Pricing can be somewhat of a dark art when approached from a software background. However, there is some guidance that can be helpful in trying to decide what you should charge for your product. Pricing is been deeply studied in both business and economic. This post cannot replace the advice of experts. In a start-up you really need an idea of where to begin. The ability to hire experts comes after you are already making money. Here are three ideas for establishing price.

  • Pricing by market neighbors
  • Value to the customer
  • Revenue maximization analysis

Pricing by market neighbors

This is the technique you use when selling a used car. You go online and look at the price of other cars ofthe same make and model. A similar technique is used in pricing homes and other real estate. This works best when items for sale can be easily compared.

In the software start-up business your goal is usually to create something new that is unlike what has gone before. This technique can still be helpful. Suppose you are inventing software to prepare taxes (yes I know this has already been done.). One way to price your product is to look at what customers are already paying for tax preparation services. Just matching what has gone before may not be appropriate. One approach for disruptive innovation is to deliver a solution at a price well below market norms and still make a profit.

see: The Innovators Dilemma by Clayton Christensen

Another approach is to address a niche of the market with special needs and then price at a premium for meeting those needs. In any of these cases, looking carefully at what people are currently paying for similar services is an important step in setting a rational price.

Value to the customer

This approach looks at your customers and their business/lifestyle to see what prices make sense. One guideline to value is the wages/salary paid to the people who will use the product. In most businesses personnel costs dominate business expenses. So for example, if using your product will free up half of the time of a person making $25,000/year then your product needs to cost substantially less than $12,500/year. If it does not cost a lot less than the savings produced then there is no advantage in buying your product.

In many cases a software product is intended the enhance the productivity of some set of people. In those cases, the product’s value is related to the value of the person it serves. You can sell a product to assist doctors making $400,000/year for a much higher price than one to help secretaries making $30,000/year. If a product merely assists a person in their job then it is unlikely they will pay more than 5-10% of wages/salary for that assistance. If a product substantially replaces the work of an employee then that replacement of wages is an upper bound on what you might charge.  In the U.S. the Bureau of Labor Statistics can provide information about what people make in various occupations.

see: Bureau of Labor Statistics wage data

Revenue maximization analysis

The basis for this pricing approach is to first collect data on what your customers are willing to pay for your product. There are a variety of ways to go about obtaining this information. You can talk to a lot of customers and simply ask them what they would pay. You can send out surveys to customers asking similar questions. The challenge of both of these techniques is their reliability. When answering a survey or a direct question the customer frequently has not thought deeply about the value of your product. Many times they are responding to you as a person more than to the product’s value. Because they have no “skin in the game” they are less careful about what they say. Survey and question data is still better than just guessing. The customer’s opinion is likely to be more accurate than your guess. If you are getting a lot of answers that are out of line with the previous two pricing techniques then do some more thinking about how you are acquiring information. It is also important to be careful about who you are asking. In many enterprises the person using the product is not the person making the purchasing decision and may not have a realistic view of what a product is worth.

Eric Ries advocates creating a minimal viable product (MVP) and just start offering it for sale. Even if there is no product yet, an offer for sale at a given price that links to a page asking for an email address for future product availability can produce concrete evidence of what people will pay. Several web pages offering the same product at different prices and produce useful information. The value of this kind of information is that you are sampling behavior (an actual desire to buy at a particular price) rather than opinion. This kind of technique can work when there is a very large consumer market. Making 100 people irritated that your product is not yet available has little risk. However, if there are only 1000 potential customers, you cannot afford to irritate very many of them.

see: The Lean Startup by Eric Ries – Chapter 6

All of these techniques should produce data that looks like the following graph. You have various prices and the number of people who indicated that would buy at each price. This data is much more tidy than the data that you will have, but it serves as an example. The analysis tool is just a simple spreadsheet.


At first look  one is inclined to pick a price of $40 because that is where the most customers said that they would buy. Getting the most customers to buy is not actually the goal of our business. Getting the maximum revenue is our goal. For that we need to look a little more carefully at our data. For a software company, the costs per item sold are minimal so maximizing revenue is a good model for setting prices. In other businesses where the costs of items are a factor, this analysis should be modified to maximize profit.

The first step is to change our data from what people said they would pay to the number of people that would buy at a particular price. If someone said they would buy the product for $50 then they would probably still buy the product if priced at $40. Therefore the number of people who would buy at $40 should not only include the people who gave $40 as their answer but also everyone who gave a higher price. This analysis is a little simplistic because some buyers associate price with value. If the price goes too low they start to expect a lack of value and then will not buy. Starting out, we can ignore those effects, particularly if we are choosing prices indicated by our other two approaches. Accumulating all buyers at higher prices produces the following graph.


Now we have an estimate of how many buyers we would have for a series of product prices. This alone is not enough. Picking the largest number of buyers would indicate a price of $10, which will not make us a lot of money. What we want to do is maximize revenue. Computing revenue is simply a matter of multiplying price times number of people who will buy at that price. The following graph shows our revenue at various price points.


In this graph we see that we make more money by charging $30 even though our original data showed that many more people would pay $40. By charging $30 we captured more paying customers and their income was more valuable than the loss of $10 per customer.

A little thought would indicate that the maximum revenue may actually lie on either side of $30. It may be at $27 or at $35. With a little more effort we can use Excel to fit a curve to the revenue data. We then take the first derivative of that curve and set it equal to zero (calculus strikes again). That will give us a more precise estimate of the price with maximum revenue. However, we must be careful to have enough data to justify such precision.

This analysis is easy to do with some customer data and a spreadsheet. I learned this from Nile Hatch. He and his company (Quanitval) have more sophisticated techniques that are more precise but this is a good rational start.