the whole time. And that’s the power of machine learning versus statistics while good old statistics is a little more static and it describes the data that you have. The power of machine learning is the fact that it can learn from every facet and every piece of your data. This is a big deal!
You can’t think about 20 characteristics at the same time in your head and draw correlations between all of them. But machine learning can! Now you can take your machine learning segments and try testing and customizing your messaging to different groups. You set up some ads on Facebook and you start
targeting “sporty people” with “athletic messaging” and “regular messaging” for your product. Click throughs are 16% higher on the sporty messaging. That’s great you’ve validated part of your AI learnings.
Now you can start A/B testing with careers which was second and your most important variables. You have so much to work with. You can now test, test, test and retrain your model on more data. Now you can probably start seeing that AI doesn’t just require data. It requires clean, useable and relevant data. It’s as if you’ve started training a young person at your company and they’re starting to grow up. And this is probably one of the most important parts of AI:
AI is a symbiotic relationship between your data and you the business minded human. How you collect the data and what you do with your findings. Is the human input AI desperately needs. Your AI models won’t
tell you anything if your data is dirty and unusable. So start by asking yourself what questions you want to answer? and what data you need to answer those questions?
reference – It’s Not Magic. It’s Data – AI for Business
Web enthusiast. Thinker. Evil coffeeaholic. Food specialist. Reader. Twitter fanatic. Music maven. AI and Machine Learning!