As a former equity research analyst I built financial models on numerous companies in several industries. I wrote on telecommunications companies (RIMM) I wrote on Software Security Companies (SYMNC) but mostly, I wrote on Internet companies.

The key to writing on companies is to understand the financials and what drives them. That is best achieved by building a model of the company and focusing most on the income statement. The income statement, after all, is the source of the revenue and EPS estimate, the things my clients cared the most about.

To get a better understanding of Facebook, I thought I would walk you, the reader of Zacks.com through how I would do this if I were a research analyst. This is a top-level discussion and all data is from a recent S-1 filing with the SEC.

The Caveat

When I did this professionally, I had a 50 call rule. I had to make at least 50 phone calls to customers of the target company. I had to make at least another 50 phone calls to the target company to talk to their sales department, the service department or anyone else at the firm to give me insight that I would not get from SEC filings or earnings calls.

I didn’t do this for Facebook as I am no longer a research analyst, but I didn’t do it for Google (GOOG)either and I was ranked #1 for Earnings Estimate Accuracy by Starmine… so take that with a grain of salt. For a small company, the calls are invaluable but for a huge company, they are an infinitely small sample from which I would base my assumptions. Google, like Facebook, has hundreds of thousands of advertisers, so my 50 calls would be meaningless.

The metrics

In order to estimate the revenue in a credible fashion, you have to rely on the metrics. At this point, the only metrics we have are ones we have from the S-1. Use the most recent S-1 when looking at a new company, or rely on 10K’s for modeling public companies that have been around for more than a few years.

The metrics will help you connect the dots on how to estimate the total revenue. For Facebook, the company published MAU’s (monthly average users), DAU’s (daily average users), Advertising Revenue, Payment Revenue and ARPU (average revenue per user). We find these metrics on page 48 of the S-1, they are embedded in the graphs so get them into a useable format in a spreadsheet. I did it in excel, but Google docs work just as well.

Facebook took the unprecedented step to break out the users and revenue by geography, giving us a clear idea how this worldwide social networking phenomena is not just a big deal in Palo Alto. It’s a big deal worldwide. I started making my model by focusing on revenue. So I listed all the metrics and then added in a few lines beneath each item to insert an equation for annual and quarterly growth. If you do this, part of your spreadsheet would look like this:

For purposes of clarity, I color code things so I know without checking what type of metric it is. Brown is MAU, Blue is DAU and Green is Revenue (for the color of money). I do not color code ARPU because it is a calculation more than a given data point (even though it is a supplied data point). I also added a % of total revenue for the revenue lines, this might help me down the road when I model out Zynga (ZNGA).

The idea is to work into a “proven” system that accurately shows how the data points are manipulated into revenue. That is done by dividing historical revenue by historical MAU’s giving us a proven ARPU. The ARPU that the company supplied us with for each geography doesn’t match up perfectly as there are times when the ARPU rate is higher or lower than our “proven rate” due to the difficulty Facebook has with coming to actual numbers of estimated users (they guess based on IP address) and assuming a 5-6% duplication of extra accounts.

I believe it’s best to do this one region at a time as opposed to doing it over the blended average, and yes that is basically 4 times the work, but over time it will pay off. Down the road after four more conference calls, we will pick up on trends by geography and our estimates should get better. The more details, the more data points the better.

Next I want to go ahead and work out what the proven ARPU rate is and break that out in terms of both advertising and payments. Basically I just divide the historical MAU’s by the desired Revenue line to get the ARPU I want. Also include the two lines to show percentage growth for both annually and sequentially. I also want to put in a graph right now that measures the different ARPU’s and their respective annual growth rates to get a good look how things trended. It should look like this:

The more astute readers will notice that I have what appears to be an estimate. I have made my estimates for 2Q12 ARPU for the Rest of the World segment and put them in blue text. This helps me understand what is an equation and what is an estimate. Why 3% for both… well I am just showing how to do it, I am not really giving my estimate.

The next step would be to do the same for MAU’s and DAU’s. Again I am using a static 5% for both, but this time I am looking at sequential growth instead of annual. I have found it best to use the time frame that fits best with your knowledge of the situation and the one that has the most apparent trend. A 5% sequential growth rate gives MAU's a reasonable annual growth rate that shows a deceleration, but still growing. Your model should now look like this:

Again, the astute reader will see that I have already stuck in a revenue estimate for both advertising and payments. Others will notice that the stated ARPU is here, as are a few extra lines that help me adjust the stated ARPU with my proven one. At this point we are just about done, if you can believe it. Just wash rinse and repeat for North America, Europe, and Asia. Once you have completed all four segments it is just a function of adding up the sums for each to come to a global estimate.

Those of you who are already good with excel know that pasting and dragging will save you a lot of time, and if you put that 3% and 5% in for all quarters for all geographies, you will end up with the wrong answer, but you will have modeled out Facebook revenue for 2012 as $4.9B or an increase of 32%. Being aware of seasonality and looking at the trends in each geography for each revenue line will lead you to creating the best estimate you can.

When you have your estimates completed, let us know in the comments below!

Brian Bolan is the Aggressive Growth Stock Strategist for Zacks.com. He is also the Editor in charge of the Zacks Home Run Investor service

Follow Brian Bolan on twitter at @BBolan1

Like Brian Bolan on Facebook