Reliability\Quality of Data

re: Reliability\Quality of Data

greg — Fri, 10 Oct 2008 14:51:06 GMT

This empahsizes the need for systems to be able to incorporate both bottom-up and top down resource planning and tracking.

Top-down is more about portfolio level resource allocations, priorities, and organization capacity than it is about managing individual projects.

To me, if you have a very heterogenous workforce and a limited need to track detailed human resource monetary costs on individual projects, the amount of effort required to capture bottom up resource data (planned or actual) is hard to justify.

re: Reliability\Quality of Data

brianken — Fri, 10 Oct 2008 19:51:09 GMT

Im confused a bit. Your statement about top down being about portfolio level and your statement about bottoms up being hard to justify seem to be in conflict. If top-down is about portfolio level allocations and priorities and bottoms-up estimates are hard to justify based on level of effort then where are projets getting their estimates and status information?

I also dont think that my post is emphesizes the need for both (though I do think that both should be in place in a good system). I think it emphesizes the need to be careful when looking at ANY data (bottom up or top down) that is far into the future.

re: Reliability\Quality of Data

jack dahlgren — Mon, 20 Oct 2008 20:59:31 GMT

Brian,

I think the thing that people really want is a seamless way of forecasting demand and availability. Surely, your precise needs three years out are going to be fuzzy, so people estimate them in some sort of spreadsheet and build their long term plans on those estimates. Perhaps they even work on a recruitment and training plan to get those resources in place. Then as they get closer they want to look at things a bit more closely and finally get to individual assignments.

Right now Project and Project Server and Portfolio Server only do this in a very awkward way. Working with FTE's (full-time equivalents) isn't supported in the data analysis views and there is no easy way to do things like resource constraint analysis.

Doing this while still displaying a confidence interval would be a great help. I expect that would take a bit of training to achieve though. Many are used to taking the one number that excel gives or the project finish date that project gives as "the answer" when in reality, it is just one of many possible answers. Until people understand that, there is little point in spending much time generating the models in the first place.