That's No Tech Startup: That's a Factory
Digital tech startups often need physical touchpoints. These can be called a "hub", a "warehouse", or even a "factory". How we operate physical touchpoints can be what makes these startups thrive — or crash.
The dream of many startup founders is to build an app, a smart back-end, and maybe some clever cloud-connected hardware, and then watch from the sky as profits come rolling in.
This is how Uber and Lyft started, after all. It was people sharing cars via apps. (Or later, like taxis, but with apps.)
Scooter companies like Wind and Lime had a similar idea, except with internet-connected scooters. The idea was that you'd leave scooters all over the city, always know where they are, and use algorithms and incentives to get people to move them around.
The reality of running startups like Uber, Lyft, Wind, and Lime with physical operations is quite different. You think you just need an app and a back end, but you always end up renting physical sites to manage customers and inventory.
Let's go through some case studies of this.
Like this article on tech operations? Subscribe for more!
Join the mailing list. Days of work go into each post.
Ride sharing: Lyft, Uber, and Bolt
Ride-sharing apps started out as an app and a matching system. Drivers go online and serve passengers who request them.
The initial idea of ride-sharing companies was that if there weren't enough drivers, the price system would adjust automatically. No drivers downtown, but a ton of passengers? Prices would skyrocket. Then more drivers would come online, and the system would balance itself out automatically.
We quickly realised that you couldn't run the whole Lyft/Uber business with automatic incentives. The momentary price spikes were outrageous. To smooth them out, we had to have a bit of predictive planning, done by someone who knew the peak times and areas of each city. Someone who knew the major events, and someone who had an eye on the whole marketing strategy. The first General Managers were hired, and their operations teams came quickly afterwards.
We began doing more and more things non-automatically, with physical touch-points. These included
- Onboarding. We figured out it's much more likely a driver will end up actually doing rides if we onboard them in person and coach them through the process.
- Customer service. Often, driver's problems are sensitive and complicated. They might involve a lot of documents. The driver might not have a great computer, or be very technically literate, or just might need to explain it to a person. So we started doing customer service in person.
- Vehicle inspections: In some states we had to do inspections of cars to sign them off as potential ride-share vehicles. In California it was mandated to be done by a certified mechanic.
- Rentals. Lyft opened its Express Drive business and started renting cars out to drivers. Obviously this had to be done in person, because someone had to physically take the keys.
All of this led to our physical locations, which became the "Retail" arm of Lyft, and the "Greenlight Hub" arm of Uber.
E-Commerce: Groupon, Amazon, Rent-the-Runway
Any online business selling things needs warehouse space (unless it's totally outsourced... but costs for that can add up quickly).
How efficiently a supply-chain/logistics operations can run has a direct impact on customer satisfaction. While in Australia delivery time of a week is still acceptable, anything more than two days is unacceptable in the US, and China has become accustomed to same-day delivery from JD and TaoBao.
Rent the Runway suspended customer operations in late September, 2019, because of "significant operational shifts". Their warehouse operations could not fulfil the (significant and ever-growing) demand from customers in a timely way, which meant that customers were receiving their clothing late. Rent the Runway's customers rely on timely delivery — they often want a dress for an event. So delivering things late was extremely bad news for the company.
There's nothing surprising here. But what surprises me is how often small companies are caught off guard by the difficulties of logistics. From one person trying desperately to fulfil orders for a huge Kickstarter campaign, all the way to a billion-dollar company trying to send dresses to many people in holiday season, smart people are often on the back foot when it comes to physical operations.
Planning from the outset to be ahead of the curve in physical ops is what makes companies like this survive or die.
Scooter sharing: Wind, Bird, Lime, and many others
Finally, all scooter-sharing companies own all their own assets — fleets of tens of thousands of scooters (or, sometimes, bicycles).
Owning assets means that these startups are worried not just about revenue, but also about depreciation and maintenance. Bird, Wind and so on are all worried about keeping their scooters alive for as long as possible.
In principle, the basic idea behind scooter companies was: Buy a scooter for about $400 wholesale and rent it out for about $5 a ride. They got 3-4 rides a day on average, meaning $15 a day in revenue. You can pay someone to recharge them for $5 a day, meaning $10 in gross profits. After forty days, it has paid for itself!
This was the dream on which investors were sold. But there were some critical things that were missed.
- People who rent scooters abuse the shit out of them. Sorry for the expletive, but it needs to be said. They throw them off curbs, and sometimes literally in the ocean. They get busted up within about ten rides — every few days.
- You can pay $5 a ride for a regular recharge, but what about scooters that are really far out? Ones that mysteriously can't be found? Ones that have been trashed or stolen? You can pay your people more, but there's an economic trade-off.
- What's the cost per unit of theft? How can that be minimised? You might imagine that your team will be fine just calling the police or letting thefts go, but prepare to be surprised. Staff have pride, and they don't like people ripping them off.
- Scooters still break under ordinary conditions. They get blown over in the wind. People will drop or crash them. Cars will run over them.
I could go on, but that's should give you an idea. Basically, e-scooter maintenance and loss adds up to over 25% of the cost per ride. I'll spare you the math (for now), and it does depend on the particular model being used, but 25% is an optimistic low-end.
Scooters startups end up looking a bit like this:
OMG what is all that stuff?? Well, some of it is caffeine (trademark coffee cups of Israeli coffee, cafe shachor).
But mostly it's the result of having to manage constant repairs of scooters.
Every night, a large percentage (between 5% and 20%, depending on seasonality and luck) of a fleet of scooters can come in the door with problems including
- Won't charge
- Some system error meaning it won't lock or unlock
- Broken controls (brake lever, accelerator, even just grips)
- Bent frame
- Damaged motor
- Damaged brakes
- Poor driving (e.g. imbalanced, uneven wheel)
- and so on...
Staff of a warehouse have to learn how to triage, categorise faults, assign them for repair, and then assign a workforce to repair them systematically to get them out the door.
They also have to learn to manage parts inventory, keep a well-organised warehouse, and conduct all the repairs of course.
The problem: most people hired into startups have no experience in managing a workshop. They may have worked in one, or seen them. They're definitely bright and able to learn quickly and have all the requirements we hire for in operations talent. But they still have no idea.
Add on to that the constant chaos of daily operations (charger teams getting into disputes and sometimes quitting, conflict between a company and authorities, or just changing operations strategy) and you have a recipe for headaches.
What's the solution? Plan ahead
I'd say the solution is "hire me as a consultant", but it's easier than that!
The first step is to be realistic about future infrastructure needs and bake them into your plans. Don't think of ops as a temporary measure until you optimise it away. Build it well, and build it to last.
The second step is to hire the right people. Flexible, bright, self-starting. If you get over-pedantic about building exactly the right infrastructure and hire for specific roles you might build yourself into the corner. But if you hire ops generalists who are on-the-ground problem solvers who can report up and down the chain with ease you will set yourself up for the future.
Finally, don't outsource all this. I see too many divisions of companies outsourcing bits of ops like logistics and deployment logic... when this constitutes a large part of their business model. If a scooter business is made up of a cool scooter product, an efficient matching algorithm, and scalable operations that put them on the ground, why would you outsource the last part? It's a huge and difficult to replicate part of your business.