Insights on Estimating Software

If someone tells you with absolute certainty what a software project will cost upfront, they’re not telling you the whole truth. Software isn’t built like a bridge, where the properties of steel and concrete are well understood. It’s built with ever-changing components (APIs, open-source libraries, cloud services) that can shift overnight. What worked last year might be obsolete today. That’s why early-stage software estimates can be off by as much as 400%, not because of bad planning but because of unavoidable uncertainty. Steve McConnell studied over 10,000 software projects and introduced The Cone of Uncertainty to illustrate this challenge. The key insight? Uncertainty shrinks over time, but only if you take the right steps: 1️⃣ Product Strategy: Defines business goals and system requirements, cutting cost uncertainty in half before design even starts. 2️⃣ User Experience and User Interface Design Translates functionality into screen designs, further refining scope and behavior. 3️⃣ Technical Design and Software Architecture: Brings cost uncertainty down to 25%, allowing teams to set realistic budgets and timelines before development begins. Every step removes another layer of uncertainty. The best teams don’t gamble on estimates; they manage risk, adapt to new information, and make informed decisions at every phase.

Before starting any new initiative, everyone will ask "how long will it take" and/or "how much will it cost us?" When I was a developer, I used to get frustrated every time I was asked this question. And even more so when I was asked for an "accurate estimate" (wait... WHAT?!) 😶 But as I evolved into leadership roles, I understood why this data was important:    - It offered clarity on the size and complexity of the work. - It informed resource plans to complete the work or impacts to other projects. So, if you're ever asked the same question... Here are 5 techniques I've learned to estimate IT projects: 1. Bottom-up Estimation: Bottom-up estimating uses a work breakdown structure (WBS) which you can take from your project plan and break down a project into its individual tasks, which are estimated separately and then added together to calculate the total project cost. 2. Top-down Estimation: This is a method of evaluating a project or budget as a whole and then separating it into smaller components. It involves creating an overall plan or budget without defining the particulars, and relying on experience and past data to produce a ballpark figure for the total cost. 3. Three Point Estimation: Three-point estimation is a way to calculate a realistic cost estimation using the average of three data points: Best-Case, Worst-Case, and Most-Likely scenario. 4. Critical Path Method: The critical path is the shortest duration between the project’s start and end. This takes into account what tasks can be done in parallel, or which ones have dependencies on work to be completed first. 5. Analogous Estimation: This estimation process uses data from similar projects to determine the overall project cost. This works if you got relevant historical data. Of course, before you do any of the above, make sure you have a clear understanding of what the problem is and what a successful outcome should look like. If this isn't available, be sure to document all your assumptions! If you want to learn more project management fundamentals such as budgeting, planning, and risk management, sign up for The Digital Butterfly membership waitlist today! 😎

Had a great chat with some fellow dev veterans about the basics of production—stuff that’s often missing on projects but makes a massive difference when done right. One big gap we kept coming back to: validating estimates. Here’s the thing: individual estimates might seem fine, but over time and across teams, they’re consistently off. That’s why it’s so important to measure and track the historical patterns of estimates vs. actual outcomes. When you do that, you start to see a truth—like a team that commits to X workload but only delivers 70% of it on average. With that insight, you can adjust future plans to set realistic expectations. This approach helps nail down milestones based on reality, whether that means prioritizing scope or sticking to a fixed deadline. If scope is king, you give the date. If the date is immovable, you define what can actually get delivered. When I interview production or program management candidates, at some point I always ask how they came up with their dates. Was it handed to them, or did they calculate it? Then I follow up with how they validated those dates were realistic? This is where most candidates stumble. Many just say, "The team told me what they could do by the date," and leave it at that. The problem is, people are notoriously bad at estimating, especially for long-term tasks and honestly, if you’re just taking estimates from the devs and passing them up the chain, I find myself wondering what value you're adding to this process in the first place. Why not skip you and go straight to the devs? In this process, a large value of the producer is in the validation step—digging into whether those dates hold up under scrutiny. After all, a big part of a producer's job overall is protecting the team, even from themselves. That means taking their estimates, validating what’s possible, and being the one to push back on both the team and communicate this out to the stakeholders. Without all that, you’re headed for crunch, every time. Learning to validate dates is one of the most important level-ups for associate and mid-level producers. Having this skill and being able to articulate how you do this can be the difference between landing a job or not. Then, once you’re on the team, it’s what sets apart producers who deliver from those who don’t. Hope this helps!

7 lessons I learned after estimating 100 projects. 1. There is always a trade-off between Time, People, and Features. Whether you are Agile, Waterfall, or “AgileFall,” these 3 are in constant tension. 2. Always calculate for the Best, Most Likely, and Worst Case estimate. (Three-point estimation) 3. Define the meaning of DONE. A zip file with the code is way different from a solution running in a Kubernetes cluster. A basic deliverable or a fully deployed solution? 4. Make it small; smaller tasks make estimation easier and more accurate. Think about smaller pieces rather than a large, monolithic task. 5. Write assumptions for obscured features; it will help you later to negotiate if something comes up. 6. You need to know your team's capacity and skill set. Each member's skill level and availability can impact the project's timeline and quality. 7. Never use an estimate as a deadline. Estimates are best-guess approximations, not fixed deadlines. What lessons have you learned?