Napkin Math Before the Business Plan

The fastest way to test a business idea isn't always to build it. Sometimes it's to find out whether the economics deserve to be built around.

I've always believed that some of the best business ideas start with a few numbers and a willingness to test them before everything is figured out.

Not a five-year forecast. Not a perfectly designed customer journey. Just enough math to understand what needs to be true for an idea to work.

Then get something into market.

That last part is where a lot of businesses slow themselves down.

Find the Economics Before Building the Experience

Years ago, we explored generating leads in the assisted living market.

We didn't have a complete experience. We hadn't built out the destination pages or secured the downstream distribution needed to monetize the leads.

But we had a hypothesis about customer acquisition cost.

So we tested it.

The initial experience had limitations, including places where the journey effectively ended. That wasn't a good long-term customer experience, and it certainly wasn't something you'd want to scale.

But the experiment helped us answer a critical question before investing in the rest of the business: Could we acquire the right audience at a cost that made the opportunity worth pursuing?

Once we understood the acquisition economics, we had a much better basis for deciding what to build.

We took a similar approach when exploring prospective pilots interested in discovery flights.

We used a lead magnet to attract people interested in learning to fly. When someone expressed interest, we didn't need a sophisticated booking platform or a network of integrated flight schools.

We manually coordinated discovery flights with schools in their area.

It wasn't scalable. That was the point.

We could learn about acquisition costs, intent, conversion, and the work required to fulfill the service without building technology around assumptions.

Build the process manually. Prove the economics. Automate what earns the investment.

Let the Math Tell You What to Test

The model behind these experiments was simple.

Start with the potential value of a conversion. Work backward to what you can afford to spend acquiring someone. Then calculate the conversion rate required to make the economics work.

Break acquisition costs into their drivers: channel, device, audience, and distribution. Understand how those inputs affect conversion and what you can reasonably control.

Early on, we often kept audience targeting broad enough for advertising algorithms to find demand while maintaining tight controls on cost per click.

As the data improved, we could move optimization further down the funnel, toward target acquisition costs and actual conversion outcomes.

You don't need to forecast every scenario. You need a plug-and-play model that shows how changing one assumption affects the result.

What if acquisition costs rise? What if conversion improves? What if a more expensive channel produces substantially better customers?

Those questions tell you what to test next.

Don't Make the Tail Rescue the Business

We generally started with a linear model: someone arrives, moves through the experience, and converts within the session.

Ideally, those immediate economics could carry the acquisition cost.

Then we considered the additional value of having a first-party relationship.

Someone might return later, respond to an email, or convert through another relevant experience. That future engagement can improve the economics considerably.

I've seen that tail become more valuable than initially expected.

But I wouldn't give it much credit in the first model.

If your idea only works because of hypothetical future conversions, you may be building a financial justification instead of testing a business.

Make the first interaction carry its weight. Let the relationship improve the economics.

A Data Point Can Be Worth More Than a Feature

There's a difference between a product that's ready to scale and an experiment that's ready to teach you something.

A startup doesn't always need the former to benefit from the latter.

Sometimes a manual process is enough. Sometimes you don't need the entire funnel built to test whether people want to enter it.

The discipline is knowing exactly what you're testing, what you can legitimately offer, and what the customer should expect. A limited experiment should never become an excuse for misleading people or collecting information you can't responsibly use.

But waiting for every page, integration, and operational detail to be perfect can be just as dangerous to a young business.

You can spend months solving problems that only matter if the economics work.

And even a high acquisition cost isn't necessarily disqualifying. A more expensive audience may produce substantially better downstream results. You just need enough evidence to justify paying for it.

Earn the Right to Build

The objective of napkin math isn't to create a forecast that looks convincing.

It's to identify the assumptions that could kill the idea, then find the cheapest credible way to test them.

Can we acquire demand? Does that demand translate into meaningful intent? Can we fulfill the promise? Are the unit economics attractive enough to invest further?

Some experiments will fail. That's fine. The important thing is finding out before you've committed months of development and a significant amount of capital.

A spreadsheet can help you understand what needs to happen.

A small experiment can tell you whether it might.

Don't spend six months building something you could have learned about in six days.