Always been curious how “workload” is assigned and estimated for software engineers.
In: 92
The preferred path is
1. Architecture: Determine business processes, functions, people and what they are trying to accomplish. “Domain” architects for things like I.T., Data, and Applications each plan out the big picture on how their domain will need to change to be able to support the vision.
2. Design: The groups begin to break up their respective change areas into logical blocks – these people support these things, this group owns this application, this team stores data. Those groups then look at the what the needed inputs to the larger architecture – oh, they need sales transaction data in order for their plan to work; well, we don’t actually collect all of our sales data the same way. Lets make a plan on what standardized sales data will look like
3. Development: You now analyze your respective systems to determine what changes are needed to create the standard data set. One change might be to update the UI on sales registers to all transmit sales data with a $000.00 format, as some only have $000.0 format. However, we know that rolling out a change to the registers at over 2000 locations will be a big change in the sense of regional testing, integration with the financial system, integration with the data repository, training, communications. Thats too much for one person.
4. Delegation: The UI change will be easy, as its invisible to users – the button now sends 2 decimals instead of 1. However, their SALES numbers will show up on their screens and in-store reports differently every day now. You need to verify every variation of every screen to make sure that the screen can fit the change and not cut out zeroes. Well, thats that just time consuming – we have 87 different regions and logging in, making the change, making sales, and verifying the UI and reports will take about 2 hours each – thats roughly 180 hours, though you’ll get faster. thats an entire month for just 1 person and so we probably want to cut that down to 2 weeks, so lets have 2 people doing that. Meanwhile you still have the financial system, the reporting server, all of the reports, etc to send the data to. We want to say “Well. some systems already send the .00 so it should be fine”. That is a big no-no, never assume technology will work. Someone may have defaulted the .0 senders to *automatically add the second 0.* That would OVERWRITE any value you sent at the second decimal. People do that crap all the time. There are 43 sales reports and we need to verify the reports. Those will take about an hour each. So lets assign that to someone. We now have 2 people working 2 weeks and one for 1 week.
As long as you understand what you are working towards, and all of the things that are in play, you can put metrics around the quantity, complexity, unknowns/risk, and time associated with each one. Those are called SWAGs – Scientific Wild Ass Guesses. Youre usually wrong, and there is always something you didn’t think of so all of you are overworked all the time and things are rarely on time or under budget.
Latest Answers