You actually wrote what is probably a large part of the answer, and then passed right over it as if for a multi-national corporation customer service would be a trivial expense and staffing problem.
There's got to be a "logical fallacy" Wikipedia page for something along the lines of "why did the Hindenburg burn so quickly (other than the large bag of hydrogen, of course)?" :-)
[1] https://medium.com/gabor/timeline-employee-count-growth-for-...
[2] https://www.statista.com/statistics/266206/googles-annual-gl...
I believe a large number of the AirBnB employees are support related (either for hosts or guests).
It's an apples-oranges kind of comparison.
They didn't need to deal with any sort of insurance for their clients (like AirBnB does for their hosts), they didn't have nearly as many tricky questionable real life situations to resolve (the host cancels your housing reservation a couple of days before your have your already reserved flight, what do you do).
Google was just providing search results, and almost all their work was limited to the virtual space. Pretty much a pure software company providing only virtual space services. You don't usually get a complicated real life situation to resolve every time there is an incorrect search result, almost every issue is resolved by making a simple code change.
Due to the inherent nature of AirBnb, it just has magnitudes more non-purely-technical problems to deal with. And you will also need technical work to resolve them too, but it is just a small component of those. For example, how do you deal with verifying hosts? You need to build some verification mechanism, but it will almost always need a human in the loop. So not only you have to implement the technical flow for it (like a verification portal or whatever), but you also need to pay people to process all of this and resolve any issues arising. And I expect that with every heavily human-centric problem, you will need a lot of human resources to deal with it effectively.
Add the fact that AirBnB has to deal with those "physical world" problems in 191 countries (all with wildly different legalities that come with short-term rentals, which is the core of their business), it doesn't seem to have such a large number of employees at all.
Disclaimer: I am not saying that AirBnB is NOT overblown in terms of employment numbers. They very well could be, I cannot tell without understanding the entirety of their business. But I can definitely say that it makes perfect sense in my head that AirBnB in 2020 would need significantly more employees than Google did in 2005.
Here's some context: https://craft.co/reports/s-p-500-market-value-per-employee-p...
To be fair, I do agree, it's higher than I think is right. It's absolutely not abnormal, IT companies are at 2.4m on average, and companies like Facebook (19m) or Netflix (11m), Apple (5.5m) or Microsoft (4m) are much higher.
But I also think we quickly get into apples/oranges comparisons, and it's hard to really make fair comparisons. For one, Airbnb has a very strong non-IT component, and 100% of its revenue comes from real-world transactions with very strong human behavioural elements of non-employees, i.e. complete strangers, that need to be managed. Selling a phone to a customer is pretty straightforward. Connecting a stranger to let another stranger into their home and treat it well, is a transaction with much more friction.
Second, it's not at all clear how employees are counted. Foxcon employs a million people and manufactures the majority of Apple's revenue. If Apple bought all its shares and let it operate independently, its valuation would not change radically (if it was purchased at market value), yet its market cap per employee would drop by huge amounts.
Employee figures alone don't say much. If anything they're a reflection of the company's (supply chain) structure. If Airbnb outsourced all its customer service, you'd see far fewer employees, but a far larger outsourcing bill. One or the other isn't necessarily better. Context matters, without it 14 thousand is a pretty meaningless number.
But again, on average, 2m is not a weird number, its very normal, quite average.
I thought this thread was pretty insightful, for a change: https://news.ycombinator.com/item?id=22822995
TL;DR - big co’s get big because hiring is an optimization problem, and each new hire gets you closer to a local maximum. E.g. if we hire an engineer to improve load times by x%, they’ll pay for themselves. More successful companies have more opportunities for positive-ROI hires, so headcount tends to scale with revenue. (And because the company is making a series of local optimizations, it may actually be worse off due to the drawbacks of being a big org.)
Have you ever built a business that operates in that many countries?