The Big Development Project: How much should it cost?

September 3, 2008 by Peter Stevens

Picture courtesy of irene.@Flickr

Some people think that agile and budgeting are incompatible. The product is ready when the product owner says it is. But before starting a project, most managers want at a least budget. So the product owner puts together his wish list and asks the ScrumMaster what it will cost to build. The answer comes back — usually a long time and whole lot of money! Then the customer turns pale as he tries to decide what it will really cost, whether he can afford it and whether it's worth it.

But there is a better way: the product owner can perform a double worst case analysis. This quick and easy tool uses the project's business value to determine a reasonable price for the software investment.

Scenario: You are the product owner. Your company, Congo Lomerate Corporation, plans to automate its internal operation by developing a tool for its staff, "TurboCongo." Over the next 5 years, TurboCongo should produce cost savings and additional revenue of €10 Million. You have developed a list of User Stories and given it to your team. Their response: "Yes, we can do it! Including support it will cost €2.6 Million." Is that a good investment?

The fundamental financial project risks are 1) the benefit will be less than anticipated or 2) the costs will be higher than anticipated. A significant delay in the project increases the cost and probably reduces the benefit. Even if both occur, the investment in the project should still be more attractive than putting the money in a savings account.

Determine Business Benefits

A product has features for the users, but it should accomplish business goals for the company, for example: increase productivity by X%, increase market share to Y%, or create a community with 100'000 active users, etc. Each of these factors can be converted into a monetary equivalent. For instance, increasing productivity implies doing more business or saving on labor costs. Even members of a community site with no revenue have a value.

Sum the value of all benefits over your planning horizon, typically 3 to 5 years. The examples use five years, but you can download the spreadsheet and experiment with other values.

Double Worst Case Analysis

Combining the two possible worst cases gives the double worst case. Less benefit, higher costs. Worst case number one: the project is more expensive than planned. Let's assume twice as expensive. If you're using Scrum or Agile, this is may seem very conservative but is actually quite common (I've had to clean up after worse!). Worst case number two: the benefit is less than planned. Let's assume half the hoped for benefit. Your management may or may not consider this conservative. In any case, you can adjust these factors as appropriate.

ROI (return on investment) is simply the profitability of an investment:

Profit = Total Benefits - Total Costs
ROI = Profit / Total Costs / Time Horizon

From our case above (figures in thousands):

• Total Benefit = 2,000 per Year for 5 years = 10,000
• Total Cost = 2,600
• Profit = 10,000 - 2,600 = 7,400
• ROI = 7,400 / 2,600/ 5 = 57% / year

This looks good, but what happens if the project goes badly? Assume half the benefit and twice the costs. Now the project is losing money. Even if the result were slightly positive, there are much safer investments which produce the same return. So a budget of €2.6 Million is surely too high, unless you can lower the risks of the project. (Hint: use Scrum and release early to start achieving benefits as soon as possible).

Development Budget

The target cost will depend on the time horizon and the minimum acceptable ROI. I would suggest 15%, but pick a number you are comfortable with.

If the expected average yearly benefit is 100, then a target cost of 72 will mean the project has an ROI of 15%, even if cost and benefit calculations are both wrong by a factor of two.

The target cost will need to be apportioned to development and operational costs. Using the rule of thumb that yearly operational costs (including support) are 1/3 of the initial development cost, the total cost over 5 years equals (1 + 4/3) times development costs. The target development costs are estimated to be 3/7ths or 43% of the total costs.

How much may it cost?

What about Congo Lomerate? What is a reasonable budget for the initial cost of developing TurboCongo?

For the budget planning, total costs should not exceed 72% of €2 Million, or 1.44, Million, of which 620'000 are development costs and 824'000 are support costs spread over the following four years. Even if the total benefit is half that and the cost double, the project will still be profitable for the company.

When to Use Double Worst Case Analysis

Double Worst Case Analysis identifies a reasonable cost for a software development project, based on the expected benefits and reasonable risks. As Product Owner, you have a clear of idea of what your project is worth and have a basis to negotiate prices and scope with you suppliers. You have a tool to assure not just the functional success of the project, but a financial success for your company as well.

Further Reading/Resources

Download the spreadsheet
Product Owner Responsibilities
Who Manages Project Risks in Scrum?

[ This article is the first in a series on Agile Project Planning ]

Permalink