The Zen of Building Financial Models: 9 Tips for Startup Founders

One. Build it to bin it. 

Your first model, your first set of answers, are going to be wrong. Remember that. So there is a reasonable chance you are going to throw that away.

If you are likely to trash version 1.0, don’t spend too much time building it.

Use the model building exercise to understand your business, explore it, make mistakes, and when you are done, bin it.

Very few of my plan A’s worked across 30 years. Very few of your first passes on a model are likely to work either.

That doesn’t mean you don’t build them. That only means you try to work through them as quickly as possible.

Build your business. Not your model.

Keep that in mind. Don’t spend too much time cleaning it up. Get what you need and then get out.

Two. Build models from the founder’s perspective, not investors.

Investors don’t mind. They like to see us founders think. If your models help them walk through your decision-making process, they will be happier with it.

We make a tactical decision every day. Build models that help you make better decisions.

There is another way to look at this. Numbers tell great stories.

Figure out what your story is, then tell it.

Growth in a down year? Scale? Increasing sales volumes? Faster servers? Bigger teams? What is the story behind the move?

Draw it out on a model and then tell it. It’s easier to map models that tell stories into financial statements, then the other way round.

Models that are data-driven, that consider a founder’s daily decisions making process. That can be updated easily when more data is available.

We make decisions every day as founders. Build models that help you make those decisions.

How do we get more sales? More traffic, more orders, higher ticket sizes, advertising campaigns, different products. One of the above, all of the above.

The right answer? I don’t know. Build a model and find out.

Three. Create sub models.

Take complex models, difficult models and break them into simpler, shorter models that are easier to work with. Divide and conquer.

Simple example. You are modeling revenues; what do you need? A model for your sales pipelines, a model for traffic and conversion rates, a model for order sizes, a model for processing, shipping, and payment costs, a model for growth, a model for customer retention, churns, and returns.

Don’t factor all elements into the model in one go, in one cell or formula. Focus on building the model one sub model at a time — separate models for sales funnel (conversion rate, click-through rate) human resources, inventory, growth, etc.

This not only simplifies the model building process and makes it easier to trace errors but shows investors how you think and the choices you make.

Easier to build, easier to explain, easier to understand, easier to debug.

Four. Build in dynamism. 

Use dynamic assumptions, don’t hard code values. Sensitivity test assumptions and results, evaluate the interaction between assumptions, and use them to make strategic decisions.

You can do this with data tables and if you are in the mood for a bit of rocket science, with Monte Carlo simulations

Five. Use real-world data.

Calibrate your model to real-world assumptions and results. Real data, real results. Ask yourself is this how the real world works. You have two worlds. The world as it exists in your head, your imagination. The other world is the real world. As it exists.

Build a bridge between the two worlds by using real-world data.

Six. Ask specific questions.

Focus your model around specific questions. Form follows function. How do we fix a specific challenge or a specific problem? Build a model to test your solutions.

Ask what if?

Borrowing from Covey, build your model with the end in mind.

Don’t worry or focus too much on the answer. Focus on the question and the process you use to answer your questions.

Seven. Build a Model for Dr. Doom.

As founders, we are often optimistic. There is no room for dark thoughts or the downside when we need to take a leap of faith. Build you need to shed the dark light on your models.

Ask yourself two questions.

One. What will kill this business? What changes in parameters, choices, values will kill my business?

Two. When I find myself in this position, how do I survive? How do I go back? How do I find the way back home?

The Upside is well and good, but an effective financial model should also consider what will make the business break. And then what can be done by the company in that situation to fix or mitigate it.

Eight. Peer review.

You are not alone. Ask for help. Find the smartest people in the room and ask them for help. Model Feedback from peers and colleagues is a gift. It adds clarity to the thought process and helps in eliminating redundant elements.

Nine. Build in Excel first. Then in the real world. 

Cheaper to build it in Excel. A typical business takes 2 to 8 years to show that you were right and that it worked. Would you rather find out at the end of year eight that you were wrong? Or on day one.

Build it in Excel. Take it for a dry run. See where that takes you. Don’t wait till the end of the line to find out that you were heading towards a dead end. Do it on day one.

Also published here.