Feel invited to ask!

Classic approach allows to treat the office rental as fixed, General & Administrative expense (GA). It's simple and tempting, as rent per software engineer is typically lower than payroll and taxes, and it is kind of fixed - companies do not relocate or change office space often. Yet, with this approach management might believe some project is say +5% profitable, while it is actually -5%.

It's crucial for a business to differentiate between positive and negative project contribution to company gross profit. In fact, it answers the "to be or not to be" question: whether you would save or lose money if dismissed this client. That's why office rent should be logically reflected in COGS (cost of goods sold).

So, our approach would be to split the rent payment into 2 parts: "variable" software-engineer-related cost, based on number of desks / part of premises occupied by software engineers, and GA expense - all vacant desks and administration. Idea behind is that hiring more people will finally incur adding more office space, and that now-vacant desks are due to premises limitation (not flexible). Also, administration rent share is no way a direct result of project- or line managers decisions.

Particularly in "2-story PnL" method, the share of monthly rent sums up with the rest of employee monthly costs - payroll, taxes, etc - in each software engineer's "personal profit-center". Line manager-owned PnL is a sum of financial results of all people reporting to him by solid line. Project manager-owned PnL, from the other hand, will be isolated from rent, payroll, sick days and other employee-related parameters.

Thus, if Software company adopts 2-story PnL, it is Line manager's responsibility to worry about if his direct reports generate revenue higher than desks rent plus payroll plus taxes and other costs. And, it is Project (and Sales) managers incentive to find a proper balance between factual scope and initial estimate + markup negotiated with client.