Having now observed or read over 150 pitches in the last 3 weeks and having been on the road a bit, I’m focusing on the overworked term of “use cases.” There’s a lot of de-risking left to be done in ideas that depend on too much human interaction in their routine business processes. In that light, the ones that have a heavy technology IP component seem to be more attractive, but we’re a long way from inventing the autonomous startup.

Allow me to pick on Uber a moment. Here’s a use case at the Atlanta airport. My flight from San Francisco dodged an incoming line of thunderstorms and encountered only one brief burst of moderate turbulence. However, by the time we got off our plane, traversed the concourse and the main terminal, and then hiked another half mile to the Uber/Lyft pickup point, it was raining buckets. The passenger loading areas were flooded. Worse yet, Uber drivers weren’t showing up. People were complaining about cancellations and 30-minute estimated pickup times. It seemed that the Uber drivers had decided to wait out the rain somewhere else. I hiked back to the taxi line, which was a no-wait and comfortable, albeit pricey, alternative. I had been spoiled at SFO earlier in the week where one merely walks outside on the upper level and Ubers magically appear. There is some major terminal construction at ATL that may be the root cause of the distant rideshare pickup areas, but, whether or not that is true, Uber’s system at that moment needed a human traffic controller to balance supply with demand. Their fundamental operating model of depending of the law of large numbers missed the mark by leaving hundreds of us stranded in the rain, and that was on the North Terminal side alone.

Just to pile on Uber a bit more, on my immediate previous flight into ATL, I was picked up shortly after dark by an Uber driver as his first ever ride. (I’m still waiting for my commemorative plaque.) He had just driven up from Tallahassee, FL, and had zero experience driving in Atlanta. I wasn’t worried about directing him to my place, but I felt bad about abandoning him in a metro area of 6M people spread across 8400 square miles. There aren’t many straight streets or right-angle turns in Atlanta; it has no logical grid like NYC, for example. I’ve seen Uber GPS systems struggle with the nuances of getting around Atlanta, so they’re no magic solution. I figured my driver could find his way home to Tallahassee if he eventually got on the right expressway headed south, but whether he could find another passenger in the middle of Buckhead is a big question. That person needed a driving instructor to ride with him for about a month to acclimate him to such a daunting and unfamiliar city.

The net of those two anecdotes that even the heavily funded and highly scaled market leader in ride sharing doesn’t have the processes to overcome the use cases where the drivers on the ground can’t or won’t deliver the rides.

I had another relevant experience when I moved from Austin to Atlanta. Naturally I first called a leading shipping service where smaller shipments are matched with haulers having extra capacity. It turned out that the lowest quote I got via that method was 2 ½ times the price of using the old standby Allied Van Lines. And, I must admit that Allied’s service was flawless, quoted accurately, delivered right on schedule, and no possessions even dinged. That particular online shipping service I know is doing very well, but I’m wondering what use cases really fit them. Everybody making a local move “knows a guy who knows a guy” who can handle those on the cheap. Long-distance moves would seem pretty common on I-20, the Route 66 of the interstate system. It wasn’t like I was going from Lukenbach, TX (RIP Waylon Jennings) to Hahira, GA (home of the annual Honeybee Festival). Surely many vehicles going my way had unused capacity, but online matchmaking didn’t find them for me. One has to wonder if a human dispatcher might earn his or her keep in a situation where the supply and demand seem so out of balance. When word spreads of too many mismatches, after a while the company growth curve surely will start to slow.

Bradley Horowitz, the new head of the Google for Entrepreneurs program, spoke last week at their partners meeting and Demo Day, which was my reason for visiting San Francisco. He related how he invested in a startup that had teams spread across multiple cities. Their core business didn’t succeed, but the tool they created to manage their far-flung enterprise turned out to be — Slack! I have used a wide variety of collaboration tools in that genre, going back to the early days of Basecamp, and my first exposure to Slack in one particular company I was advising was a good experience. Now I am on seven Slack’s for multiple companies, each of which has a long list of channels. Plus, many of the people I’m working with switch seamlessly among Slack, iOS texts, FB Messenger, email, Google groups, or whatever else seems handy at the moment. The net result is lots of notices on my iPhone screen of incoming messages, followed by a process of calling up those notifications and trying to track each one to its particular carrier. Then comes the question about where was a certain file appended? You know you saw it zoom by, but where do you look first? One of the stage presenters at the SXSW Accelerator was a company which purported to solve this by linking all messaging services together so that no matter which one you wanted to use you’d find anything addressed to you from any source. They illustrated their show with a complicated plumbing diagram of interconnecting and overlapping pipes. While that would be a highly desirable product, frankly, to me it seemed to be a “pipe dream” (couldn’t help that). How could one possibly get around all the security issues and constant versioning of each individual messaging service and make them work across multiple companies to everyone’s satisfaction? One can’t fault Slack for the creativity of its user base that has led to such complexities, but the original use case has been lost in the rush. The masses have found a way to turn a neat and orderly tool into chaos.

The net of all these rants is that as you design your product or service, you really must peer around the corner to the use cases that are not so easily anticipated. You must decide how you are going to populate your management team with the right operating skills to deliver a dependable and compelling experience for your ultimate providers of your revenue. You can’t let rain showers put you out of business for an evening when you have many customers begging. You can’t let your users run amuck with your product to the point it loses much of its utility. When people who are behaving rather normally catch you unprepared, it’s your problem, not theirs.