In a recent article by Debbie Madden, CEO and co-founder of Stride, she argues that a budget is often better than an estimate.
Her argument is that getting a good estimate of a project often takes a lot of time and cost. And in the end, a lot of software projects run “X”% over the estimate. So she advocates for a budget.
Because software is always a creative process, her “budget” is not what we normally think of as a budget. She is actually advocating for a form of Bayesian Analysis. This type of analysis is where a statistic is assigned to outcomes. It works like this.
Let’s assume we’re building a book selling site. And we need a shopping cart. We ask experts if the cart can be build within our £200,000 total budget for the project. They may give us these estimates of the shopping cart as percentages like this:
- there is a 100% chance we can deliver the cart for £200,000 or less.
- there is a 90% chance we can deliver the cart for £20,000 or less.
- there is a 50% chance we can deliver the cart for £2,000 or less.
In other words, we get the technical team to give us estimates based on our budget. If they say we need the total budget to make the shopping cart, we kill the project because we know we need more software parts to complete the whole project.
In this case, the technical team feels there is a 50/50 chance to complete the cart for under £2,000. By just asking the developers to give a percentage for the major parts of the project, we can see if the whole thing can be built without the need for a formal estimate – saving the time and expense of running the numbers.
By breaking the project into major parts, and getting the technical developers to assign a probability to a budget number, the project can be bracketed. It will fall into the “can’t be done for this budget,” or “Possible” bucket quickly.
And the percentages can be discrete. The teams can give percentages like this:
- 50% say this phase can be completed for £100,000
- 50% say this phase can be completed for £10,000
To see what the experts are really saying, we do the math. We total the two, and multiply by the percentages. That’s £110,000 x 50% = £55,000 given these two estimates. If that phase, and all the other phases, fall within the total budget for the project, then management has confidence that the project can be completed within the total budget.
Debbie Madden’ point is that a formal, estimate is not really needed in most cases. What is needed is the budget and a confidence level from the technical team. The latter is the key.
The only flaw in this is when a good analog or similar project cannot be found to base the budget on – whether the outcome is based on value or technical success is another question. Otherwise, this is worth a look and could save the time, and the expense of a formal estimate.