Pricing software is tricky because it's difficult to establish the cost per unit, and there are many external factors that can increase or decrease the effectiveness of whatever price point you choose.

When you're selling a physical good, there's a known cost of raw materials, production, and distribution, so you can just pad the cost with some profit and round up to the nearest 9 or 5, avoiding "bad" numbers like 13 and 666.

But when you're selling software, nobody knows the true cost per unit. The largest cost behind software, by far, is its development. But development is a static cost - it will cost you the same whether you sell 20 or 2,000,000 copies. How much money did Microsoft pay designers, programmers, managers, and testers to make your copy of Word? Who knows? What if you consider all of the previous development that led up to the current version? What about future updates and patches? What if this version generates 30% more support calls than expected? What if it sells 30% fewer copies than the previous version?

There's a very good reason why so many people in software are insanely rich, and so many more have gone out of business: most software pricing is completely arbitrary.

I registered an excellent shareware OS X RSS reader, NewsFire, for about $20 (€16.99). I thought, "I guess this program is worth 20 bucks to me." But why did the developer choose to charge $20? Would I have purchased it at $25? If the developer released a premium edition and charged $50, how likely am I to decide to spend the extra money for the premium edition?

It's time for some economic theory!

That's probably not the most exciting phrase you'll hear today. Most people have sat through an economics class and either thought "This makes no sense" or "This is all just common sense." One of the most useful lessons I figured out while daydreaming in a college economics course is that much of it is completely wrong, because it's fundamentally based on the incorrect assumption that people will behave rationally.

How is anyone supposed to make sense of all of this?

They're not. But I can at least offer my own baseless theory to explain everything.

Sandwich economics

The perception of your product's price among consumers is important. You might be able to sell 100 units at $10, but 800 units at $5. Why isn't it linear?

People have different perceptions of different price levels. That's why financing is such an incredible money-maker: a person might happily pay $19 per month for 24 months without thinking about it, but would carefully debate paying $299 immediately for the same product.

There are five categories of purchasing decisions:

1. Throwaway: People are willing to drop this amount of money into a tip jar, leave it behind in a cashier's penny tray, or give it to a bum. It's guaranteed to be a complete loss.
2. High Risk: People are willing to spend small amounts of money without any guarantee that it will be "worth it", such as tickets to an unknown movie or food at a new restaurant.
3. Moderate Risk: People might be willing to spend more, but would feel more comfortable if they're given some assurance that their purchase will be worthwhile. If a video game costs $50, many people would want to play it first or read reviews before buying it.
4. Low Risk: Very few people will want to take chances on products in this price range, such as cars.
5. No Chance: People absolutely won't buy your product in this price range because they simply can't.

The Throwaway category isn't very important here because it would be nearly impossible to make a decent living selling software for pennies. If you can accomplish that, you clearly don't need to be reading this.

Now, the part about sandwiches.

The Sandwich Boundary is the approximate price of a great sandwich for your potential customers. Depending on who you target, this is probably between $5 and $8. It defines the High Risk category, from which all other categories are defined.

This is very important.

People don't think twice about spending money for a great sandwich, even if they've never tried it before. At that price level, people are generally willing to buy a product without trying it first or considering the financial consequences. It's a frequent purchase. How often do you buy something for $5 to $8? If you saw an item in a store that looked interesting for $5.99, would it take much convincing for you to buy it? Would you hesitate to sign up for a useful service that cost less than one day's lunch each month?

Below the Sandwich Boundary, it's difficult to make enough money. But for each order of magnitude above the Sandwich Boundary, the number of paying customers will drop significantly.

The longer your customers need to question their purchase, the longer they'll have to convince themselves it's not worthwhile and walk away. And the more questions they need to ask themselves (or their loved ones or business partners need to ask them), the longer it will take. This is why salespeople are always nodding "yes" while speaking and trying to get the sale completed quickly, before you have a chance to think about it.

Each price class adds a thick level of customer questioning. At $20, NewsFire is near the top of my High Risk class. I tried its demo for a month, and when the time came to purchase it or stop using it, I knew that I'd get a decent value out of it so I purchased it without much debating. If it were $100, in my Moderate Risk class, I'd have to convince myself that it's actually worth that price, and I wouldn't be able to.

Here's an interesting effect of sandwich economics: While the price class will greatly influence the number of customers, the actual price you charge within that class is almost completely irrelevant.

If a useful little program is $4, it probably won't sell many more copies than the same program at $9, even though it's less than half the price. They're both in the same perceived price class to people, so if someone won't buy it at $9, he probably won't buy it at $4 either. Similarly, NewsFire's sales probably wouldn't be very different if it were $15 or $30 instead of $20.

The tricky part is figuring out where your potential customers define their price classes. Most software shouldn't exceed Moderate Risk pricing. But what's Moderate Risk for your market?

Photoshop, $649, is one of the most pirated programs. Most Photoshop pirates are young amateurs who use it more for fun and tinkering than professional image editing work. For high school and college students, $649 for nonessential software is solidly in the No Chance class.

Adobe realized that they were missing out on a lot of potential sales, so they released a stripped-down version called Photoshop Elements for $99 with some "home user" features and without most of the professional-level capabilities. While $649 is No Chance for many Photoshop pirates, $99 is Low or Moderate Risk. Ignoring the portion who are unlikely to pay for any software, Adobe probably converted many Photoshop pirates to Photoshop Elements customers.

Within Photoshop's target market of photographers and designers, however, $649 for an excellent, indispensable tool is a Low Risk purchase. These customers are willing to spend three times that amount on a single camera, and they're likely to need additional cameras before they need additional Photoshop licenses.

Given Photoshop's market, Adobe could probably charge $899 and remain in the same price class, retaining most sales.

Who knew sandwiches could tell us so much?