Five Tips for Estimating Requirements

Years ago I worked on a large effort to reengineer a distribution center for a large retailer. We provided an estimate for both the business analysis work and for the entire project, which would involve the organization’s first use of Electronic Data Interchange (EDI), new business processes, many software changes, and the purchase of new barcode scanners. The business analysis effort took far longer than we anticipated, and at the end of it we refined our estimate for the total project. When we reported the new estimate to the president of the company, he literally pounded his fist on the table and asked, “How did we get to this point? Why didn’t we know sooner? You’ve already spent all this time on the project and what do we have to show for it? Nothing!. Absolutely nothing!”

I have always thought of business analysis as the most ambiguous and the most fun of the project phases. However, for many years it was my least favorite phase to estimate. I felt like I was guessing, simply pulling numbers out of the air. As a business analyst I thought it was project management work. No wonder we were so far off.

Estimating the business analysis phase(s) is not easy. It is not hard, but it takes a willingness to think about exactly what work will be produced, and many business analysts do not have the patience.  So for those of you who do not have the “stomach” to spend the required time to estimate business analysis, here are four tips.

1. Break the effort into manageable pieces. We can estimate a whole lot better when our business analysis phase(s) are small. It’s easier to estimate a user story than an epic story, or one specific business process than business process modeling in general.
2. Choose your approach. We’ll estimate differently if we’re using a plan-driven approach (Waterfall) than if we’re estimating in a change-driven (Agile) environment.
3. Use a variety of estimating techniques. On many projects we cannot be precise about our estimates when we’re first asked how long business analysis will take. We usually use analogous estimating, or experience with a previous project. If we have good history, we might be able to use parametric estimates. For example, if we know that it takes four hours to model a business process and we have five processes to model, it will take twenty hours to model business processes.
4. Brainstorm.  Talk with the people who are actually going to do the work. They usually have a more realistic idea of what needs to be done and how long it will take. I also like yellow sticky notes, since they can be easily added, taken away, and moved.
5. Identify all the deliverables/artifacts.  Make sure you know your deliverables before attempting to identify the tasks needed to produce them.  Here are a few examples of deliverables: user stories, agendas and minutes, “as-is” business process model, traceability matrix, to name a few.

Of course the real, real key is having the courage to communicate bad news. Which brings me back to the president pounding his fist. What I should have done was communicate our status regularly, rather than surprising him after months of effort. What a lesson learned!