This paper presents a simple, fast method (or management tool) for the anal
ysis and improvement of software-intensive complex products and systems (Co
PS) called software analysis-software improvement (SA-SI), The tool relies
on outside intervention, rapid data collection and structured in-company wo
rkshops. The distinctive feature of the method is that it focuses on the 's
oft', human side of the software development process and examines and compa
res formal (or rational) processes ('what should be') with real, actual pra
ctices ('what is)), in order to identify problems, their causes and strateg
ies for improvement. The tool complements existing formal approaches by del
ivering a 'bottom up', grass roots, practitioner view of real processes in
action. The purpose of SA-SI is to help overcome the severe problems of mea
suring, analysing and improving performance in large scale, complex softwar
e projects. An illustrative case example (Company X) is used to show how th
e tool is applied and how it confronts the problem of actual/real processes
differing from ideal/formal processes. It also shows how SA-SI is used to
identify process 'hot spots' (severe problems), analyse their causes and id
entify solutions. The paper provides guidance on typical problems encounter
ed in running SA-SI and how to overcome them. It also shows how the tool ha
s been modified and extended to deal with other complex domains and innovat
ion management issues. Although SA-SI cannot be a substitute for a change p
rogramme, it can play a useful part in complementing ongoing improvement ac
tivities. From a research perspective, the method helps link up studies fro
m the organisational development and software fields and assists in 'closin
g the loop' between innovation research and business practice.