bmc::KnowledgeFlow

I was once on a defence project that had been running for 18 months and was 18 months behind schedule. This same riddle is faced by most software projects at some stage and is easily solved. It stems from the inherent problem that once you start the journey you realise there is much more required to be done than originally anticipated.

One of the techniques used especially on these large projects to track progress is Earned Value Analysis. On large projects, it is an essential tool to allow the accounting side of the business to be able to claim revenue. Used properly is also a very simple and powerful tool to give an instant snapshot of the health of a project. How much work ($) has been expended vs how much we thought we needed ($).

I have found in software projects (or any projects that primarily rely on human intellectual output) this can be dangerous.

The reasons are simple:-

• Most people are capable of focusing on the following timeframes; daily, weekly, monthly.. Whereas corporations focus on monthly, quarterly, yearly.
• Most people leave things to the last minute (or probably more correctly, prioritise “just-in-time”)
• So most project plans are set up to show people working on multiple tasks over a monthly/quarterly time frame.
• When the project manager, asks for the status how far are you complete on the document you are doing this month; the reply is usually based on elapsed time not actually progress. Hence your EVA looks fine, but a storm is may be potentially brewing. Especially if the task is not easily defined.

My 14 Steps to successful Project Management:-

1. Tasks have one of 3 states; either Complete, Not Started or Started.
2. For tasks that are started, progress is reported as either “on target” or new “target is”.
3. Break tasks down to avoid multi-tasking…so people can focus on what they need to specifically do today & this week.
4. No tasks can be longer than a week. (any longer you don’t fully know what you are doing)
5. No task shorter than 0.5 day (you are managing to too much detail)
6. Don’t try to use a project plan to track cost & progress at the same time; If you must track cost use timesheets (my preference is just allocating your time monthly to project codes-statistically it is just as accurate)
7. Don’t forget to factor in vacation/sick leave/public holidays….I usually create a bucket of time per person/quarter and then allocate as people request time.
8. Senior Managers/Technical Leads are supposed to multi-task and are always trading-off priorities. Other than tracking specific outputs/dependencies from them I would recommend keeping them off the project plan (it only confuses the plan).
9. All staff need to help people not on the project, attend meetings, do small activities relating to the project…just put the time in a bucket and so you know they are not available 100%.
10. Keep the project plan up to date and have no more than 150 tasks/project manager (3 printed pages)
11. Get individuals to make/own the estimates;
12. 12. Give the person working on a component that spans multiple weeks, one chance to revise their original estimate;
13. Have status reports to report up to Friday and not allow the use of weekends to play catchup/hero stuff;
14. Stress that you do not like late surprises; e.g.  one day before something is due, you are informed it will take another 5 days.

expanding on (14). This is one of my top hot buttons. My rule of thumb is that I will accept a slippage only if the extra time required is no more than the current planned amount of time remaining on the task..i.e. if you have 1 day to go on a task, you can slip by another day. This encourages a behaviour of letting you know earlier if there are problems (when you have time to correct) rather than when it is too late. As they say in the air force; “The altitude above you is useless”.

There is a potential downside to these techniques. The software industry (& the people who get the bonuses) has been founded on obfuscation, last-minute heroes, martyrdom and everyone going over the trenches to work through the night to hit the deadlines. The ones who manage themselves professionally and ensure that their evenings are spent with the family or working on open source projects are somehow considered to be less competent & committed – go figure?