The Rising Costs of Scaling a Startup in San Francisco
tomtunguz.com
tomtunguz.com
I am a long-term open-source and commercial dev, former serial CTO, with an additional specialization in recruitment software, but I also work as a recruiter. Recruitment in SF sounds like a nightmare for everyone involved. Agency fees are relatively low (10%ish, compared with 15-20% in the UK), but the service the candidate gets (incessant LinkedIn spam for inappropriate roles) is poor, the service the employer gets (scatter-gun approach, loads of shitty CVs, very little value-add from the recruiter, other than they'll spam LinkedIn for you) ...
I'm guessing it's a combination of the generally slightly immature state of the US agency recruitment market (compared to Europe; not meant as a judgement, just a fact), insane competition for candidates, very small take-up of and limited competition in job-boards/channels, and the big competition with corporate recruiters against FB, Google, and so on.
Every so often I wonder about trying to break in to the SF market (rather than London, which is where I put 90% of my time), but I wonder if the effort would be worth it, given what feels like the dire state of the market.
All the recruitment startups I can see are trying to build a better ATS to remove the agency recruiter (Lever.co, Greenhouse.io, SmartRecruiters.com, TheResumator.com, and even ZipRecruiter (an ATS disguised as a multi-poster)), yet no-one seems to be trying to fix agency recruitment.
Sorry for the slightly rambling post - happy to give free advice to anyone trying to recruit (see my profile), but it sounds like a real uphill battle in the Bay Area...
It might also be easier to find qualified people in the bay area. Certainly they'll cost more than just about any other place, but how important is it to you to find say 5 people within the next month or two instead of the next say 6 to 24 months?
I've never needed to actually do either of those two things, so I might just not know what I'm talking about, but for a startup that isn't bootstrapped I'd think that getting a long runway (for wherever you are) and advisers/investors who are knowledgable about your company's market are more important than the raw monthly expense figure, and that quickly getting qualified people working on the product might be more important than saving, say, $75,000 in salary per employee per year.
> but how important is it to you to find say 5 people
> within the next month or two instead of the next say 6
> to 24 months.
While I am led to believe that the easiest money is in SF, and there are some awesome investment partners, don't assume the best advisers are necessarily there; it's not been true for any of the three industries I've worked in, namely high-end fashion, recruitment software, and news/PR analysis.
Digging specifically in to recruitment, the fact that every recruitment startup appears to be another ATS (Applicant Tracking System) - by far the most competitive space, also the one where the established players will be some of your hardest to displace (the reasons are dull but real (ATSs have long tendrils in to all sorts of business processes)) - seems testament to my hunch that "fixing hiring" companies are being guided by advisers with little in the way of recruitment experience. Possibly I'm wrong, and disruption needs to happen from outside, but the prevailing view that the expertise in tech in SF translates to expertise in every thing seems suspect to me.
I'm not surprised at all that there are probably better advisers outside of Si Valley for many industries. I'm a little surprised about recruitment, but not shocked at all for fashion or news. SF doesn't really have much in the way of those industries.
I've read good things about Berlin, but by-and-large investors in Europe are far more cautious and more heavily taxed than the US.
> immature state of the US agency
> recruitment market compared to Europe
I'm a Russian living in London and always mention Russian recruitment industry as an example of smoothly-functioning market with most inefficiencies resolved in a way you would expect in a modern economy. In short, agencies in Russia only deal with hiring top-level execs, equivalent to someone on £200K+ salary in the UK (i.e. less than 0.1% of the country's labour force).
The rest is handled directly by companies directly through two of the largest websites -- job.ru and hh.ru (it actually feels weird to describe this on a forum that talks about disrupting something all the time!) The traffic on those websites is enormous, the number of applicants for any vacancy is much higher than any recruiter I used in the UK ever brought in.
It costs the company a fraction of recruiter costs; it shows the applicant the company name immediately so he or she can do their own due diligence without listening to things like "top-tier investment bank" that you and your colleagues have invented. It's better for everyone. Compared to that, UK recruitment looks like car boot sale compared to eBay.
Best companies like Yandex normally also advertise on their websites with few questions dropped into a vacancy to filter out non-qualified people (see, for example http://company.yandex.com/general_info/jobs/develop_cpp.xml).
Now, please, tell me about the added value recruiters in the UK bring anyone.
1) recruiter effectively steals a percent of your starting salary;
2) company is so cheap they cannot afford a recruiter;
3) big tech companies are paying extremely low in comparison to the primary sector - finance (and the most interesting stuff will still be developed in USA).
re 3 is mainly related to the poor social standing of STEM employees in the UK
> 1) recruiter effectively steals a percent of your
> starting salary
I have only lengthy experience to back this up, rather than anything I can link you to, but:
* For a contractor, companies budget a maximum amount to spend per day. How that gets cut between the recruiter and the contractor, they don't care about. If you're going through a recruiter, you could be charging the employing company exactly what they're being charged by the recruiter, and pocketing £70-£120/day difference.
* For permanent staff, it's accounted very differently. Companies large enough to have people employed in an HR capacity will treat recruiter fees and salary very differently. Salaries will normally have an upper-band, the salaries of other employees and future employees will need to be taken in to account, the precedent set will need to be taken in to account, generally payroll and hiring costs are accounted for differently, and so on. A three-person company will see it all as cost, but a larger company will treat the recruiter fee and the salary as completely different. Additionally, it's in the recruiter's interest to get you the best salary they can, as they take a percentage from the employer. They will know the limit the employer will pay, and will try and get you the best salary they think the employer will go for; generally they are far better negotiators than developers!
It looks Russian companies do more in house recruiting and employ FTE to do this where as in the developed world this is out sourced to a large extent it doesn't make sense to have full time staff recruiting unless you are the size of Ford or Google
> please, tell me about the added value recruiters in the UK bring anyone
Firstly, the last 5 candidates I've placed were not looking for jobs. Each was in a role where they were slightly unhappy, but inertia was keeping them put. Having chatted to them, I was able to find them roles they enjoyed more, with a significant pay rise. You could see this as introducing liquidity to the market. A job-board would not have found them.
Secondly, of the last job ads I've posted to job-boards, fully 50% of the applicants came from a single non-European company, and the applicants had no right to work in the UK. Of the remaining 50%, almost all were totally unqualified by CV alone. Of the handful who were qualified, a very quick technical chat on the phone weeded out approximately half again. Then the ones whose salary expectations were out of line with the hiring company needed filtering out too. When you're in a position to actually hire people, you'll get a feel for how time-consuming it can be to hire a single role. When you have 5 to fill, it can become a full-time job.
Thirdly, as an immigrant to the UK (welcome!), you'll be well-acquainted with the foibles of the UK immigration system. Candidate A's wife is on a Tier II inter-company transfer visa, and he wants to get a role; does the hiring company need to be on the UK BA list or can you just hire him? What if it's a subsidiary company of a larger company that is on the UK BA list, but it doesn't have its own registration? Is the candidate worth pursuing? What if the candidate applied for ILR four weeks ago? Is it safe to make him a job offer?
Fourthly, as you point out, the UK job-board market is not centralised, and it's not a monopoly. Where should you post a Python job? LinkedIn? Monster? CV Library? Twitter? Reed? Jobsite? Facebook? Jobserve? Any number of the scammy Python-specific job-boards? As a direct employer, your media cost is £200 a pop. Filling out each form take half an hour. Each of these systems has a different way to get the candidates and format them. As a recruiter, however, I'm using a multi-poster, and I'm paying ~£10 for the media cost, and I have the time and motivation to post the job to a wide-range of channels, but I also have a pretty good feel for where my return on investment will be.
Fifthly: What's a reasonable salary for a Ruby developer in London? What if you need them hired in the next four weeks, urgently? What about a contractor? You've probably got a gut feeling, but look, all the job ads show a range, and not an actual salary. How much would you pay for a superstar? How much for a junior? Are you paying more or less than the competition? Two types of people have access to salary data across the industry: HR people, who aren't allowed to share it, and recruiters, who share it as part of their value-add.
There are sixthly, seventhly, eighthly and ninthly here, but who reads long comments anyway?
> Russian recruitment industry as an example of smoothly-
> functioning market ... through two of the largest
> websites
> that people like you (yes, you) will lose their jobs and
> suffer
> who reads long comments anyway?
> the Russian way is better because there is a duopoly
> in the job-board market, rather than meaningful competition
> Fourthly, ... Where should you post a Python job?
> Thirdly, ... Is it safe to make him a job offer?
> Secondly, ... you'll get a feel for how time-consuming it can be to hire
> a single role. When you have 5 to fill, it can become a full-time job.
Would you argue that a person sitting next to you, being able to talk to hiring person (me) in case of any doubt, speaking on behalf of your company (emailing candidates from @company-mail.com, not @recruiter-company-mail.com) and costing fraction of recruiter cost is better than a recruiter from an outside firm?
> Firstly, ... You could see this as introducing
> liquidity to the market.
I understand this conversation is a bit meaningless. It's as if someone would talk to me saying that programming profession is outdated and needs to be replaced with something else. I would refuse to believe that, of course.
> I understand this conversation is a bit meaningless.
> It's as if someone would talk to me saying that
> programming profession is outdated and needs to be
> replaced with something else. I would refuse to believe
> that, of course.
You are talking to me as if I think the agency model is perfect, where the originating comment specifically stated I thought it needed to be fixed.
You are talking to me as if I am primarily a recruiter, where the originating comment mentions that I am in fact primarily a tech guy. I have been programming commercially for longer than you, and I have been a CTO at a company two orders of magnitude larger than yours, and I have had to fill headcount larger than your entire company - a task that would have been impossible without recruiters.
You are talking to me as if I see the world through an agency recruitment role, where the originating comment mentions that I have a specialisation in recruitment software. As it happens, many of the world's largest companies (and many of its smaller ones) used the software we built at my last role with their in-house recruitment teams - I was responsible for a software system through which 70% of all job-ads posted in the UK passed. My understanding of recruitment comes through working with recruitment teams from 2 person companies to gigantic multinationals, from personal experience as a hiring manager, from countless times as a technical candidate, and - of course - from the time I spend working as an agent.
A startup's office space is just one way that venture capital is getting funneled directly into a landlord's pocket. And office rents are rising like crazy, while pushing lots of other types of businesses out (particularly in SOMA).
If a landlord thinks they can come out ahead by evicting a long time tenant and refurbishing a mechanic's garage into office space, then you're clearly paying a much higher rent than that space used to command.
Salaries are still a larger overall expense for just about every startup, but consider how much of that employee salary goes straight to rent. 30% in pre-tax dollars is completely normal at this point, and I expect for some people it's even more. Ownership is out of the question for anyone who hasn't had a stock windfall at a previous company (or who lived here 20 years ago). When a 1 bedroom in a shitty neighborhood is over $3000/mo, there's a minimum to the salary you can accept -- even doing the normal salary-for-equity trade with a startup. Just to hit 30% of pre-tax dollars, you need to make $120,000/yr for that 1 bedroom.
There's no reason for landlords to stop until they've milked the tech economy dry. In most cases they're entirely within their rights to charge whatever the market will bear, and there is currently no shortage of people willing to pay more for less.
In the long term this will hurt San Francisco and the tech industry. The money won't last forever, and eventually the city will be so expensive that the benefits of locating here become irrelevant. SF landlords can't levy a tax on the innovation of the entire tech industry forever.
Even at San Francisco's current relatively frenzied building pace, there will only be another 4000 units or so this year. That will not make even a slight dent in rent for a city of 700,000+ people.
The other problem with building more units is that a lot of the available space sucks. There are whole huge swaths of the city that may as well be on the moon for how accessible they are to locations with offices (SOMA, Financial District, etc.). You can build condos in Candlestick Point, but how are people going to get to work?
The infrastructure of public and private transit in the city (and the larger region) is woefully inadequate for even half the number of people that currently live here, let alone if the population significantly increases. And these kinds of projects work on 10-20 year timetables. So again, it's way too late for this to be fixed in this economic cycle. Maybe next time.
So where I live (Pune, India), we are also having an huge surge in population due to people moving in for new jobs. We don't really have a housing crunch.
Unlike SF, you are allowed to simply build new housing societies as needed (more or less). Here is NIBM, one of our outskirts, about the same distance from Koregaon Park as Candlestick Point is from SoMa: https://pbs.twimg.com/media/B8cMeVTIMAEMyzk.jpg:large
Strangely, all these terrifying issues like people being unable to get to work simply don't happen. They take two wheelers, auto rickshaws, or 6 seaters. Housing societies sometimes provide shuttle buses.
Bus service? Public not the Google Buses. Will always help to encourage movement out of a crowded city centre and (relatively) low capital requirement. Do some deals on season tickets/passes/carnets
Or you could take the single bus line, which cuts your walk down to half a mile or so. You'll have to transfer to another bus that goes on the highway to downtown (so it will be absolutely awful during rush hour). Expected transit time is again 1 hour.
Muni's on-time rate is 60%, so these are pretty optimistic numbers. Bus's don't keep to their schedule at all -- if they get to a stop early, they take off. The supposed 15 minute interval is laughable given the city's traffic patterns, it's just as likely for three buses to arrive simultaneously because they were all stuck in traffic for an hour.
The city's infrastructure is simply not built for outlying neighborhoods to be viable living destinations for people who commute into the city for work. The highways are clogged and poorly designed (and don't connect through the city). The surface road stoplight system is one of the worst synchronized I have seen in the US. The bus and subway system is poorly run and starved for money on basic necessities. The subway system itself is terribly designed: all the lines run through the same underground portion, on the same rails, so the slightest hiccup blocks up every line.
It's all solvable if you throw a lot of money and political will at it, but neither seems to be there. At least all of the buses are supposed to be replaced within a few years: they SORELY need it.
The rent I charge on my (nice, two-story, good neighborhood) property that is not located in the Bay Area is about 1% of the price I bought it at. In the part of the bay area I live, rents are about 0.3%-0.4% of the typical sale price for homes that are much smaller than my rental. I could buy 6 of my rental properties for the cost of one such home, have similar net appreciation in real estate value and be earning more rent.
I don't have hard data, but my intuition is that this is not the case. The raw price of storage / compute has only declined a little bit since 2009, but the number and variety of development tasks you can outsource to (broadly speaking) "cloud services" has dramatically increased.
In 2009, Heroku was still a relatively primitive platform with only Ruby support - today, it can save you the cost of a sysadmin or two. Mailgun didn't exist and SendGrid was just coming out of TechStars - if you wanted email in your app, you needed a dev to build it. Twilio couldn't do text messages yet, and we didn't have Stripe for payments or Mixpanel for analytics - both were glimmers in their founders' eyes.
The sheer amount of grunt work you need to do to get you app up and running has plunged since 2009 - now you can buy these core services for 1-2% of what you had to pay to develop them. These wins are most pronounced for seed- and A-stage companies, but they substantially cut your development and operations costs throughout your scaling lifecycle.
I'm willing to bet that if you plot the median number of employees it takes to reach a seed, A, B, or even C round, you'd find that the leverage gained from this ecosystem of services more than makes up for increasing personnel costs. Of course, startups everywhere benefit from these cost reductions - this has nothing to do with San Francisco - but it certainly makes the increase in salaries and real estate expenses an easier pill to swallow.
the author has a well laid out case, researched and backed. you come in with feelings.
This guy also has these ridiculously generic datasets and the geekerati goes nuts because he's a vc.
The market has driven up salaries ridiculously high.
[1] at a 40% tax rate, 3.5k/0.6 * 12 = 70k
Here's an open secret: companies willing to allow remote working are able to hire in a -fraction- of the time it takes to hire onsite, and often substantially cheaper.
To many people want to have their cake and eat it
Wages aren't ridiculously high in SV--they're ridiculously low, just like share of equity is.
Has it really? My dad owned his first home freehold after working for 5 years. I'm almost there after ~15 years. And I've had help from my wife.
Real estate has driven up salaries ridiculously high.
I'm not going to uproot my entire family for an equivalent lifestyle so realistically I wouldn't move for less than $240k/year.
I'm not saying I'm worth that kind of money; I'm just saying it's not worthwhile for me to work in the valley for less.
If you are describing an increase as a contributory factor, then you are obviously missing the time period that goes along with that increase you are talking about, so readers can go and look up the historical real estate pricing data.
At 6-figure high-competency-developer total compensation package (and salary) levels in the US, you are looking at the 28% marginal tax bracket for the employee. High-wage employees like these developers will typically pencil out to 100-150% fully burdened to the company. For top 20 metro areas in the US, you will tend to see most such employees spend closer to 45-50% of gross pre-tax income on shelter costs (counting PITI plus maintenance), rather than the 33% that is the historical prudent personal financial planning guideline (and those guidelines should be taken with a grain of salt for reasons I won't get into here).
With my CFO hat on, I look at the ratio of third-highest decile (with 50% pinned to the median) of PITI plus maintenance in the parts of the city I anticipate most such employees would prefer to live, to the total burdened cost per developer, to establish just how much the company is paying for shelter costs of its employees. It's a surprisingly large percentage for the top 10 metro areas, though everyone's specific numbers will be different based upon neighborhoods they choose, time period, etc. The ratio of course goes up as you travel down the skill (and payroll cost) ladder.
Even with highly-compensated and -skilled employees towards the higher end of the skill ladder, I was coming up with ratios closer to 10% than 5%. I've seen companies move mountains to capture far, far smaller payroll savings percentages.
What is much more difficult to quantify however, are the soft benefits of ready access to the large, concentrated ecosystem that grew up around the industry in that geographic area. Clearly, many companies today consider those soft benefits worth the incurred hard total compensation package costs. It's a calculation that you and your advisors should make and continually revisit once in awhile for your venture's own individual circumstances.
For some smaller outfits who already have an assembled team and are in their early development stage, it is arguable that those soft benefits do not outweigh the immediate cash flow benefits of avoiding the high cost of living areas.
Thus, the vast majority of the increase in costs that it takes to run a startup is in employee salaries, rather than office space.
Especially for a startup in like a stealth stage where the initial talent pool is already selected, and you don't really need all the touted soft benefits of the SV talent ecosystem when you are head-down in secretive development, it's advantageous to stay out of the high living expense metro areas. Living in flyover country during that period with the fully-communicated intent to the team that you will move to SV/high-COL-metro-area once you reach a specific level of financial success with specific commensurate salary adjustments for them, yields 15%+ salary cost savings from residential shelter cost differentials alone (plus other cost of living adjustments on top of that), and lets your initial employees plan ahead life events.
Now that startups are increasingly tapping the Millenial generation cohort for their talent, who statistically are already more indebted at their stage of life than previous generations, starting up in a very low cost of living area can be a bigger selling point for recruiting.
All else being equal, a Silicon Valley/San Francisco based startup is able to achieve a much higher valuation than an equivalent startup outside San Francisco. However, this Silicon Valley/San Francisco based startup has to deal with higher real estate and salary costs.
> All else being equal, a Silicon Valley/San Francisco
> based startup is able to achieve a much higher
> valuation than an equivalent startup outside San
> Francisco.
flying in for a few 9-5 days a month is not the same.