<i>1. The potential partner turns you down because they decide to build a similar product themselves. This happened to us a number of times. I think part of the reason was that there was a lot of market buzz around “big data” and machine learning which lead to the perception – rightly or wrongly – that those capabilities needed to be owned and not rented.</i><p>This.<p>Doing NLP and ML consulting through MetaOptimize, this is our biggest barrier to acquiring clients. As Chris Dixon says, ML and NLP are "hot" right now---for good reason---but this hotness can get in the way. So a lot of people want NLP/ML, but want it in the wrong way.<p>At one extreme are companies that want something difficult and highly custom that's strategic, but then decide they'll try to figure it out in-house. That's fine, because if they can't do it right and it's truly necessary, we can help them in the future, or train them as they build it.<p>At the other extreme are people that want a baseline implementation for something strategic. This gets strange when they want a textbook implementation as a prototype, but then want to entangle MetaOptimize with onerous contractual conditions (non-competes, exclusivity, etc.) while paying the (low) price for a prototype. I can't tell you about numerous companies who want to be the only company in their space we provide LSA for, or a baseline recommendation algorithm for, and don't want to hear that exclusivity comes with a price. You can't hire us for a five-hour project and then expect to own us in some space.<p>And, in the middle, are companies that have a NLP/ML need and want to solve their need in a reasonable way. Either it's strategic/core and they will pay market value to own it, or their need is not strategic/core and they are comfortable with licensing it.<p>Just my experience. Even though Chris Dixon is talking about one product (hunch) and its backing technologies, everything about this piece rang true for me.
The more important lesson of Hunch is how often bizdev deals lead to acquisition.<p>Very rarely does a startup directly benefit in a major way from working with a big company. Quite often though, if the big company is smart, they can get a bargain on a team/tech that creates value in ways they couldn't have themselves.<p>This kind of thing happens all the time. Another example is Reddit. They created a whitelabel version of Reddit for Conde Nast (lipstick.com IIRC), which failed, but lead to them being acquired.