Tweaking the Single Usability Metric (SUM)?

I think Jeff Sauro's proposed Single Usability Metric (SUM) is a great idea but it doesn't work for my testing purposes.

One problem I have is that the task time metric is assessed on whether users complete the task within the specification time. However, for the product that I'm testing, I did not have the resources to think of a specification time and then time how much longer users go past it. Instead, I had to set a cutoff point of 12 minutes. This makes the task time metric redundant in my case because it's basically just performing the same role as the pre-existing completion rate metric.

Is there any way I can tweak the SUM computation so that the task time metric is not percentage of users that are within the specification but, rather, the average speed (in seconds) in which they complete it?

Alternatively, is there a different combined usability metric that I should be using?