Why Deep Learning is the most important discipline in Product Management to tackle right now

Deep Learning in Product Management

I can’t really say what the exact trigger was for me this week, but I had a huge flashback to 2010 when I started to go down my path as a mobile product manager.
Adding the only ‘minor’ difference that I actually had abandoned the concept of ‘disciplines’ in product management a while ago, that mobile is now absolute commodity and we’re confronted with technologies like deep learning and neural networks which will shape and iterate features autonomously.

While the challenges emerging from this shift potentially couldn’t be bigger for my professional development, the circumstances leading up to this point are frighteningly similar.
By 2010, a lot of people were already using smartphones and their potential for daily consumer needs was clear to see. Nevertheless, the impact on existing business models was neglected my many companies. Leave alone the rare number of people in product, design and development being able to actually build and ship excellent mobile products.

While it may not seem as obvious to everybody how omnipresent deep learning is already in some of the world’s most used digital products (my favorite example from are the suggested replies in Google Inbox) it kind of feels the same to me. The value for users is already somewhat tangible, but few companies beyond the  GAFA players can see and adapt to the potential for their business model. Leaving the complete lack of talent out of scope of this equation as well for now.

So, even though I myself was a ‘domain’ product manager (being focussed on a certain discipline), I personally rejected this way of thinking about 1 year ago. It wasn’t a particular experience which led to this insight, but rather the professionalisation of product management in general.
I then believed (and still do to a certain extend) that a good product manager is able to deploy it’s skills against every discipline she gets presented with. Whether it’s ecommerce, payments or mobile.

The domain-specific knowledge can normally be learned fairly quickly. But before for example mobile became a natural part of all digital products, it was common sense to have mobile specific product people which knew the pitfalls and levers of mobile operating systems. The ‘other’ product managers were then left focussed on their existing products and rather asked for consulting from the specialists.
Over time, the advantage of mobile (and other domain-related) product managers became less relevant and working in disciplines is more like a simple addition to your job title.

But now I think we’re at the threshold of seeing the necessity of domain focussed product managers again. Products which will be mainly driven by deep learning algorithms not only need skilled engineers, but also product managers and designers which are familiar with those technologies.
While I will touch on some of the specifics of what this shift means for the tasks we’re facing today later on, I think the biggest changes will take place in how we’re going to discover and test the critical hypotheses of new products.

If you’re sharing this gut feeling (of course it’s not more then that at this point) I highly recommend Ken Norton’s perspective on what machine learning means for product managers.
And while you’re at it, why don’t you join me for this great educational series on coursera?