Methods and Tools has been running a poll on the proportion of software budgets that are being devoted to software maintenance.

The poll question is “What part of your software development budget [my emphasis] is devoted to maintenance. Maintenance is defined as process of correcting, enhancing and optimizing deployed software.”

Only 12% of respondents noted that maintenance accounted for 75% or more of their budget.

I was quite surprised by this figure given that the following studies have recorded figures varying from 75% to over 90% of effort being spent on software maintenance:

Is this simply explained by a difference in the terms of the studies - with M&T looking at budgeted costs and the past empirical research looking at effort? Are people budgeting much less for maintenance than they actually spend? Are M&T poll respondents or is the past research atypical? What do you think?

For further information Jussi Koskinen has a good page summarizing the past research in software maintenance costs - although it has not been updated for a few years.