>comparing yourself to another company’s valuation based on some metric like registered users
I really doubt anyone does valuation like this. What is more interesting is engagement and time using product. # of people is more a 'vanity metric' - it looks nice, but it doesn't mean much.
You should discontinue doubting that. Valuations are done using every available metric (users, revenue, etc) and some metrics that have to be "triangulated".
Valuation experts select the one they think is most relevant or mix several metrics to arrive at a value. Some acquirers view different metrics as important depending on what they need. Some companies need top line growth to keep their multiple so they buy less profitable companies with a better chance to grow revenue for example.
You can argue all day whether you think that is right/wrong, dumb/smart, but that is the way it is currently done and the smartest people in finance constantly work on new ways to value companies but most of those ways involve coming up with new metrics.
Most valuations are done like that. It is called the comparable method.
Essentially all valuations have 3 components:
.
1) Comparable Method
This is the method you are talking about and it is ALWAYS used. In some cases, with companies with revenue and net income analysts will value things like Dividend/Price, or Price/Earnings.
In cases where small changes in the business model can yield huge changes in Earnings for example, analysts may use other metrics to get a more 'accurate' measure of profitability. In cases of internet companies it may be users, or may be 'content share' or some other metric, and they compare it to public companies like google or facebook.
The downside of course is that no one knows for sure what the value of each 'user' is for the new company. But the upside is it gives the analyst another way to think about a tough problem.
2) Discount Method - First the analyst estimates future cash flow over a specified period using assumptions about how the company will perform. He then 'discounts it', divides the future cash flow by a discount rate, and arrives at a value.
The downside of this method is that the assumptions are very hard to come up with.
3) Net asset value method - taking the companies hard assets and valuing them. (I'm not sure if anyone is doing this, but I would guess that the cost of hiring awesome, top rate engineers is probably modeled and incorporated in the a NAV when making talent acquisitions.
Now, with instagram you might be saying, wait, they don't have a metric that gives them a $1B value. That is the entire reason they didn't IPO rather than sell. Because facebook thought that their were synergies in the deal, whether they be future earned cash flows, or a reduction in negative future cash flows.
Zuck may not have done a model, but he very well may have said to himself...
The buyside (stock buyers at mutual funds and hedge funds) all are asking me about competition, and how I'm going to grow this business over the next 3 quarters (what matters), and he says, for 1% of my valuation, I can probably get a 10% higher stock price at exit.
This is also why Yahoo sued Facebook a month ago as well. They figured the lawsuit would cause investors to balk at the IPO which would cause Zuck to settle it quickly for more than Yahoo could get otherwise.
In the long run, valuations are complex, and so is business. Most programmers/hackers etc. look at the Wall Street Journal and think that everyone is crazy and that there is a bubble. There may be a bubble, but everyone isn't crazy.
I really doubt anyone does valuation like this. What is more interesting is engagement and time using product. # of people is more a 'vanity metric' - it looks nice, but it doesn't mean much.