The short answer is with an estimate! Early decisions are a big deal when it comes to software development and delivery. Whether its agile or waterfall, we need to figure out what the work is going to cost and how long it’s going to take, oftentimes without detailed requirements confirmed. Estimates give managers a good way to start the conversation with internal stakeholders and with clients. Should we take this project on? Is this going to cost 5 million dollars or 10? Do we have the resource capacity to fulfill the demand? Should this take 6 months or 12? Management needs to know the answers, ideally before spending major resources and before detailed planning takes place.
By looking at thousands of completed projects, QSM has found that big money can be saved by taking a quantitative approach to finding those answers. Early data-driven estimates give us the ability to set realistic targets and manage the uncertainty that goes along with early decision making. I am referring to “Big Picture” data-driven estimates, before sprint level planning takes place.
With the SLIM-Collaborate analysis below, we can see a staffing profile that shows a gold estimate along with a more conservative green one; a two column chart showing a comparison summary; a scatterplot showing a risky effort target compared to a more reliable alternative and an industry trendline; and we see a risky gold cost estimate compared to a green high assurance one. The data shown here is saving this company from making a bad decision, a decision that could cost them a lot of money, time, and quality.
To arrive at these numbers, we were able to take a small amount of information and leverage empirically-based models to generate the estimates. You can see in the dialogue box below, this estimate asked for the scope that was required and the duration and staffing targets. This is just one example. If we didn’t know the scope or one of the other parameters, we could solve the problem from a different angle. We can work with the information we have available, then leverage the historical data and the estimation tool to bring everyone together on the same page.
If we have time to run estimates later in the decision-making process, when more information becomes available, we can certainly do that. But in this example, we didn’t have that luxury, so we leveraged the risk management capabilities and the empirically-based models to generate a more conservative estimate with buffer built in to show the 80% assurance level. This was with no additional project information provided.
Early cost, scope, and schedule decisions on software development can be tough to make, since less information is usually available at the beginning of the planning process. Leveraging reliable estimation methods with historical data can go a long way to making better decisions and getting the conversation started.