Comment by Silhouette | Hacker News Reader

Comment by Silhouette | Hacker News Reader

These documents all look very smart, but how does anyone have any idea whether the numbers in them are even close to realistic? I can make up some cute hockey stick numbers for any of my businesses, but why should anyone trust them? Particularly for businesses that haven't even launched yet and therefore have literally zero hard data on whether customers will actually pay them money or how much or for how long. How many startups in that position, with no financial data and no evidence of product-market fit, can credibly predict even a year ahead that they won't pivot to some completely different plan, never mind 3 or 5 years out?

When I started my first B2C, a long time ago now, it was actually the bank where we opened our business account who asked these kinds of questions. We sat down, put our best guesses at plausible numbers into a spreadsheet for things like acquisitions and churn, worked out the money that would result.

Barely any of the key assumptions we made were within an order of magnitude of reality, and they were all in the wrong direction. For example, we have far higher churn than any example startup business plan I have ever seen just from card charges that fail with no obvious explanation each month where we don't subsequently recover and continue that subscription. That problem remains one of our biggest pain points to this day, and that effect alone has turned many an otherwise profitable month negative and reduced that business to a fraction of the size it would otherwise have been by now if everything else was held constant. No-one here saw that coming. No example plans or startup guides or financial advisors we consulted even mentioned the possibility, never mind giving any concrete figures for what we might expect.

The major value of doing this modelling is to be able to do sensitivity analysis on the key business factors [0]. You want to know what things really matter and what don’t.

There is also great value to be gained just from building the model as you are forced to estimate the various factors that will drive the business. Whenever I have gone through this exercise I have picked up at least one important assumption that I had missed.

0. https://en.wikipedia.org/wiki/Sensitivity_analysis

I'm one of the founders of Causal (#12 on the list) — this is one of the key things we've tried to address in a couple of ways:

1. Ranges vs point estimates: instead of assumptions being fixed values (e.g. my 'Growth Rate' is 10%), you can define ranges like '5% to 15%', and we run thousands of simulations to figure out the actual range of plausible outcomes. We also make it trivial to set up different discrete scenarios.

2. Interactive dashboards where you can tweak assumptions: if you think an assumption is too ambitious/unrealistic, you can actually just go in and change it yourself to see what happens. These changes don't get saved to the model, and there's no chance of accidentally breaking anything.

Here's an example: https://my.causal.app/models/1269

Sure. I guess my point is that even picking ranges, your error bars are so wide before you have any concrete data on revenues and product-market fit that I don't believe you can draw any intelligent conclusions anyway. Once you get something into market, and you start spending and making money, and you start hearing back from paying customers and running experiments and iterating, then you can get a meaningful sense of how your product or service is going to perform and choose ranges that might not be so wide as to be completely useless.

The exercise of modeling your business helps you become aware of all that you don’t know, which helps you stay humble and curious and focused on answering the right questions.

I can only speak on behalf of my model (#3), but that's meant entirely for companies who want better visibility in their existing operations. Say, they want to figure out how much cash they have in the bank in 6 months, or can they afford to hire 3 more ppl next quarter.

Personally, I have found it hard to work with pre-revenue companies, especially if they come to me with a plan to hit tens of millions in revenue in just a couple of years since launching. Maybe a small percentage of them do, but given how many don't make even a single dollar I've tried to steer clear of pre-revenue startups. Companies with real revenue and growth seem to be a much better fit.

That is the same logic of later stage VCs. Which makes me appreciate those earlier stage investors even more as it is really tough statistically to pick the winners.

I feel like there 2 very different use cases for financial models:

- Early-stage fundraising. The numbers are wrong, everybody knows it, but you have to show that curve going up and right.

- Later-stage (maybe 1-year post-revenue?) when there is some level of robustness behind the numbers, and you do it because it's useful to pilot the company

Agree 100%. Very different needs with the two lots.

Also agree here.

Rational financial planning is, of course, essential for managing a business sensibly.

However, vague intuition + random luck generator + huge uncertainty != a useful financial plan. I feel like a lot of startups could summarise their financial slides with something like "We anticipate an outcome somewhere between failing within three months and becoming the next Facebook, with somewhere between Ramen profitability within six months and a unicorn exit at 8-10 years being most likely."

Early stage is also about ensuring that the levers are understood and that the thinking about the right things to measure and what to action are appropriate in understanding the business. Less about the numbers, more about the variables.. why are you measuring this, why arent you measuring that? What actions will you take to try to get to this number? This would give the investor a better sense of how the founder thinks.

These models are a great example of why it's important to have these retractions to cull out people who are just riding the wave and happen to be in the right place at the right time.

These models all look like toys. Things to play with. They have not grounding in reality, just dreams and ideas. I'm always shocked to see people use models like these to invest in early-stage seed and Series A even though we all know that dialing in the right product and dialing in the right sales/growth model mean way more to a startup at this stage's success than 1.5 vs 2% churn.

Reply on news.ycombinator.com