In this and my next blog post I will be talking about rules of agile planning, based on my experience in agile development and project management. But before I get there, I will need to introduce some fundamental concepts of project management first:
- “Iron Triangle” (a model of the constraints : time, money, scope)
- “Cone of Uncertainty” (accuracy of effort estimation vs. time)
At very begging of a release, it is important to decide which project lever — time, cost, OR functionality—cannot be compromised to launch a successful product. These three levers are also referred to as the “Iron Triangle.” Fixing time, budget, and functionality is not possible; at least one has to act as a release valve. Decision which dimension is less critical is not the choice of a project manager, but failing to ask PO and stakeholders before the start of the project this question is either sign of lack of maturity or courage. As a project manager, you must have both.
Some project management cynics would point out that there is always one dimension more you can “play with” if you get in trouble – and that is quality. Therefore, the best way to evaluate teams and project managers is by checking the quality of the consecutive releases of the same project. (that is also the reason why one-ff CDE projects lack trust in relations – such projects are always full of SLAs and metrics).
Even with a shared product vision in place, the product’s exact functionality is not always well known up front due to:
- Technical challenges
- Customer feedback (which might impact design choices or add extra functionality)
- Sometimes we need to dig into the surface to learn the true nature of things
- Sudden change in “market requirements” – which often means new development in the current release (that largely depends on the release lifecycle)
Three main project management practices differ only in how they see a change of the functionality during the project lifecycle.
- The waterfall model (“mechanical view”, no changes allowed)
- Spiral development (can cope with change, via prototypes and risk analysis)
- Agile development (“designed” to allow changes, higher value on adaptability than on predictability)
In agile, it is recommended to fix the time and “flex” the functionality.
Identifying the launch date is facilitated by the product vision – a roadmap, which is something that PO needs to bring to the table. It sketches the future product and describes its target customers. Vision allows us to determine the window of opportunity, the time frame in which the product must be launched to achieve the desired benefits.
Note that choosing the launch date based on the work in the product backlog is difficult, as it forces the team to freeze requirements at an early stage. If that is done too early, it results in extremely poor estimates. The accuracy of the software estimate depends on the level of refinement of the software’s definition. The more refined the definition over time, the more accurate the estimate – which is a definition of the “Cone of Uncertainty”.
“Walking on water and developing software from a specification are easy if both are frozen.” Edward V Berard
This is exactly where the waterfall guys got this problem wrong, as they have tried to reduce the uncertainty by insisting on detailed requirements too early in process – which resulted in lack of willingness to adopt to later changes, poor estimates and huge waste at the begging of the projects. One more problem which came with waterfall, which I particular dislike was covered in one of my previous posts.
Fixing functionality is a good strategy only when the product lacks some of the key features that competitors already have or if you want to bring something new and really cool on the market.
For instance, the best way to describe two different release planning strategies is to imagine yourself trying to bring a new smartphone on the market:
Either you have to have the phone in the shops before Christmas, as this is the time when kids get new gadgets, or you need to build a new phone with two antennas, as the phone with only one is not good enough.
When it comes to the cost, using stable scrum teams makes determining a budget straightforward—assuming that labor is the decisive cost factor. If you have to scale your project, accurately forecasting the budget is more difficult, particularly for new-product development projects.
If the budget is in danger of getting overrun, the product owner has to make a choice: Either ship with less functionality, or increase the cost by asking more people to join the project—as long as there is enough time for the new project members to increase productivity. But beware of Brooks’ Law: “Adding manpower to a late software project makes it later.”
OK, now that this is covered, let’s get to the real thing, here is my post on agile planning.