Hey startups, here’s why you should build your product like a Rube Goldberg machine

Editor’s note: Joe Procopio is the Chief Product Officer at Get Spiffy and the founder of teachingstartup.com. Joe has a long entrepreneurial history in the Triangle that includes Automated Insights, ExitEvent, and Intrepid Media. Joe writes a column exclusively for WRAL TechWire. It is published on Tuesdays.

+++

RESEARCH TRIANGLE PARK – A lot of great startup advice comes directly from parenting. And if I were the kind of parent who gave parenting advice, a lot of my parenting advice would probably come from startup.

I’ve been a parent for over 17 years and an entrepreneur for longer than that. But even with all that experience, I don’t feel like you need to hear about how I talk to my kids about failure, so I don’t write about the parenting angle very often.

That said, every once in a while, a parenting scenario perfectly crystallizes a particular bit of startup advice, making the advice super easy to understand, remember, and follow. Especially when the anecdote involves me uttering the term “Minimum Viable Product” while helping one of my kids with a science project.

Joe Procopio (Photo courtesy of Joe Procopio)

My goal isn’t to get you to build a Rube Goldberg machine, far from it. It’s to get you to make good decisions while you’re building your product.

Here’s what happened

My youngest got assigned a science project to build a Rube Goldberg machine, a device named after a cartoonist who drew up elaborate plans for systems that performed simple tasks in purposefully complicated ways.

My son’s lab partner was dropped off at my house early Saturday morning. I spent the day out of the house doing various things, then I returned to find him in the middle of his room with cardboard and string and marbles and junk everywhere.

He had failed, spectacularly and with great frustration, to get even the first half of his machine working. He was now alone and surrounded by a mess of his own making, with steps of the machine either not working at all or only working under the most perfect conditions.

Every startup winds up in this scenario at some point. If this was the intent of his science teacher when assigning this project, then well played, dude.

Lesson 1: Build one modular piece at a time

Here’s how most products are built: Great designs are drawn up to create a dozen or more major feature sets, all of them linked, all of them working together in perfect harmony to complete one awesome use case for the customer.

My kid and his lab partner put (actual) pen to (actual) paper and drew up a great design to run a marble through a dozen steps in their Rube Goldberg machine. At the final step, the marble would drop onto the spacebar of a laptop, which would play a queued-up pop song that all his classmates were into.

I don’t know, Spice Girls or whatever.

Here are a few of their mistakes:

They designed the entire system from beginning to end, none of the steps were modular.
They chose their goal (play the song) at the outset, and they stuck to it.
They designed way more steps than the assignment called for, because they wanted to impress everyone.
They chose the functionality of each step because that functionality was awesome, not because it was necessary to move the marble from step to step.

Now, I understand that the point of a Rube Goldberg machine is for each step to be inherently complex (and awesome). That’s fine. But the first thing we did was take it all apart and focus on building each step independently.

We stuck to the original design, but we agreed to be flexible. The functionality of each step could be as awesome as they wanted, provided that awesomeness resulted in a repeatable outcome. In other words, at the end of all that awesomeness, we needed to know exactly where the marble would be.

Lesson 2: Decide what to build first

This is where the term “Minimum Viable Product” entered the project. Almost every startup struggles trying to decide which piece to build first.

If the opportunity is a huge market gap or some other scenario in which there are a lot of customers waiting for a solution, the first step is usually the one that gets built first. This is because the most critical step is capturing market share. The market drives the solution.

On the other hand, if the opportunity is to solve a costly problem with a new solution, that solution, the last step, usually gets attention first. The solution drives the market.

However, those two situations are rare, and most viable business opportunities fall between the two. Thus, focus is usually put on both ends of the machine: Find a fertile initial market and give them a robust (and awesome) solution.

But not being able to determine the right balance between the two ends kills most early startups.

So the prudent thing to do is build the most critical step first. In other words, start by building the step that provides the most value to the widest range of customers and is the hardest to pull off. This will give you a lot of information about how valid the idea is, how difficult the execution will be, and even how successful the revenue and margin equations will be.

Once the most critical step is solid, you can tack on duct-taped versions (in our case, literally duct-taped versions) of the rest of the machine, and get an idea of what your product is going to look and act like. This is your MVP.

Lesson 3: Test your viability

Our end goal was to impress all his classmates with a push of a spacebar on a laptop that would play a song (I’m compelled here to add that it was not the Spice Girls). Because the last step required a certain amount of accuracy and force, we built the next-to-last step first.

Turns out the force and angle needed for a glass marble to depress a laptop spacebar with any accuracy is the same amount of force that makes me nervous about damaging a $1,000 laptop. Noting there was not a $10,000 prize waiting at the end of the science project, the risk far outweighed the reward.

So before we even built our first and most critical step, we realized we needed to pivot, and we did just that. We replaced the laptop with a toy that made Darth Vader talk when you pressed a big, easily-pressable button. Equal value to the customer, risk reduced by $993.01.

As you’re building, continue to justify the start and end points, as well as every point in between. The sooner you catch pivot-mandating moments, the better.

Lesson 4: This is what regression testing is for

As we built additional steps, we tested each step individually, then slotted the step into our MVP and tested our growing machine thoroughly. This is regression testing, and it’s one of the most overlooked steps in the process.

By the time we got to the third step, my kid caught the “close enough” bug, and I let him run with it. We minimally regressed the next few steps, resulting in a lower and lower success rate. After that, it took a long, long time to figure out which of our newly added steps was contributing how much error to the error rate.

Lesson 5: Check your team, manage your time

Now, my son’s lab partner is a good kid and a smart student. He’s also a good guy to have around – funny, energetic, and creative.

But, like my son, he’s an idea person.

Earlier that day, they spent a good couple hours dreaming up their perfect Rube Goldberg machine, taking extra time to invent the most awesome steps they could. They used color markers and a bunch of posterboard to draw up their plans. They used science, lots of it, to justify the awesomeness of each step.

And they spent maybe the last hour of their allotted time actually building.

On the other hand, I probably turned his fun science project into a slog, but as he saw progress, he was energized by the goal. This taught him that he had to bring the same enthusiasm for brainstorming new ideas to a task as mundane as holding an incline in place for minutes at a time to figure things out.

And to be fair, I didn’t do any of the project for him. In fact, when his lab partner returned the next day, I got them started and then popped in and out of the room to check progress and sling advice.

Make your own joke about what my role was. I’m going with “board member.”

Pre-launch and launch

Their Rube Goldberg machine was to be demonstrated at an assembly the next morning, so the last thing we did, once all the testing was done and all the glue was dry, was very carefully test walk the machine from his room to the car. This was our test launch to make sure the system held up under the duress of production.

The actual launch was a smashing success, and all his classmates appreciated the Darth Vader quote (“Impressive”), even if they didn’t appreciate all the work that went into getting there.

+++

Hey! If you found this post actionable or insightful, please consider signing up for my weekly newsletter at joeprocopio.com so you don’t miss any new posts. It’s short and to the point.