What methodology do you use for making sure that estimations are accurate?



As part and parcel of my professional life, I have come across numerous instances wherein the Higher-up's have gone to varying length's of extremes asking me to come up with "Estimates", Yes Accurate and Precise estimates.


And when asking for the same, there have been very-many instances where-in the Grey's in the "To-Be-Requirement-List" have outnumbered the whites 1:1000.

Yet the demand has always been the same "Give me the estimates". In some places I've had the luxury of time, In some places I haven't had any.


The story does not end after estimates are given, after a period of time has elapsed, folks from the Quality Assurance team, the Quality Systems team, come knocking, this time for metrics.

Along my journey through all of this, I have gone ahead and picked up some tricks-of-the-trade, that I would like to share with you, so that the estimations that I give out end up being as Accurate(closeness to the truth) and Precise as possible.

At the onset, let me clear my conscience and state, initial estimates are most likely always Guess-timates, and over time they get better or in peculiar scenario's murkier.

Here goes:

  • Try to identify if teams are under-committing and over delivering, 
  • Try to identify if teams are bumping up story points\LOC's\FP's\Bug-Fixes for meeting KRA's, 
  • Try to identify if teams are taking a lot more time to complete coding the defined user stories. 
  • All of these are symptoms that our initial estimate and the actual reality are awry. These are yet some symptoms that we are not going to get precision with respect to estimates made.
  • An easy way to estimate the accuracy:
    • (Estimated end-date - Actual end-date = as close to 0 as possible) 
    • (Estimated cost - Actual cost till date = as close to 0 as possible)
  • Another derivative of the same is:
    • (Estimated number of hours - Actual number of hours taken = as close to 0 as possible) 
  • A Scrum derivative of the same is
    • (Estimated team velocity) - (Total story points burnt in time period) = as close to 0 as possible.
Most of the time the above mentioned help to give out metrics to the Quality team, But for me to hone my skills, and for folks reading this to keep estimates as precise as possible, do try doing the following on a regular basis.
  • Try to collect as many metrics from as many axis's as possible.
  • Calculate the average of your collected metrics.
  • Calculate the difference of collected metrics - average. 
  • Closer the difference to zero better is the estimate.
  • Use the collected metrics the next time anybody asks for an accurate-precise-estimate.
  • Keep the loop running.
I would love to hear your views. Please leave a comment in the comment sections so that I can learn from your experience.

If you liked the article...please share with your network!, In case you would like me to speak/train/coach about this stuff, do get LinkedIn with me.

Regards,
Manuswath

Comments

Popular posts from this blog

DumbCharades (DC) and its different rounds or concepts

Gaalipata .... Hype ???? Hope Not !!!

Idhu Mussanje Maathalla Mussanje GoLu