This is a pretty good introduction into the nuance of COGS, but I find that it really isn't as useful of a valuation metric unless the company primarily sells inventory of some sort. MoviePass is an interesting example because it is sort of a mixture between a tech company and a traditional retail company with inventory, and its failure seems to be somewhat due to it trying to act like a pure tech company leveraging a deficit spending strategy.

For tech companies, the important thing is to understand the breakdown of your total costs: fixed and variable. Compare MoviePass to another tech company like Facebook. They both have similar fixed costs like rent, servers, and developers, but Facebook's variable costs per user are bytes of data which cost fractions of a penny, whereas MoviePass had larger variable costs per person because they actually had to buy the movie tickets as "inventory". You can pay a single developer a fixed salary per year whether 10 people are running their software or 10 million people, but your costs will be vastly different if your users see 10 movies per year compared to 10 million. MoviePass was gaining $10 per month per user, but since each user was costing more than $10 per month, no amount of scale would have saved them.[0] This is very different from companies who have virtually zero variable costs and actually increase in profitability (and valuation) from each additional user they recruit.

[0] Unless they pivoted, renegotiated costs, or added other sources of revenue.