Hacker News Reader

Comment by pdimitar | Hacker News Reader

I mean, yeah. How many engineers do you need? 5-6 for the backend, 3-4 DevOps, maybe some more for the various frontends... say, 20?

I can't imagine the whole thing can't be done with between 30 to 50 engineers in total.

And everyone else, what do they do? Cold calls to the entire planet to go subscribe, or what? Also alright, some lawyers and "compliance" people, financiers, marketers,...

Don't know. I'd struggle to fill a roster of 200 people for Spotify.

> Clearly not working on the core functionality of the app - heh.

Sadly, yes. That they become so tone-deaf is something I'll never condone though I do understand why it's happening (or so I think: is it the gobs of money that make people comfortable and disconnected from the bread and butter of the company?)

View on HN

Internal tools for licensing, library ingestion, tools to support data science, customer payments in however many countries they're in, artist payout, search, recommendations, encoding/playback (probably even multiple teams for multiple platforms), whatever their ad sales platform looks like, ad selection, concert listings, merch sales, wrapped (which is probably their most loved feature)

Some of these are a team per region, some are a team per platform, some are entire divisions

Spotify has a lot of front ends to manage, I think more than 20 would be required.

Mobile (iOS, Android), web, vehicle native integrations (Tesla, Volvo, VW, BMW, Audi, Ford), gaming (Xbox, Switch, PS4, PS5), desktop (Mac, Windows, Linux, Chromebook) and voice assistants (Alexa, Google).

Maintaining that variety of front ends must be very labour intensive.

Hm, solid point, thank you.

Let us not forget:

TVs, Wifi connected speakers, home theater systems, Chromecast etc.

Spotify uses Chromium embedded for their desktop clients and webviews for their mobile clients, so that's largely one codebase for 6+ clients.

A lot of the external frontends are also not made by spotify themselves, but implemented by the manufacturer through Spotify's SDK.

You still want a bunch of dedicated staff to support all that, but it's not as dramatic as it could be.

> 5-6 for the backend, 3-4 DevOps, maybe some more for the various frontends... say, 20?

At least 10x that, easily. In a very optimistic scenario.

The back-office part is much bigger than people think (bands management, content management, rights management, recommendations, etc).

Then you have the services like Auth, streaming, encoding, managing CDNs, caching, etc.

>I mean, yeah. How many engineers do you need? 5-6 for the backend, 3-4 DevOps, maybe some more for the various frontends... say, 20?

So Spotify should bring you in as the CTO, right? How can you make such a confident claim that you know better how to run the company that has beat the daylights out of every other music streaming platforms (including Google, Amazon, and Apple)?

Whelp, even for tech-core products, number of engineers are often very insignificant compared to other stuff(legal, license, buying, acquisition, hr, advertising, marketing, sales, security, management, decision board, product management, product owners, agile team, release team, test/QA, support etc.).

I used to think that, tech companies could do with like what 200 people max, but after working on few places for a while now, I am no longer surprised, specially when your service spans the globe(or even multiple countries or continents), you really need a huge team to keep troubles out and the wheels going.

According to LinkedIn, Spotify currently has 13,900 full-time employees, with 3620 in Engineering (26%), 1850 in Arts & Design (13%), 1000 in Media & Communications (7%), and roughly 800 each in Marketing, Business Development, and Sales (5-6% each).

Over the last 12 months, headcount has risen dramatically within Sales (+32%), Arts & Design (+19%), and Business Development (+21%). In comparison, Engineering has seen just a 2% rise, Media and Comms at 0%, and marketing at +10%.

If I had to guess, lots of these layoffs will begin to affect their headcounts within these functions that have experienced rapid year-on-year growth, and affect their Engineering function (despite being their largest) proportionally less than these other functions.

I would be interested in how you came to your conclusion of needing only 200 employees for a company of this scale? Any company of spotify's scale will have entire functions that will be distributed globally and working on a variety of projects or products. For example, Spotify has almost 400 data scientists. Off the top of my head, I can't fathom what I would have 400 data scientists working on, but I can easily believe that a company with over $12bn in revenue and 574 million listeners this year could find a use for them.

Lmao, I love this comment so much because of how incredibly uninformed it is. 5-6 backend engineers to run Spotify in 185 countries... I've seen a lot of ridiculous comments when it comes to company sizes, but yours is probably the best one I've seen in my life, thanks for the laugh.

Is it because you've never worked at a non-startup or how come you have that opinion? Like, you just have no idea what's required/useful and you can't even imagine it?

I'd love to hear how you split the workload between these 5 backenders.

Obviously I can't be well informed since I don't work in there, I am saying that the core functionality of moving bytes around the globe is not something you need hundreds of people for -- because I worked on similar teams. In fact the DevOps team was times bigger than the programmers which might be one clue in Spotify's case.

Have you ever worked at a company with more than say 1k employees that's active in more than one region? I'm honestly intrigued in your opinion, it sounds similar to what I thought back in uni before I had any work experience, but from your profile you don't look like a complete junior, so I'd really like to better understand how you can think Spotify could possibly operate with 5 backenders.

I am 43. :D And with 22 years of experience, 95% of which in backend and some sysadmin-ing.

Though I have only once worked in a huge corporation (and I couldn't understand what did they need all the people for either).

I was almost always working in smaller tight-knit teams that got a LOT of stuff done (too much contracting for my now 40+ y/o self).

So I err on the side of "be efficient" and that's not even for the purposes of cost efficiency. It's more about being able to iterate with a reasonable speed. My observations from my career support what Bill Gates and others said i.e. that the productivity of a tech team starts to decline when it goes beyond 7 people. Generalization, sure, but it's very often true.

As for the 5 backenders thing, OK, my perspective might have been too narrow i.e. "writing code to move bytes from our servers to CDNs to user's devices can't be that hard" and I mostly stand behind it. Sure you might need much more devs to author complex login systems, SSO and such (if you even need it) but again, after the product somewhat stabilizes, how much backenders do you really need?

I am also interested in your opinion. My entire career has been a proof that small and tight-knit teams get sh1t done and everyone else drowns in bureaucracy.

For larger companies, there's a lot of "hidden" functions that you're probably not aware of as a customer. Just payment integrations in 100+ countries, with all the local regulations and reporting requirements is probably going to be rough to handle with a single team of 5 backenders. Then we have functions like marketing content management (every country has different copy and will probably want to be able to surface things slightly differently to maximize conversion), artist/podcast/audiobook tooling, the ad platform, hardware integration (for Spotify in your car/speaker etc), legal stuff (including GDPR and its equivalents), metrics & data analysis tools. And much more, and those are just some of the things you don't really see as an end-user.

I absolutely, 100%, agree that a small and focused team is the best way to get shit done, but for a large company the size of Spotify the amount of work is absolutely massive. I wouldn't be surprised at all if many teams at Spotify are small and tight-knit and doing great work at delivering kick-ass anti-fraud systems or moderation software to detect and report child porn etc.

Not to mention the obvious thing where the higher your revenue, the less percentual impact each employee needs to have to more than pay for themselves. While you might think it's ridiculous to have a full team dedicated solely to the main marketing page, that could be extremely worth it if that team increases conversion by 10%, as an example.

Okay, that's all valid.

But as another commenter pointed out -- it's actually 13900 people.

Again, everything you say it's true but I am finding it hard to imagine the scale and the degree of the problems that mandate ~14k people. Sure, 1000. Maybe 2500.

But 13900?

> I can't imagine the whole thing can't be done with between 30 to 50 engineers in total.

How WhatsApp served 1 billion users with only 50 engineers. https://blog.quastor.org/p/whatsapp-scaled-1-billion-users-5...

edit: https://news.ycombinator.com/item?id=28985169

At the time WhatsApp was only forwarding messages from user to user. They didn't deal with different contracts and license deals in 180 countries. Nowadays they have a lot more country specific regulations to deal with, and also a much higher headcount.

I remember when rolling out Yandex.Music (a similar streaming service) in 2010 we had maybe a dozen developers rounding way up, a half dozen strong BizRel team, perhaps three DevOps and a few managers. Totalling around 30 people, though dozens more would do some things by the virtue of being integrated in much larger Yandex team. Obviously we had way fewer regions/catalogue/listeners/platforms, especially back then.

Also you just need rsync for dropbox, right? :P

It is difficult to understand if this is satire or not.

Why would it be a satire? I've worked on much harder problems than moving anywhere between 1MB to 20MB files from a CDN to user's devices, and we were 11 people, responsible for dozens of terabytes per workday.

Obviously I am not well-versed in all the legal requirements and many other commercial aspects, but to me 9000 is quite insane and surely can be optimized away.

And apparently Spotify agrees.

I wouldn't be surprised if Spotify moves a petabyte of images a day, I can't even guess the amount of data from their songs and videos. But obviously just moving data isn't what requires a lot of employees.

Why would it be a satire

Because it's like a cliché example of an engineer with the view that some relatively successful real-world product—that they have no real insight into—is far easier to implement than the people who _are_ familiar with it have. You see this literally all the time, to the extent that it's become a meme, and it's hard to believe anybody would make that argument seriously.

Spotify might have too many engineers on-staff; reducing the service to "moving between 1MB to 20MB files from a CDN to user's devices" is a flatly uncurious approach to understanding what engineering challenges they might face or if that's really the case. It's a service that _adds_ 100k songs a day, for goodness sake.

Don't you think there is big difference between moving files to some users' devices vs moving files to literally half a billion users with almost 100% uptime? Not considering problems associated with the scale of Spotify and instantly dismissing the task of managing that as something trivial (or at least easier than your work) makes you sound arrogant.

Reply on news.ycombinator.com