What we could have tested with a phone number and a spreadsheet

Sending a car for service costs you a morning. You drive in, you hand over the keys, and then you are standing in a service centre in Ampang with nothing to do until someone calls your name. The alternative is to bring a friend in a second car, drop yours off, get a ride home, then repeat the whole thing in reverse that evening.

I got tired of it. That was the entire origin of RocketValet.

The idea was simple enough to explain in one sentence. You tell us where the car is and where it needs to go, we send a screened driver to collect it, and we bring it back when the work is done. We started operating in December 2015, three of us, each coming from a different professional background. My co-founder from RePark was one of them.

What it actually was

The mechanics were unglamorous, which is usually a good sign. You gave us the car model, the rough distance and the drop-off address. We quoted you. If you agreed, a driver came.

We charged RM1 per kilometre and RM15 an hour for waiting time. We partnered with a handful of workshops around the Klang Valley, a Perodua service centre in Ampang among them, so a customer who did not have a preferred workshop could just take our recommendation and a fixed price. If they did have one, we drove there instead.

Most of our drivers came from Uber or Grab. They already spent their days moving other people’s vehicles around the city, and they were already used to being tracked and rated. We screened them and gave them an app so the customer could follow the car in real time.

Vulcan Post wrote about it in June 2016, and the coverage was kind. Amanz and The Vocket both ran it in Malay, which mattered more for reaching actual car owners than the English coverage did. We turned up in a round-up of Malaysian startups going after drivers, and the company still has a profile sitting on e27.

Then we got into MaGIC’s ASEAN Accelerator Programme, Cohort 2. Demo day in August 2016 had 42 startups pitching, from Malaysia, Taiwan, Vietnam, Australia and Sweden. Afterwards Vulcan Post published its sixteen picks from the 42, and we were on the list.

By every measure available to us at the time, it was working.

The number nobody asked about

Here is the number that sat in that same article, in plain sight, underneath the good news.

Thirty to fifty bookings a month. Around 200 to 300 customers in the first seven months.

Those are not embarrassing numbers for a young service business run by three people. They are real customers with real cars paying real money. But fifty bookings across a month is under two a day, and at RM1 a kilometre a typical run to a workshop and back was worth a small number of ringgit to us before we paid the driver.

I spent that period thinking about how to get more bookings. That was the wrong thing to be thinking about.

We built the tech first

We were a technology team, so we did what technology teams do. We built the thing.

A booking flow. A driver app. Live tracking so the customer could watch their car move across the city. Screening, ratings, dispatch. All of it working, all of it before we had any real evidence that a meaningful number of Malaysians wanted to hand their car keys to a stranger on a weekday morning.

Here is the part that took me years to admit. You did not need any of that to test this business.

Three of us, a phone number, a WhatsApp group and a spreadsheet could have run the first two hundred jobs. Quote by message. Dispatch by phone call. Track the driver by asking him. It would have been ugly and it would not have scaled past a few dozen bookings a month, which, as it turned out, is exactly where we landed anyway.

That version costs a few weeks. Ours cost far more than that, and the expensive part was not the money. It was that we spent our best months building instead of finding out.

Most businesses can be tested without the technology. The technology is what you build once the answer comes back yes, and we had it the wrong way round.

What makes sense is not evidence

The deeper problem was not the sequence. It was that we never seriously validated the business at all.

The idea made sense to me. I had the problem myself. Everyone I described it to nodded, because it is an easy thing to nod at, and I took that nodding as a finding. We did not study the market properly. We did not work out how many people send a car for service often enough to change their habit for it, what they would actually pay, or whether the annoyance was big enough to beat simply asking a friend for a lift, which is free.

That is the mistake, stated plainly. I decided what people wanted because it made sense to me, and something making sense is not the same as it being true.

The market is the only thing that answers that question. You test, and it tells you. It does not care how logical your reasoning was on the way in, and it will not tell you anything at all while you are still building.

I made a version of this mistake with RePark too, where I put my effort into the plate lookup, which was easy, and barely thought about getting strangers to trust the app, which was the actual business. RocketValet was the same error wearing a different product. It took seeing it twice before I understood it was a pattern in me rather than bad luck in the market.

The person we did not have in the room

There is one more thing, and it is the one I would change first if I could go back.

We had no experienced business person around us. Three founders from technical and professional backgrounds, all roughly the same age, all reasoning from the same starting assumptions, none of us having run a real operation before. The accelerator gave us structure and a stage. What it could not give us was somebody who had done this for twenty years sitting across the table asking the unglamorous questions early.

How many of these does a customer buy a year. What does one job actually earn after the driver. What happens to your margin when you double the drivers. Who else has tried this, and where did it stop.

None of those questions need a technologist. They need somebody with scar tissue.

I understand now that this is not a mentorship gap. It is a governance gap, and it is the same function a board performs for a company large enough to have one. Someone whose job is to be unimpressed by the pitch and interested in the arithmetic. We did not have that, and at that age I am not sure I would have listened anyway, which is its own lesson about who you let near your assumptions.

Today I sit on the other side of that table for other people, and the questions I ask are almost exactly the ones nobody was asking us in 2016. The service ran, the press was real, the accelerator was real, and the customers were real. What none of it told us was whether we had a business, because we never actually asked.