As someone whose obsession with all things bot and AI related borders on unhealthy, I struggle to keep up with the ever-growing universe of bot platforms, bot building services, and NLP services.

I fancy myself something of an expert on the subject, yet when asked by someone considering building a bot which platform is right for them, I am often at a loss for words. Perhaps my terrible memory is partly to blame — I do routinely walk into rooms having fully forgotten what compelled me there in the first place. However, the deeper issue is revealed when reading how all these new products describe themselves.

About Us sections are awash with words that seem descriptive, but in reality don’t say anything. Empty terms like “fun” and “easy-to-use” are presented as though there are competitors out there who hang their hats on being boring and difficult.

Testing your pitch

I am reminded of my time working as a millennial marketing consultant in NYC. In countless client meetings, people on the other side of the table believed that the very act of targeting millennials was enough to properly endear their brand to a whole generation of people.

I was fond of using this simple litmus test with clients — take your value proposition and reverse it. Can you imagine any reasonable company making these claims? If not, you probably aren’t actually differentiating your offering.

The problem is many still think of the AI and chatbot space as being so new and cutting edge that they ignore the fact that it has become a highly competitive marketplace. Therefore, many tout the general benefits of AI technology as though it only applies to them.

This is, of course, a terrible strategy for selling in a competitive marketplace. What if Ford released its 2017 Fusion with bold claims like: “It can move you long distances!” or “Keeps you dry, even in the rain!” or “Smells much better than public transportation!”

While these may all be legitimate reasons for buying a car, it is certainly not the way to sell a car, especially not in a highly competitive marketplace.

The car market is a great example of what the AI and chatbot market needs to develop into. Even if you only know cars well enough to pick out which one might be your Uber when the license plate is still too far away to read reliably, you probably have a strong sense of which kind of car suits which type of people.

We all have preconceived notions of which cars are suited for a particular driver, whether it’s a minivan, a truck, or a muscle car. The point is, if someone were to ask for advice picking out a car, I would probably have some idea of which products are best suited to them.

Finding your niche

So how do you establish what makes your offering unique to users in the crowded AI/bot marketplace? An intense week-long Ayahuasca trip in the nether regions of the Amazon? Maybe, but for the sake of saving on airfare, let’s start with asking other people.

Surveys are tempting because they allow you to fire and forget, but for something as nuanced as this, I recommend asking people in person. This gives you a chance to read their facial expressions and see how easily descriptions are coming to them, or how much they are really digging deep to try and describe your offering.

The difficult part is picking out which people to ask. If you are operational, actual customers are the best option. I would also advise asking people you don’t have a strong relationship with, as they will feel less compelled to spare your feelings. If you aren’t live, pick people who you believe best embody your target demographic of users. The main point is, just don’t ask other like-minded developers who are working on a similar product and would never use yours.

The easy part is the questions; no need to overcomplicate these.

  • How would you describe my product to someone else?
  • How would you describe which users my product would be perfect for?
  • How would you explain to them why they should try my product?

Resist the temptation to try to finish the other person’s sentence, like an old married couple, trying to get them to say what it is in your mind. Let the person struggle to find their own words; let their “thinking pause” draw out until it’s awkward. The most important this is that you don’t taint your test results.

Another good exercise is letting another person describe your offering for you. I recommend using networking events for this, where total strangers begrudgingly ask what you do. Instead of mechanically diving into an elevator pitch, I like to jokingly insist that my friend is a master with words and is far better at explaining it than I ever could. I promise it’s a lot less awkward sounding in person — alcohol helps.

Exercises like this give you a real insight into how people perceive your offering, because now they are truly trying to explain it to someone new, not just tell you what you want to hear. Be mindful of the words they are using to describe it; it’s not a good sign if they default on just naming other companies.

“It’s like IBM Watson, but worse.” OK, hopefully your designated pitcher isn’t that drunk, but you get my point.

You may get them explaining how they feel your platform is better suited to the needs of non-developers who need a news bot. You may feel yourself push back against being pigeonholed, even wanting to correct them and explain how your platform can actually help all people. Fight the urge to interject.

Remember the car example. Can a muscle-car guy drive a minivan? Sure he can, when his Corvette is in the shop (it still is an American car, after all) and he needs to use his wife’s car to get to work. Just because it’s possible doesn’t change how manufacturers target their marketing.

AI and bots platforms large and small need to adopt this rigor if they are to thrive. Frankly, even the big players in the space are arguably not differentiated enough in the eyes of users. If the small teams are to have any luck gaining market share against all the biggest names in tech, they are going to need to take a page out of Darwin’s playbook and figure out what makes them unique.

After all, starting targeted doesn’t mean you will forever stay niche. People tend to forget that when Facebook launched, it was a way for a couple thousand Harvard students to share party pictures.  Now with over a billion users, it is one of the most widely used products of all time, meaning your weird uncle can share his awkward selfies with people in all corners of the globe.