Guessing is not a strategy: How to build decision velocity with AI and real-time data
June 10, 2026 | Live Webinar
OK, the title of this blog post is a bit misleading. I don’t think I could count how many different ways there are to do estimating, I’m sure there are plenty of tools and techniques I’ve not used in my work as a project manager.
But there are some high level options for thinking about, presenting and sharing estimates with key stakeholders, and that’s what I wanted to dive into today.
I’m drawing on what PMBOK 7 says about different types of estimating, so you can refer back to that for more information. The guide talks about 5 different approaches to estimating, without mandating that any particular approach is used at any particular time – that’s up to you as the project manager to assess and choose the right estimating technique.
Deterministic/Probabilistic
Deterministic estimates are those that represent a single point. For example, 16 months, or $5,250. These are the easiest to understand but (I think) the hardest to come up with unless you have a lot of past project data and simple tasks that repeat.
Probabilistic estimates – why is that so hard to type?? – a those that represent a range. This is what I use most often in the thinking, even though when putting plans together we rarely represent the range in the schedule visually. This type of estimate often includes three factors:
The estimate itself with a range e.g. $5,250 +/- 10% or 16 months +1 month/-1.5 months.
This represents a weighted average based on several possible outcomes (like when you work out a three-point estimate with the most likely, optimistic and pessimistic outcomes), alongside what the range will be for the work. The result is often calculated by scheduling algorithms, but you can work it out manually, or even use professional judgement to make your best guess, as long as people know that’s all it is.
That brings us to the second factor: confidence.
A probabilistic estimate includes a confidence factor that represents how confident you are in the numbers. A 99.9% confidence factor would mean you’re pretty sure this is exactly how it is going to turn out. A 60% confidence factor – not so much.
Thirdly, if you are using computer programs to simulate scenarios and calculate estimates, you will also have a probability distribution, that shows the data and the range.
Absolute/Relative
Absolute estimates give you a specific number that stands alone. A deterministic estimate can be absolute. An example would be that to buy one machine costs £12,000.
Relative estimates are figures that only make sense in the context of other data. For example, buying more than one machine would be cheaper, but we don’t know how much by as the procurement team hasn’t finished the negotiations yet.
A better example, and the one in the PMBOK® Guide, is planning poker for agile estimating. Story points and T-shirt sizing help teams size the work in relation to other pieces of work but alone, “XL” doesn’t mean much to anyone else.
Flow-based
The final flavour of estimating talked about in PMBOK 7 is flow-based estimating. This type of estimating is useful where you know the flow of the work: the throughput and cycle time. If you know how many items of work can be completed in a particular time period (throughput) and how long it takes one item to go through the process (cycle time) then you can estimate how long it will take for a number of items to be completed, assuming other factors like sizing are taking into account in the calculations.
Regardless of how you do estimating, the data is only useful if people are bought into how it has been calculated and what the effort, duration or cost represents for them. Get the team involved in working out estimates and providing the assumptions that they are based on. Remember to adjust for risk, build in contingency where it makes sense to do so, and to keep estimates under review until such time as your confidence levels are sky high… which is usually once the work has been delivered and the task can be marked as complete!
Posted on: April 11, 2023 08:00 AM |
Permalink