Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!samsung!brutus.cs.uiuc.edu!apple!sun-barr!newstop!sun!stpeter!cmcmanis From: cmcmanis@stpeter.Sun.COM (Chuck McManis) Newsgroups: comp.sys.amiga Subject: Re: Re:Byte by Byte / WordPerfect Message-ID: <132421@sun.Eng.Sun.COM> Date: 2 Mar 90 03:03:47 GMT References: <87.25D6A8B2@afitamy.fidonet.org> <27063@cup.portal.com> Sender: news@sun.Eng.Sun.COM Reply-To: cmcmanis@sun.UUCP (Chuck McManis) Distribution: na Organization: Sun Microsystems, Mountain View Lines: 148 In article <27063@cup.portal.com> Classic_-_Concepts@cup.portal.com writes: > This is pure speculation, so easy on the flames, and I'd love to hear >from some of the developers who have stayed away from the Amiga, BUT, are >prices of Amiga software too low to provide incentive? And I'm not talking >about incentives for big profits, I mean just enough to cover development >and after-sale support (which is a big chunk of the cost)? Understanding these components of software development can often mean the difference between successful software companies and unsuccessful ones. Since a lot of software developers try to be their own business managers it might help to cover some of the territory again. When calculating any business venture, one first has to estimate the profit, and thus the return on investment for that venture. This is a fairly simple to comprehend Profit = Gross recipts - Cost of goods sold Return = Profit / COGS Profit is easy, just take what you spent and subtract it from how much you made to see what it is. But what isn't to clear is that the number "Gross Reciepts" is actually Gross Reciepts over Time. So lets say you wrote a program WhizCalc and sold it for $50, and over the course of the year you sold 1000 copies. Then your gross reciepts for one year are $50,000. The cost of goods sold (COGS) is a bit tougher to calculate because it has some fixed costs like the price of the disk, the manual, and the box. Some non recurring costs such as advertising, trade shows, actual development, and sampling. And it has some more nebulous costs such as the rent on your office, support personnel, and telephone bills. Some of these like the rent on your office also include time periods. Others such as tech support have to averaged over the cost of all copies sold. So pulling some numbers out of the air, suppose the fixed costs of the program are $5 (disk, manual, and box), the cost to develop it was $25,000 (about 3 months developer time), and your annual cost on the office is $24,000/year. (For simplicity we will assume you *only* make WhizCalc so all of the cost of the office have to be applied to it's cost, if you made WhizPaint as well you could split the cost of the office between the two programs. Now as a "visionary" you have to guess how many copies of your program will sell and over what time period. (Sometimes this is called the product's "legs") lets say WhizCalc will only sell for 1 year before it is pirated or otherwise replaced. NOTE : Making this guess gets better with experience, there is no "formulae" to calculate it. Now we can calculate its COGS by adding up all of the numbers : For 1 yr, Rent $24,000 Development $25,000 1000 units $5,000 Misc $5,000 ======== Total $59,000 Using our formula for profit we come up with -9000 dollars. Ouch! Our investment return can be calculated as -9000/59000 = -15% per year (since we did everything in years). Clearly, making WhizCalc is a bad idea. You might as well give away $9000 early and save yourself some hassle. (The interesting thing is that naive or uninformed developers will say "I made $50,000 on this program in one year!") Now lets lighten up the picture a bit. Lets say that WhizCalc will sell 1000 units/year for *two* years. Now the *yearly* costs for Development are halved because you average the development cost over the "life" of the product. That makes our numbers : Rent $24,000 Development $12,500 1000 unit $5,000 Misc expense $5,000 ======= Total $46,500 Now your profit is 50,000 - 46,500 or $3,500. Now we're talking, in the black at last right? But wait, what is our return on investment ? (50,000 - 46,500) / 46,500 = .0753 = 7.53% Guess what, money market accounts are paying 8.2% / year. So as a business you have to decide that it just isn't worth it to develop WhizCalc, you would do better to put your money in a money market fund and just sit around for a year. The variables that most affect the monetary return on software are the number of copies sold, and the value received for each copy relative to it's cost. (The price you sold it for.) If you use our example above you can see that if you could double the units sold in a year to 2000 it would raise costs by $5,000 but raise your gross receipts by $50,000. And that would raise your return to 94.17%. At this point it would be silly *not* to write WhizCalc. Relating all of this back to Julie's question, yes there was a question, most forecasters use an empirical rule for calculating the number of units sold, and this is often based on the installed base. So for a computer with a million units installed base, and a rule of 1% of the owners will buy your product. You can use as a starting point that you will sell 10,000 copies. Remember you have to consider things like only 40% of that installed base is in the US, so if you are targeting US distribution then use 1% of 400,000 or only 4,000 units for your base calculation. Then you have to determine the "sale price" of your product. This number is usually much lower than the "list" price because the people who distribute your product and the dealers who stock your product want to make a little money on it in the process. The factors that go into this calculation are "What are the costs of similar products?", "How much does the machine sell for ? (which will give you an idea of the budget of its owners. No one sells a $3,000 CAD package for a $99 C-64, but will for a $20,000 Sun workstation)" and "How much markup will I allow the Distributors and Dealers?" This last one may determine whether or not your product gets any shelf space at the dealer. WhizCalc might have to sell against MiniCalc and if the dealer stands to make $5 selling a copy of MiniCalc and only $2 if he sells WhizCalc you can bet he will push MiniCalc. This same effect affects distributors as well. Presumably, all developers have some idea of their own costs like rent, supplies, and their own salary. For people like Microsoft it's even easier because they have pay records. Costs like advertising and such can be determined by calling up various magazines. Given all of these numbers, one can estimate the return on investment for any given software project. Most companies have a cutoff point where they won't invest if the ROI drops below that point. If the estimate is 94% then clearly you should invest, if it's 10% then you might invest but it can be risky because a schedule slip or unexpected costs and drop you down into the single digit numbers. Startups generally won't do risky stuff on purpose, people like Borland can, but they might have six other projects with less risk available and so they don't even bother. Because a lot of the numbers are interdependent (like units sold ~ advertising costs) the ROI becomes more of a "region" with error bars in both directions. That is when you hear stuff like "The return should be between 15 and 30 percent." The better you forecast, the more accurately you can make these decisions. Part of the problem of the IBM PC is that with an installed base of 20,000,000 a very silly "phonebook" type program that costs $10 and takes a week to develop can have an ROI of 50% and wipe out any chance of that whizzy DTP package for the Amiga which will cost $300 and have an ROI of only 20% of being developed. Whew! --Chuck McManis uucp: {anywhere}!sun!cmcmanis BIX: cmcmanis ARPAnet: cmcmanis@Eng.Sun.COM These opinions are my own and no one elses, but you knew that didn't you. "If it didn't have bones in it, it wouldn't be crunchy now would it?!"