What with the hoopla around the possible JBoss (and other) acqusitions in the near term, it set me to thinking about how open source companies should be valued. Once they "grow up," of course, like Red Hat, they're valued in a somewhat rational (P/E) fashion.
But how does one value the frontier of open source business models? How does one value downloads? (I remember Kevin Harvey of Benchmark and David Skok at Matrix both telling me that part of their investment calculus in open source companies/projects hinges on downloads, but how do these factor into an acquisition?) Do downloads equal "community" and, if so, is "community" something that can be valued? (In my conversations with IBM, they certainly seemed to think so, though they didn't share with me how they'd do so.)
And what about downloads or users (as in the case of a XenSource (Xen), Zend (PHP), or GroundWork (i.e., Nagios)) that don't directly relate to sales of a company's products? Can one have a highly successful project without having a highly successful business? Of course one can. These downloads/users are leading indicators, perhaps, but not always revenue-determinative. Relatedly, should MySQL, with a reported 70+ million downloads, be valued higher than a PostgreSQL-based or Berkeley DB-based company that has fewer downloads, but a higher conversion rate into paying customers?
At some point, of course, open source companies will likely get to serious revenues sooner than they have in the past. The trailblazers (JBoss, MySQL, Red Hat, etc.) took many years to get to profitable, meaningful revenues. New companies (including SugarCRM and my own Alfresco) seem to be on revenue trajectories that will get them there faster, largely because of the paths these trailblazers established.
In the meantime, we need to figure out rational valuation models, and be realistic as sellers. This point came home to me last night when I bumped into a good friend and former controller of Lineo, my first open source company. We laughed about how we had scorned $225+ million quasi-offers, and remained scornful up until the day the company was sold off as scrap metal once the promise behind died. We were foolish and greedy. I suspect we're in the midst of another bubble time for open source companies and would do well to be prudent in our expectations. Downloads do not a profitable, revenue-rich company make.
In thinking this through, I stumbled across this excellent report [PDF download] that identifies common valuation methodologies for software companies. I've included a fair amount of relative detail here - try to figure out which you would use to value your favorite open source company.
They are:
Earnings Focused Value MethodsSo, what do you think? I imagine open source M&A will use composites of the above-mentioned models. Some open source companies have real revenues and so will likely be valued against those revenues, primarily, while other companies with strong communities but tepid sales will hope for (and likely get, while the open source euphoria lasts) more of a Replacement Model valuation, whereby the cost of building a community will balance out the lack of revenues.Asset Focused Value Methods
- DCF - Discounted Cash Flows: This is the most common method used in mergers and acquisitions and is being used more often for other valuation purposes, especially for larger companies. The authors of the report feel this is the best method to use.
In this method:
- The value of the company is based on the free cash flow to investors, expected to be generated in
the future.
- The value is the sum of the net present values projected for future years (normally 3 to 5 years) plus the value of "continuing operations" after the projected period.
This method is frequently the only method applicable to companies in the start-up stage and to companies expected to have extremely high growth rates driving even larger growth in their profit margins. The difficulty in using this method primarily relates to determining the risk that the company will not achieve its projections. The old adage that the higher the risk, the higher the return, really applies in the DCF model. Thus, the greater the risk of not achieving the projections the higher the required return required (discount rate).Discount rates applicable to DCF models for most software companies will vary from 20% to 70% depending on the risk to be assumed by the investors. The better management can support the projection'?s assumptions, the lower the discount rate required and the higher the value indication for the company.
[Asay note: This might be the authors' preferred way, but it bears no semblance to "reality" as it exists in today's Web 2.0 world. I suspect that intangibles would drive this rational valuation to huge heights, well beyond revenues.]
- Free Cash Flow - LBO model (i.e., Investor assumes that the company's worth is equal to the down payment plus the debt that the free cash flow will support after reserving some of the cash flow (typically 5% - 15%) as a margin
for error). or model whereby a company is worth three to eight times its cash flow represented by EBITDA (earnings before interest, taxes and depreciation and amortization). The valuation process starts with a multiple of approximately four to six times EBITDA and is adjusted upward or downward based on the qualitative aspects of the company. In this variation, it is assumed the investor will be happy receiving his/her capital back in three to eight years depending on the risk assumed;
- Replacement Value: The replacement cost method is based on the assumption that the company is worth what it would cost to replace all of the company's identifiable assets [Perhaps this is a good place to measure downloads and/or community?]. Replacement cost values tend to be most useful in two situations:
- First, for young software companies, with few, if any, sales, user base or dealer base. This may be the highest value indication possible when the primary investment has been in the technology or product.
- Second, in a company with an operating history which does not reflect its investment (time and money) in it product. The company maybe making a major business model change in its platform, versions or distribution channel.
[As the authors note, one problem with this model is that a great deal of damage can happen between the cup and the lip. Take, for example, the execution risk Oracle faces if it were to acquire JBoss, or...well, I'd better not go there. :-) The point is, much "replacement value" can be lost in the very process of trying to make an acquisition that looks good on paper, also look good in practice. Particularly to a skeptical open source community.]
- Liquidation Value: Liquidation value is the value of the individual assets if the company were to be liquidated today. [Not likely to be used in many successful open source companies.
Market Focused Value Methods
- Internal Transaction Price: This method assumes the stock's current
price should at least be equal to the last price paid for the company's stock or the last transaction price. [I can't see this being used very often, however much a buyer might prefer it. There's little incentive for either the VCs or the companies to sell out at parity (for the VCs).]- Public Company Revenue Multiple: This method looks at analogous public software companies and calculates a revenue multiple by dividing the public company'?s market capitalization by its revenue. [Note, however, that private companies tend to have lower revenue multiples than public ones.]
- Private Similar Company Earnings Multiple: Similar to the Public Company Revenue Multiple, except it uses the revenue multiples calculated from private M&A deals.
- Public Company Earnings Multiple: The public company earnings multiple is very similar to the revenue multiple, except it compares market capitalization to some level of company earnings. The concept behind using a multiple of earnings is that earnings more effectively reflects the difference in the return to the investor between companies than the
revenue multiple.- Private Similar Company Revenue Multiple: Like the private company revenue multiple, the earnings multiple is derived from transactions that took place in the mergers and acquisition world. The earnings multiple is computed by dividing the transaction price by the appropriate earnings of the company.
Posted by Matt Asay on February 8, 2006 08:09 PM
