Task Estimate Confidence
Steven Souther
I train & consult with PMOs to remove chaos, reduce costs, and improve services ROI. Contact me for free consultation & assessment.
What level of confidence do you have in your Task Estimates? Who estimates the tasks and what methodology do they use?
I have found that despite all efforts to construct a schedule that delivers the project on time, if the task estimates are wrong — the schedule will fail.
I have built many schedules that required re-baselining. This caused me to study why these initial schedules were so poor. What I found primarily was that the task definitions were poor, and secondarily that the task estimations were poor. As a result of these findings, I added 2 techniques to my methods:
1. The task must have a proper definition — confirmed by the resource. You can read about these methods here:
- Task Clarity (https://lnkd.in/fYg2pRJ)
- Releasing Tasks (https://lnkd.in/fpeXyrw)
2. The Resource who works the task is the one who estimates the task using a pre-defined method for estimating.
__________________________________________________________________________
It is very important to coach and build proper rapport with your resource. They need to know that you care about their success. You can do this by asking a few simple questions and listening carefully to their response:
- Have you reviewed the task definition, and do you fully understand it?
- Have you worked this task before?
- What confidence do you have in your estimate of the time required to complete the task?
So, let’s take a look at how resource confidence and the task estimates are related. Here is a typical distribution of the time needed to complete a task:
The distribution plots how frequently a specific time to completion occurs. The area under the curve is the total probability of task completion.
Reading the distribution from left to right notice that:
- No tasks are completed in zero time
- As the time to complete the increases, the number of occurrences increases up to a point of maximum occurrence
- After the maximum occurrence, the occurrence decreases as the time to complete increases
__________________________________________________________________________
Now consider your resource. What odds of success do you want them to have — 50/50? Not likely, and I hope not. You need your resource to be confident in completing their task on or ahead of schedule.
There is a balance to be found between high confidence and an overly inflated task estimate. I have found the sweet spot to be about 85% and encourage my resources to provide an estimate where their confidence is at least 80%.
The statistics show that at an 80% confidence, the time to complete the task is approximately 3 times greater than the time to complete it at a 50% confidence. In other words, asking the resource to estimate the task at an 80% confidence will produce an estimate that is 200% larger than at a 50% confidence. I define this as the task Safety.
Some schedulers remove the safety and build their schedule using the 50% confidence estimate. They might store the safety as contingency elsewhere in the schedule, but this drives the resource to work at 50% confidence level knowing that failure occurs half of the time. Personally, I believe managing this way is disrespectful and a demotivating force on the resource.
What has worked well for me is to trust the resource, give them the confidence for success, and support them in every way possible to achieve it.
Project Management Leadership Coach: 15+ years * Project Management Transformation * Certification Success * Unique Technical Leadership Mastery with PQ Positive Intelligence * AI Project Solutions
5 年Wow great stuff.
Independent Consultant
5 年Good article. Made it easy to understand the asymmetric bell curve. Thank you
?? Visionary PMO Leader & AI/ML/DL Innovator | ?? Certified Cybersecurity Expert & Strategic Engineer | ??? Organizational Transformation Architect | ?? International Best-Selling Author & Keynote Speaker ??
5 年Hi Steven, great article! Will we be able to track and factor in the estimating history of individuals doing the estimating? I.E., track the estimates that people have made and the results of those estimates. Then use that data to create some sort of reliability rating that can be included in the estimate in a way that speaks to the expected outcome of the task(s). By the same token will be able to pull reports that show the estimating history of users making estimates so we can identify estimating strengths and weaknesses and work to improve weak estimates and estimators?