How many times have you seen an engineering project get delivered behind schedule or with a reduced scope? Probably more often than you'd like.

We can all agree: Estimating is hard. Software development has a lot of known unknowns.

That said, seasoned leaders have found repeatable methods for improving estimations, which in turn help them allocate resources more confidently.

In a recent Faros webinar, Mustafa Furniturewala, SVP of Engineering at Coursera, shared his tips for improving predictability, including how to scope, estimate, and communicate your commitments.

Scoping: Invest a little more upfront

Mustafa is a big fan of the Shape Up methodology from the folks over at Basecamp.

Shaping involves an upfront investment in project clarity and risk management that pays off by making the development process more focused, efficient, and likely to succeed in delivering value.

The process helps identify the core elements that will make up the solution and how they fit together, without going into the minutiae of implementation. “The gist is that you shape the product in the right way to understand what the actual requirements are and to scope them more accurately,” says Mustafa.

By the time a project is shaped and ready for development, it should be de-risked, meaning the major uncertainties have been addressed. This allows teams to work with confidence and reduces the likelihood of major hurdles or project stalls during the development phase.

Estimating: Use the right tool

A common anti-pattern is trying to achieve extremely accurate estimations, down to the number of hours. Given that R&D work always has some unknowns, what is the point of getting to that level of detail?

Instead, Mustafa recommends using higher-level estimation tools like T-shirt sizing or story points, which are closely related.

T-shirt sizing is an estimation technique where projects or tasks are categorized into sizes (XS, S, M, L, XL) to represent the complexity or effort required, rather than assigning specific hours or days. It focuses on the relative size of a project rather than exact durations, so requests can be estimated quickly without getting bogged down in details.

Story Pointing involves assigning a point value to tasks or user stories to indicate their complexity, effort, and risk, using a predefined scale (often Fibonacci-like: 1, 2, 3, 5, 8, 13, etc.). Story pointing enables a more nuanced understanding of effort and complexity for teams that have a good understanding of their velocity. It’s commonly used for sprint planning and backlog prioritization.

Given story points' strong association with time (e.g., how many story points can fit into a two-week sprint?), some leaders are more partial to T-shirt sizing for high-level estimations. “I prefer using T-shirt sizing as much as possible, and then breaking it down into weeks if I need to,” shares Mustafa.

Committing: Set the right expectations

Communication is your friend. When you share an ETA with your stakeholders, it’s wise to set the expectation that this is only an estimation. "Expectation setting with stakeholders is important to make sure they understand that there are some risks here," advises Mustafa. Timelines may change as you begin to figure things out.

"Communicating is also extremely important," he continues. Part of getting your colleagues comfortable with the risk that timelines may shift is the commitment you make to them that you will inform them promptly if an ETA changes.

By showing you understand the importance of timely updates, you will strengthen the trust between engineering and its cross-functional partners and stakeholders.

Faros AI helps engineering leaders stay informed of how key initiatives are coming along and where the bottlenecks are, so they can be more proactive with that communication.

In summary, while there is no silver bullet, if you want more precise estimations, you have to spend the time to spike and design what you want to build in a bit more detail.

And, equally important, you need to improve your ability to identify delays promptly, address them if possible, and communicate new ETAs promptly.