I guess a significant number of employees are just working really hard on things that don't matter. When every single thing that needs working on is already being worked on, a lot of people are going to be doing busy work. There was a post by Stevie Yeggie where he talked about working at google. he looked really hard to find a new project to lead but absolutely everything was already being worked on, he found. eventually he just couldn't see himself leading or doing meaningful work there and quit.

the lesson here, i think is that there are limits to how much can be done, without moving into new industries and spreading horizontally.

But isn't that just bad management / prioritization? Isn't the solution to just shuffle people to more useful projects instead of firing them? Especially at a company that hires general engineers who can work on any project, and also at a company that has a lot of internal tooling which makes ramping up new engineers slow and tedious.

They'll probably grow back to 190k within a few months, but those 8k new hires will probably take months if not a year to fully ramp up.

In my org at the Goog, we didn't have any layoffs, but we also aren't expecting any new headcount any time soon. Projects that need more staff are expected to cannibalize it from other projects under the same VP.

I feel like this is actually a good way of doing things, and wish Google did this more broadly. Just think if everyone who would have been laid off was instead transferred to a project that needs their skills, or given training to move into a role that's needed. Can you imagine what type of employee loyalty and morale that would engender?

Google used to be exactly like that up to a decade ago. Talent was considered core for the org so you would be relocated rather than cut off.

Values shifted with the focus on earnings and growth getting less impressive.

Rewarding managers by headcount is like rewarding software engineers per line of code, yes, but that is how basically every company rewards managers.

Imagine how obnoxious code reviews would be if everyone wanted to submit as many lines as possible for their salary to go up, that is how management headcount requests works today.

The problem with this (and I was also in an org that didn't get any headcount -- for several years) is that those non-growing teams often ultimately just wither and die, whether it's through internal transfer, attrition or layoff. Google doesn't really reward managers for their managerial skill, and there's nothing that says "managerial incompetence" at Google more than having the same size team for years on end.

Maybe that’s okay though? It’s not like every team is supposed to exist forever. You could argue it’s ineffective or inefficient at this, perhaps.

The idea that a manager’s success should be measured in the number of people below them is truly idiotic. That is how you end up with companies like Google increasing their headcount by massive amounts during the pandemic only to have to turn around and do layoffs a year later.

Middle managers serve a role but honestly many of them can be replaced without really disrupting anything. At a certain point, you become too far removed from the actual work that is being done to really matter.

On the other hand, there are many low level managers with teams of fewer than 10 people who can be very valuable and hard to replace. People with deep expertise in the specific area they work on.

Measuring a manager by their headcount is just a lazy way to evaluate impact and calculate salary.

the problem is there's not enough high value projects to do. to much labor, too many employees and too much funding relative to the number of projects that actually bring value.

Because all the high value stuff(high ROI stuff) is already done or already being worked on. all that's left is low ROI projects which take up huge amounts of time but bring little value. employees and managers know they generate little value, so they end up doing more of them to compensate which makes the problem even worse as there's less and less high value stuff to do.

there's just too few opportunities, hence the need to expand into new industries.

Of course there are enough high-value projects to do. You just need to be willing to experiment, expand horizontally, and let your employees go out on a limb with new ideas; something that no corporate leadership anywhere has ever been willing to do. They didn't run out of valuable problems to solve, they just have shit-ass worthless leadership just like every other (public) corporation in America.

well, it's not like they're not trying to branch out. the list of failed project at google is incredibly long.

i sometimes wonder if they would have more success with atoms based innovation rather than bits (reference to avc article on the matter). but that would necessitate hiring something other than software engineers.

> Isn't the solution to just shuffle people to more useful projects instead of firing them?

Yes. However how do you find those less useful positions? And identify whether it's the developer who is less productive or the position?

For top management the easiest way is to fire people broadly and assign it all over equally by pushing the decision down the management chain. Medium managers know their teams a tiny bit (but often not much ...) than too management and identify the least troublesome positions.

And then you let them bring up cases to keep their headcount or hiring new.

Works, after a short bump in productivity, if it's just (human) resources and numbers in a sheet. (Similar to other budget cuts, where you give a Mandate to reduce cost by X% everywhere and then see who cries the loudest and who adapts) Of course has impact on trust etc. which has an invisible cost.

You're pointing to poor leadership which isn't an uncommon refrain coming from people who know Google.

Good luck finding useful projects for 180k people. That's just a ridiculous amount of people and projects.

There are more than just SWE's employed at google. Custodial, food service, administration, operations, HR, etc...

A lot of those are going to be contractors employed by agencies.

Google also outsources SWEs. My friend in Mexico worked as this. They have an acronym for this kind of worker, I can't remember. But for example, they contract out work to SWEs at https://www.epam.com.

Not contradicting your point really, just a note

TVC

Yes! Can't believe I forgot. He was an interesting guy - a biologist who was deported to Mexico after living in the US since childhood without citizenship, and then tried working in a lab in Mexico, but the pay was so bad he learned to code and eventually became a TVC.

Extremely cool individual who I miss. Maybe I should say hi to him.

Just wait, they will outsource to “AI agencies”.

I don't think you realize how many products Google has. Cruise had ~ 4k employees at its peak. Waymo is just one tiny project in Google. Splunk had 8000 employees, GCP has a competitor to it. Bigquery is a competitor to Snowflake, who has around 5,000 employees. Databricks had 4k employees, which GCP also has a competitor to. This is just a fraction of the software just in cloud.

That's not apples to apples, though. Snowflake has 5,000 employees in the entire firm. The BQ Product+Eng+partnerships team is a fraction of that, and the SG&A functions are all shared services across all of Cloud, just like your Databricks comparison.

I think that says as much about those other companies as it does about Google. Namely, that despite Google's reputation for having become a bloated bureaucracy, there are in fact economies of scale happening at Google, and in fact it's those other companies that are bloated. 8,000 employees for collecting and searching logs? 5,000 employees for a data warehouse? Even accounting for the share of sales, marketing and G&A employees that contribute to Google's equivalent products, I'd be shocked if Google has anywhere near as many employees working on these things.

In fact it's even worse than that. A lot of Google's employees are dedicated to maintaining tools and infrastructure that Google built up over decades and hundreds of thousands of engineer-years because "the cloud" didn't exist when Google started. These other companies get all that "for free" (or rather, for nominal pay-as-you-go fees -- but certainly without the massive engineering investment that Google put in over the years) by building on AWS, GCP, or Azure. Without public clouds and the vast growth in open-source cloud infrastructure tooling, those other companies likely wouldn't be viable businesses.

Waymo and Cruise are valid comparisons. GCP competitor to Splunk (Stackdriver) is a complete joke. I would be shocked if more than 500 people are working on that thing

I do realize Google has a lot going on. They have to, to keep all these people even remotely busy.

180k people? That's not how many people were fired. 8k were fired, which is less than 0.5% of the company. Finding useful projects for that many people doesn't sound that ridiculous.

It's quite possible for the engineers to work really hard in ways that even directly harm the company, either directly, or by increasing the political costs of making a better decision.

My favorite here is Bridgewater and their HR/Evaluation/Alignment system. There's books written about it. Developing it and running it took a high percentage of workers and contractors. Its implementation had all kinds of negative effects on everyone else, including the people doing the actual investment. But it took a pandemic, and the CEO being sidelined, for the whole thing to be basically thrown away.

I'm aware of their alignment system from the book Principles, but I'm curious why/how you came to the conclusion it was a negative?

Pandemic and larger-than-life CEO stepping down both seem like the major outliers in the dynamic you described. Bridgewater was very, very successful for quite some time with that policy in place.

I had always assumed (perhaps cynically) that Bridgewater's elaborate system was part of its marketing to people who run endowments and pensions. It gives people a sense that they are working with the best and hardest working, even if in reality their returns are equal to other less notable firms.

Read "The Fund" - an entire book on the matter.

You don't just multiply employees x time = revenue. Google develops IP, systems, and processes over the course of years which pay out on long time scales. You would probably start feeling the financial effects of these layoffs in a couple years, and continue feeling them basically forever.

This is anecdotal, but whenever I read details about layoffs at FAANG companies ("$foo people were fired from the $bar team"), I'm shocked by the headcounts.

I get it, developing and supporting some feature at a company like Google requires more people (both in engineering and elsewhere) than if you banged out a prototype of something similar at a startup. But I've read stories where hundreds or thousands of employees were laid off, and they were all working on some small service (or even an obscure feature of that service) that never had a chance of moving the needle for the company and offsetting their costs.

I'm sure most of those employees were hard-working, motivated, and tried to build the best product possible, but there are no other companies on earth that would pay that many people top-of-market salaries to work on anything that isn't existential to the company. Other companies would simply go out of business if they operated this way, but in big tech companies, this is lost in the noise when times are good. But when things slow down, big tech companies start to pay very close attention to this.

The key word is "meaningful" and that's going to vary depending on the employee's interests. I transferred a few times at Google. When I looked at internal job listings, there were lots of jobs that I didn't want to do.

I suspect that many good jobs went to people who learned about them in some way other than official job postings. (This is generally true.)

Talking to some Googler friends, it also seems like everyone there think they are "leaders". The reality is that most people at google end up working on something meaningless. That's the reality when you are in a company of 200k where the core business is at complete maturity.

> Talking to some Googler friends, it also seems like everyone there think they are "leaders".

I guess we have different friends, but that doesn't match my experience. Of course, pumping yourself up and making yourself look more impressive for perf/promo is a huge thing, and accounts for a lot of pointless stuff being done just so you can say you "launched"/"landed" something. But the flip side of that is the recurrent memes about stuff like so-and-so over in that other division solved some incredibly hard technical problem, and "I'm just here copying protos around." Lots of Googlers realize they are just worker drones.

I think this is just unwinding from overhiring during the pandemic.

- 2021: 156,500

2021: Yay! Crazy hiring!

2023: What?!?! Laying off a fraction of recent hires? The world is ending!!

Obviously this sucks for the laid-off, but…you don’t need a very long memory to put this in perspective.

More than a fraction of recent hires. I had long-tenured teammates (10+ years) who were let go, for no particular reason.

Sure, it’s not LIFO, and this churn can be a nasty way to get rid of the expensive oldsters. But as a percentage, the bigger companies are generally still well above 2020 staffing levels.

I bet either their comp was very high due to their tenure relative to their output, or they spent too much time at the same level and weren't showing upward trajectory. I know that you only need to be an L4 officially to get past the up or out stage, but that doesn't mean that they won't find people to lay off who have been an L5 for 12 years

Top grading is the MBAs dream. Replace someone who doesn’t stand out and get someone who on average will be average, so presumably better than the 10y vet you just exited. Don’t worry about what they know, because coders are monkeys.

Yes, some tenured people suck. But it’s their salary that drives this behavior.

It's was more like

2021: this is clearly a bubble

2023: execs do "shocked Pikachu", "we have hired for a different economical environment!"

Agree. But will many execs pay for that groupthink mistake? Nope, somebody else gets to pay…

That doesn't include the army of contractors they have/had that were laid off.

To me it shows that Google doesn't know how to generate significant revenue with those eight thousand people. I am worried it indicates Google is running out of good bets and has limited future growth.

Of course, the counter argument is that technology allows Google to do more with less people. I am not convinced by this argument, because Google's competition has about the same ability to do more with less people. If you can improve individual productivity than why not scale that across more people.

To summarize, this indicates to me that Google is reducing their value on human capital.

One year is barely enough time for the marginal impact of those 8k people to matter.

Say that made 3 revenue products revisions slip by 2 quarters, and not ship six cost-saving optimizations (if you make google search 0.001% faster, you save millions in compute). Further, say that made 100 companies not renew their 7-figure ad contracts. If all of this was evenly spread throughout the year, it still hits revenue and profit only a comparitive small amount in one year.

It takes time for those sorts of changes to propagate all the way through the system until they finally impact revenue and costs. The effects are long term.

I'm on record as saying that Google hired way too much and way too fast, but it isn't quite as simple as, X-people produce Y-revenue, so (X-1)people produce Y-Z revenue.

The folks who make up "human capital" have been reducing the value that they place on Google the employer, too.

I mean that in several ways: Google doesn't have nearly as good a record of successfully recruiting (or retaining) top-tier talent as they did just 5-10 years ago. It is no longer seen as a dream position, or even a strong candidate's first choice. Similarly, hiring managers who see Google on a resume don't treat it as highly as they did 5-10 years ago.

Google is rapidly on its way to becoming just another big boring technology company, like SAP or Salesforce or Oracle. Now the culture of those around it has finally started to catch up.

google doesn't want more bets. It wants best bet possible and for now this bet is ads, that's why so many products are killed

Clearly, those 8000 employees were producing -$10,000,000,000 per year.

A Google SWE costs the company around $750k average, fully loaded cost. So if those were all SWEs, $6B/yr isn't an unreasonable estimate.

Where did you get the $750K figure? I'm in the Bay Area, Sr Engineer, on the higher end of the pay scale, and no-where close to that figure.

If I were to guess, I'd say that this estimate is including hardware, cloud resources, software licenses, benefits, stock options, and probably more that I can't think of in addition to salary, but I too would like a more detailed explanation on how they came by that number.

Probably also the HR, legal, etc. support staff amortized across headcount.

How do you know how much your benefits and support staff cost?

Fully loaded cost. Typically you estimate that an employee costs 2x their total compensation, across various expenses.

Where is that typically? Most of those are had costs for an employee, not % based. Google has employees from $50K to $1mm+, I very much doubt that all expenses double as income doubles.

Assuming those SWEs produced absolutely zero value to Google, which I bet is not true

Also, I'm guessing legal fees and the raw cost of severance are no joke.

Nothing about my post was estimating the added value of the SWEs, only the cost.

The median in 2021 was close to $300k as per this[0] source.

[0] https://www.spglobal.com/marketintelligence/en/news-insights...

Math checks out. Imagine what happens when they lay off 800k!

I wonder how the TVC headcount has changed y/y. They don't publish that data.

Well, if total average costs per employee for G would be very roughly 1 million, there you have it. I think it may be half of that if at all, but it impacts the increased revenue significantly in any case.

Costs do not impact revenue only profits.

I agree with the sentiment, but costs do impact revenue when those costs represent an investment in something that can be used to increase revenue, like employees.

+1. Employees can generally boost revenues (if used correctly).

The post I was replying to was implying that cutting employees boosted revenues somehow.

I reckon they’ll be a deeper cut

> The company added that for 2023, it's taken charges related to employee severance of $2.1B, with $1.2B in exit charges taken for the last three months of the year. It's also continuing work to optimize global office space.

https://seekingalpha.com/news/4060311-alphabet-falls-4-as-ad...

Revenue increase of 10 billion is about 3%, essentially tracking inflation.

Short term revenue gain (2~3 years) can be all done by "optimizing" core ads businesses but that's not sustainable for a decade. The real question is how to build a longer term growth potential for Alphabet, and probably more than half of the employees are working on less certain business areas to figure it out.

I got the total employees for the past few years for a comparison:

2019: 118,899

2020: 135,301

2021: 156,500

2022: 190,234

2023: 182,502

https://www.macrotrends.net/stocks/charts/GOOG/alphabet/numb...

It is absolutely mind-boggling that a company as large as Google could increase their headcount by 62% in three years.

Not really, their annual revenue went from 161.857 B (2019) up 89% to 307.394 B (2023).

Nah, I'm pretty sure that's just also mind-boggling lol

Has Google done anything new and major in these years? Like as a product.

So could there still be 10-20% extra as workers?