Do you separate OpenAI usage by project internally?

We’ve started using OpenAI more regularly at work and I’m trying to figure out how companies are keeping track of the cost when the usage is spread across different projects. We’re a civil engineering firm and have multiple projects running at the same time so two people can be using the same tools but for completely different client work and as of right now the overall spend isn’t really the problem like it’s that everything eventually rolls up into one number and I can’t tell very easily what project or type of work is actually responsible for it.

My question could be more specific to civil engineering but I wanna hear from anyone and everyone using OpenAI across a team, are you separating usage/cost by project somehow or do you just treat the whole thing as a general company software expense and also how granular it’s worth getting before tracking the usage becomes more work than the money you’re trying to account for.