[danpalmer]

I've long felt that consumer acceptance of this pricing correlates with whether the subscription has a cost-of-good-sold component.

Desktop software does not impact COGS, and people near universally hate subscriptions for desktop software. File storage obviously has a COGS impact for the physical drives, and no one questions Dropbox/etc charging for their cloud storage (even if the price is an order of magnitude disconnected). Notably, customer support is not usually considered part of COGS, and doesn't scale in exactly the same way as the general variable costs associated with delivering a service.

[gruez]

>and doesn't scale in exactly the same way as the general variable costs associated with delivering a service.

You underestimate the technical sophistication of the average user. Even with perfect docs, there's going to be 1% (or whatever) of customers that call into technical support asking questions. That's essentially COGS.

[danpalmer]

Sure there's always going to be someone, but it's hard to attribute a specific figure for support cost to any given sale (you don't know about future support requirements, or for a subscription when do you account for support costs). This is why COGS usually doesn't include support. In the same way, marketing costs vary wildly per sale and are very hard to attribute, which is why marketing is usually not a part of the COGS.

[gruez]

>Sure there's always going to be someone, but it's hard to attribute a specific figure for support cost to any given sale (you don't know about future support requirements, or for a subscription when do you account for support costs). This is why COGS usually doesn't include support.

By that logic, do cloud companies need to factor storage costs into COGS? Most people don't store exactly 1TB when they get the 1TB plan, and many don't use anywhere near that amount. Does that mean dropbox can pretend their offering has 100% margins? Sounds unlikely. The principle of accounting is that the numbers should reflect the actual business. Customer support costs don't have to be allocated to sales with 100% accuracy, but they can't pretend it doesn't impact unit economics either.

[danpalmer]

Look, I'm not an accountant, if there's one here to correct me on this I'm keen to hear it. I can understand that it may be useful to associate customer support costs to accounts in some ways. It obviously impacts unit economics. But none of that is what COGS is about (again, marketing is a major factor in unit economics and not traditionally considered in COGS).

COGS is about me as a customer buying a widget, receiving a widget, and the seller having to have made me a widget. That's an incomplete view of a business, but incomplete views can still be very useful. Understanding customer support cost for example must be done over some time horizon, i.e. what's the 6 month post-purchase support cost, which means we don't know the cost for 6 months. COGS is known much faster.

As for whether it should include storage, or any other particular piece of technology, for a SaaS business, I guess that probably depends on the business. The important bit for accounting is the direct cost of providing the service. If you're renting out GPUs by the hour then clearly a GPU hour is a direct cost. If you're just hosting a web app and not selling any particular slice of infrastructure then that's probably not a direct cost. Fixed size plan storage is probably somewhere in the middle, although cloud storage is clearly a direct cost.

Still, my point is really that businesses should only use subscription pricing when there's ongoing COGS, because that's what people intuitively associate with ongoing value most of the time.