Financial projections of new business ideas are totally made up / not accurate, so being of by a few million here and there would not matter? We can make quick changes in our financials in the presentation slides, and then "forget" about updating our financials spreadsheet with the new information. 

While the absolute numbers of your financial model might be totally pulled out of the hat, it is the thought process of how you got to them that is still valuable for investors. How does your business model work? What would I have to believe in order for your year-5-dream-scenario to come true?

And that model should be consistent across all your documents: presentations, spreadsheets, budgets, everything:

• Discrepancies make you look sloppy (a little preview of things to come when you need to work together with an investor on a Board)
• A consistent model of totally made up numbers makes sure that everything is, well, consistent. If you just slashed sales & marketing cost by 50% but maintain the same amount of sales people, something goes wrong.
• Inconsistencies make it harder to understand your story for an outsider. If sales are $50m on one page and $49m on another people get confused. You established "$50m" as a mental shortcut for year-5-sales-in-the-most-optimistic-scenario, and all of a sudden you create a new shortcut.

So, even if nobody can predict the future accurately. there is still value to create a consistent financial model the same way as you would for a business in which you know every single detail (next year's budget of an established company for example).

What can you do to incorporate the fact that numbers are highly uncertain?

• Round things up to whole numbers (no $49.569m sales in year 5)
• Minimise the number of assumptions you put in the model and make cells that are guesses highly visible (I usually mark them bright yellow).  Rather than "guessing" the number of customers for each year, and the number sales people for each year (10 assumptions over 5 years), you could assume a % growth of customers, and a fixed ratio of sales people to customers (2 assumptions).

Slightly more complex models might actually be simpler to understand.