It's the expense post acceptance - the big T&M expenses that overrun and are under-estimated. These are not adding functionality - capex, these are to get what's been built, capitalized and accepted to work - actually work in a true environment with real customers banging on the whole solution.
The fact that it's an over-run is not relevant. In other words, your ability to estimate does not affect that amount of costs that are capitalized. Actual costs either meet the criteria or don't, and those criteria do not involve whether the costs were more or less than estimated.
Costs that are necessary to "get something to work" sound like capital expenditures to me. For example, if you buy a piece of machinery, the costs of installing it are capital expenditures, even if it costs more to install than you thought it would. Costs of modifying the "true environment" to work with the software, however, are not capitalizable as software costs, but might be capitalizable in their own right. Costs of training are not capitalizable.
For example: to the Provider, having to re-design, re-build, re-test, re-implement an interface so it actually works when integrating all of the disparate legacy systems, migrating that data into the new data schema and everything collapses because the legacy data set isn't really what was in the original design....to the Provider it's an enhancement requiring a change order and more money. Not necessarily to the Customer, like Disney. There's no new functionality.
Depends. If you throw out the old work and re-do it completely, then the costs of the work that is trashed and abandoned are expensed, and the cost of the "re-work" are capitalized. If you use the old, and need to add capability, it's all capitalized. Costs of building interfaces are costs of internal use software. Again, if an interface is built and doesn't work, the costs of that interface are then written off and the costs of the replacement interface are capitalized. Costs of migrating data are never capitalized, either on the first attempt or subsequent.
Or those occurring when it actually hits the hands of real users - not the small handful who participated in the true alpha and beta testing prior to acceptance - often the same ones who wrote the spec's for the system and approved the results of the feasibility. Things such as key information not displaying in the correct order to complete their job, or it's so slow to display that a reconfig where the data is virtually staged must happen. Big issue to fix. Big expense, big change order and enhancement to the Provider.
Costs of making the information appear the way it's supposed to sound like software development costs to me. Why would the fact that they are incurred late in the process change anything? Costs of making the software run fast enough for its intended purpose are also development or enhancement costs. Both capitalizable.
Please correct me if I'm wrong. But, if E&Y came in for an audit and found that Disney categorized as capex those above expenses incurred to correct issues or get it to really perform, after Disney signed off on and accepted the product (and wrote the specs)... they might let them categorize them as capex, but they might have to explain why.
You keep inviting me to correct you. If E&Y came in for an audit, Disney would kick them out, as PwC is Disney's auditor. Beyond that, my thoughts on the accounting requirements are above.