The AI Trap
AI is Magic! (Actually, no)
Can artificial intelligence solve the hard problem of finding your next customer for you? We all WANT this to be true, because it would be truly game-changing. In fact, we want it so much that we’re willing to accept “black box” solutions that claim to be able to discern buying intent or ideal customer profile fit.
It’s the “black box” part of this you should be wary of.
Vendors are naturally reluctant to share proprietary IP, but if they won’t tell you how their algorithm works after you’re under NDA, it’s appropriate to wonder why. Maybe they’ve developed a statistical model that has predictive value for a given dataset. That’s great, but if it doesn’t learn and get better as it encounters new data, it’s just a statistical model dressed up with “AI”.
Is your AI a picky eater? Anyone who has tried to apply AI or machine learning to complex, unstructured data can encounter this problem. If your data isn’t clean and in the right format, it will probably fail to return a valid result (or any result at all). Information that points to real buying intent is inherently messy
Many systems get around this by simplifying, like ascribing buying intent to simple content consumption. Yes, there is some predictive value here, but it’s pretty weak.
Many AI systems, particularly deep learning systems, learn to recognize things in the real world by ingesting huge amounts of training data. Facial recognition is an example of this. There are now AI systems that can learn with far smaller amounts of training data, but it’s still early days.
The most compelling applications of AI are those that incorporate human subject matter experts into their process. “Human-in-the-loop” AI is what you should be focusing on. It’s certainly where we see the most near-term promise, and it’s the model we’re using to build our own AI-augmented software platform for sales & marketing intelligence.
In the meantime, finding (and more importantly, closing) your next prospect doesn’t actually require AI. It just takes a good researcher who understands how to connect the dots that predict a sales opportunity. We wrote about how one salesperson figured this out in another blog post.
You might want to check it out here