Federal Contractor Games Part 3: Can an IT contractor be your PARTNER? – Estimating Level of Effort (LOE)

 This is the third posting on federal government–contractor ‘partnerships’ for IT work.  In the last few days I’ve seen more discussion and marketing of these types of relationships. President Obama’s comments on fixing the procurement system for IT are likely to spur more agencies to seek ‘partnerships’. For the agencies contemplating such a relationship I hope that they get some take-always from this series.  But more importantly, they need to seek the lessons learned from agencies that have already engaged with “partnerships”.

  In these relationships, the agency expects to get a good product at a fair price.  The primary driver of price is typically the labor content of an IT development project. Another tactic that a profit-seeking contractor may attempt to use in partnering relationships with Federal IT organizations lies in the proposed Level of Effort (LOE) for an undertaking.

Estimating effort for tasks and projects

  In a partnering relationship where there is no immediate or continuing competition wherein there is pressure for ‘honest, or bare-bones LOE estimating’, the agency needs to figure out how to independently assess a proposed level of effort for a task or project.  All contractors have their own staffing and pricing methods or models.  The agency must either have it’s own models for estimating LOE or must ensure that it can use the contractor’s model and duplicate results.

Typically these models are driven by parameters such as estimates of: complexity; function points; number of interfaces; number and type of users; response time; and the nature of the system (reporting, real-time, web-based, compute intensive, and many other types).  So, unless the agency understands the inputs to the model as well as how the model calculates the LOE, there is potential that the contractor/partner will ‘overload’ some of the parameters to bloat the LOE beyond an objective set of parameters. 

 

A good approach would include the agency having an agency person or team versed in several models which they would use to develop the agency’s own estimates of LOE.  Some of the commercial models use a database of past projects for comparison, others use algorithms that have been derived from past industry experience.  Each approach has its pros and cons, and produce differing results.  An agency should use multiple types to establish a probable RANGE of LOE, rather than a POINT estimate of LOE.  Since ALL of the parameters that drive the models are ESTIMATES, the agency needs to establish upper and lower bounds, and use that range against which it would compare the partner-contractor’s estimates.

  A ‘too low’ estimate from the partner can be more worrisome than a ‘too high’ estimate.  A low estimate probably indicates a lack of understanding of the requirements, and will lead to problems during development (e.g. corner cutting: using lower qualified/cost staff, skimping on essential steps such as testing).  A high estimate may also indicate lack of understanding but provides padding the LOE to cover the risk.  In either case, the agency itself needs to have a deep understanding of the target system’s internal technical characteristics. 

To paraphrase a friend and colleague “You can’t outsource your thinking”.

P.S. The software estimating models that we used on most projects included: SLIM, PRICE, PRICE-S, @Risk, COCOMO, but there are many others.  An agency should develop expertise in several.

Leave your email above to receive new postings