Scaling In Lower Cost Locations
This is a topic I’ve written about a bunch over the years. I feel like it is becoming more urgent every day.
Last week I heard some shocking numbers about salary levels for certain kinds of engineers in the bay area. I checked them out with a few of our bay area portfolio companies and they were more or less corroborated.
The tight technical labor markets in the bay area, NYC, and a number of other regions in the US are making it hard to scale software businesses without burning massive amounts of cash.
At the same time, we see a growing number of our portfolio companies succeeding with scaling engineering/technical teams in secondary labor markets in the US, as well as going outside of the US to build engineering locations.
I feel that the ability to spin up and then successfully operate remote engineering locations is a skill that technology companies need to develop earlier in their development than used to be the case.
It seems to me that once you get to 100-200 people (or 50+ engineers), you should be thinking about this. The most important thing is not where you put your first remote location. The most important thing is learning how to do this successfully. Because once you can do it in one location, you most likely can do it successfully in multiple locations.
This post explains how Stripe (a USV portfolio company) started with remote engineering hubs in Seattle, Dublin, and Singapore, and then evolved into a structure that supports remote workers anywhere.
The move from a centralized engineering structure to a decentralized one is a process and takes time to get right. And so I think it is best to start building those capabilities long before they become necessary.
In other words, it’s labor arbitrage. Outsourcing IT services companies from India have been making billions of dollars in the past couple of decades just on that value proposition. Most of their top customers come from the largest industrial countries (US, UK, etc) where white-collar jobs have become quite expensive.
Is this quite the same thing that Fred is on about? i would call it ‘outhousing’.
I wouldn’t call it arbitrage. There are a lot of great engineers working at companies out there that don’t have the stomach for startups. Plenty of computer science majors graduate each year from engineering schools in the US, but they are in such high demand from all companies they have choice. I think Fred is right that you need to learn how to do the remote team. I have seen early stage companies start with this concept and just build it into their business practice. Communication is the key
It’s labor arbitrage at the global level. Sorry for being pedantic about it, but I’ve spent a few years in that industry to know what it was about.
I agree this is labor arbitrage at the global level because I am involved in the same business 🙂
Coming from the game world we did this early.I was traveling to and working with remote teams and studios since the beginning of my career.Zoom btw has made this process a lot easier.
In terms of location, is there much activity or potential in Eastern Europe? Prague, for instance?
In my experience yes.A buddy is building a project on the side and his development team is in Romania.
Having done this for the last twenty five plus years one does this for – time to market,- skill augmentation andone benefits on the cost as a by product.Over the years I have found that- the earlier one starts the better- ensuring multiple points of necessary communication are key- having the same rigorous hiring processes independent of the location are important- incorporating the remote teams into mainstream work and going through the resulting challenges will pay dividends laterand finally- the ability to think in the language of the parent location / compnay is an essential ingredient- initial patience will create a strong foundation
Hanover, NH (home of Dartmouth College…alongside White River Junction, VT, Lebanon, NH, Norwich VT)… we are doing everything possible to keep our HQ here for our startup. Not easy, but still hanging in there.
Why not easy?
I’ll try to be more clear:(1) Solid Stuff. In information technology, e.g., computer science, software engineering, software development, Internet communications, project management, system management, analytics, data science, machine learning, artificial intelligence, etc. essentially the only source of solid material is academics, especially the better US universities and associated books, papers, video clips, etc. That’s a fact, Jack. Of what actually IS solid, that’s where it is and essentially the ONLY place it is.(2) Doofus Land. The practical information technologists in Boston, NYC, Silicon Valley, etc. add nothing of much value to (1) and otherwise add ugly graffiti to good or great work.(3) The stuff from (1) is readily available nearly anywhere in the US from a small, rural crossroads to anywhere else. For the rest, the main issue is in just one word, DOCUMENTATION.E.g., computer science professors commonly have little or no experience with the System32 calls so long just key to Microsoft’s operating systems and software. But there is NOTHING at all very technical or advanced about those calls. Instead the main issue is just that one word, DOCUMENTATION. With a good start on (1) and some good documentation, can learn the rest QUICKLY.Commonly in practice the bottleneck is JUST documentation. Getting the information is commonly a combination of mud wrestling and a self-inflicted, unanesthetized root canal procedure with NOTHING fundamentally difficult except really BAD documentation. Bluntly, when hiring an expert and paying the big bucks, you are hiring someone who has some, typically not very good, practical skills with the poorly documented stuff.Soooo, paying those big bucks is DUMB de DUMB DUMB. Instead, borrow from the F. Brooks idea of a “chief programmer team”, that is, some one person who understands that stuff and passes out the info to others. MUCH better still is just to hire such a person as a consultant for a week, as I outlined. ANYONE with anything like a competent background in (1) and some GOOD documentation can learn the rest with blinding speed. Here can start with any well motivated computer science graduate from nearly any college at nearly any rural crossroads and have them fully up to speed on any of several parts of the latest of the rest of the junk-o stuff in a few months or weeks, sooner than can recruit an “expert”. Actually, much less will often be sufficient — can get someone with no college and do well. Often can do well with a bright sixth grader — literally. And that is true for the current hot topics in analytics, machine learning, artificial intelligence, etc.Generally, the techs are NOT very good at technical writing. E.g., nearly all the classes in writing are for Belle Lettre — sick-o. But, in addition, it appears that long ago Bill Gates saw that there could be some big advantages in having poor documentation — the nerd, techies who did get some skills with the Microsoft products could have careers with a barrier to entry and become part of a marketing channel. That theme has remained largely the case in practical computing.The significant stuff is ALL grounded in US academics and changes very slowly. The rest is trivial and can be learned quickly given JUST some good documentation. You really SHOULD have the documentation around anyway, so get the documentation and then hire local computer science grads or less at quite reasonable prices.That Silicon Valley actually knows anything significant beyond what is at hundreds of US colleges is just total BS.E.g., now even a LOT of Google searching won’t say anything about the Gleason bound. It’s a very nice, even important, result. So it remains that the only source I know, and an excellent source, isDonald E. Knuth, The Art of Computer Programming, Volume 3, Sorting and Searching, ISBN 0-201-03803-X, Addison-Wesley, Reading, Massachusetts, 1973.Right, not nearly new.If they don’t know the Gleason bound, then they don’t know very much. And the rest is easy given some decent DOCUMENTATION.Ah, but what about the latest, super hot, far advanced, way out artificial intelligence !At one time I was an expert in that suck-o garbage. My wife wanted to know some of it. She was in mathematical (statistical) sociology, and never got through calculus and had never had a course in computer science. So, I gave her a 15 minute lecture on using an editor and writing Rexx code, and in a week she was up on that and had some nice code. Then I gave her 15 minutes on artificial intelligence, and in two weeks she had some nice code running. Then I explained a little more of the smoking funny stuff the MIT guys had in mind, and in another week she had a really good, first artificial intelligence program.Currently artificial intelligence is a lot of data, some routine techniques based on curve fitting of the logistics curve, next to nothing in meaningful theory, nothing to do with intelligence, even of a housefly, some routine software, and a lot of throwing work against a wall until it appears to stick. No reason to pay big bucks.
Hahaha – I sense some frustration with popular front end tastes nowadaysI’m long remote work. Started doing it years ago. Now with slack and video the distractions of office space are not something I’d pay for the privilege of.
Great post. I’ve been dealing with overseas engineers for the past decade, and currently have my full team in Romania, right outside the technical University of Cluj.I feel they are churning out great talent there. However, I find the market for highly competent engineers shrinking. Go to Elance or other hiring sites, and you’ll find thousands of engineers whose resumes seem to show high skills, and competency. However, do a few interviews and you will find a very different story and most of these guys can’t answer basic engineering questions properly.So you really do need to be careful who you hire from overseas, when looking to take advantage of lower labor rates.I’ve been down the path of using a recommended overseas engineering firm to do our first mobile app. Big mistake. Luckily my team knew right away there were competency issues with basic engineering architecture. The problems we would have encountered if we continued would have been massive. We got rid of them after only a few months, at a small loss, and found a much better mobile firm locally in Romania. If we hadn’t done that change, the mistake could have cost hundreds of thousands for us. Not huge I guess versus the big companies mentioned here, but would have hurt us financially and the time lost would have been tough to recover from as well.For any new overseas hire, make sure your superuser ensures they pass the sniff test first and I would use those first few months as a test to see if they are engineering the right way, in a timely manner, and that they fit in with the overall culture and mission at hand. It’s a always a bonus when you land a competent, reasonable affordable engineer who also possesses some ui/ux design skills. That’s the real challenge
“shocking numbers” – go on then, shock me.
Two separate dynamics I see going on here. Off shore, or remote engineering centers and distributed, de-centralized development teams. The former has been around for years in many technical industries, not just software. The latter seems more new and will be interesting to observe.As an aside, rising wages theoretically drive inflation, which can lead to monetary tightening that restricts the availability of easy money. Perhaps this is an early indicator for the current economic cycle.
Remote/distributed companies will become more and more the norm. The trend is clear. The upside is measurable.
No surprise seeing the revised thinking since it was blogged here 10 years ago: https://avc.com/2010/09/out…I’d say that until 5 years ago, lots of VC’s were condoning these ideas only as a necessary evil.From around 2 years ago it’s been more like, “ok I guess we should consider something that leverages a blended on/offshore, as a hedge against local market insanities”.And now, most investors are saying, “What’s your global decentralized dev plan and how are you executing?!”My group still specializes in helping startups big and small make sense of options– we do this all day long, and I feel like it’s finally becoming really exciting due to factors including that people are getting very used to Zoom meetings, telecommuting, multiple time zones, etc.Here’s our URL (if that’s a “plug”, fine, but would love to hear from people here): https://loremlabs.co
Real estate in the big cities is expensive and at the same time infrastructure is buckling under the population increases. Remote work is a way to try and blunt the impact of this but also it is worth trying to encourage government to do better and advocate for policies that help scale transit and build housing.
Great developers, product managers and programmers can be found throughout the southeast. Look at Chattanooga, Knoxville, Orlando, Charleston, Charlotte….
May I post my twitter thread here?https://twitter.com/CindyBi…
We get you. Remote teams are extremely valuable but staying in compliance with payroll, benefits, taxes and local law is an ongoing challenge. Here’s a resource you can use to help you meet specific regulations. https://papayaglobal.com/co…
I’ve had a few searches where the client was open to hiring remotely and the talent choices improved significantly. While at executive levels and certain job categories, regional differences do not factor as much into compensation, lifestyle issues do factor in.I wonder how opportunity costs factor into decisions to having all team members onsite, whether limiting hiring to the local market or paying for relocation (which is limited to those willing to relocate).Even if not open to a remote team overall, there are at least some roles that could be remote.
As a start-up, unless you have specific ties to an offshore region (like a resident co-founder), the costs may outweigh the benefits. High turn-over, language barriers, lack of start-up cultural alignment, lack of passion, timezone offset, etc. IMHO, near-shore/on-shore is a much more effective model.There’s a definite trend underway, with start-ups and software companies in the tier 1 tech hubs committing to growth in tier2 tech hubs to combat the costly competition for top talent in SF, NYC, Seattle, and Boston. Consider the model that we at EmployeeChannel have employed. We expanded from SF to downtown Rochester, NY, foreseeing the high tech, entrepreneurial resurgence currently underway.For us, it didn’t make sense to pay a premium to hire and relocate top talent from universities like RIT, when we can build a high-performing team locally, and provide that team the same start-up challenge but a better economic future? Rochester was ranked #1 in “Jump-Starting America” by MIT professors Jonathan Gruber and Simon Johnson. If you are looking for top talent with an entrepreneurial spirit, you owe it to yourself to check ROC out.Why our Silicon Valley company chose Rochester https://rochesterbeacon.com… with Rochester Beacon
We are seeing this trend play out across multiple Upstate NY cities at a lot earlier stages than 100-200 people. Entrepreneurs from both coasts (often alumni and expats) are returning to build out their teams in “saner” more livable environments. Upstate has 100+ colleges to recruit from, commutes measured in minutes vs hours, and housing that costs a fraction of what one has to pay in the big tech hubs. https://uvc.org/coastal-tec…
Definitely seeing companies trending towards lower cost but equally talented regions – like Upstate, NY. In Rochester, NY recently saw Live Tiles, Employee Channel, Catalant, and others – including a large Silicon Valley VC, open offices here for the talent and opportunities. Grow a company, and have a life – good combo. Look at “Jump Starting America” for all the reasons Rochester ranked #1.
If you only have the mindset of getting cheaper labor, you won’t really find talented people.Talented people outside the US can also negotiate good salaries, probably not always in the 200k to 300k range, but certainly much more than the 50k – 70k range you are looking to pay them.The greatest advantage by far is that you broaden cultural and technical knowledge from all over the world.
Thank you for discussing remote workers inside of Stripe’s structure. I think more companies will adopt remote-friendly policies.
Or… you could go “Distributed First” and see how that works!
I live in a Light Industrial Area in Outback Australia. It’s highly cost-effective, very grounding and it’s exactly where I need to be to build the world’s best workplace safety tech.
When is Trump getting removed from office like you said?
When boomers die US companies will go 100% offshore. Value of tools like Upwork is amazing.
Why on Earth would a SaaS company need 50+ engineers until very late in the private phase?Some maybe, if they need to integrate to a LOT of platforms. Even then it’s iffy.Back in the Bad Old Days, when we had none of the slick dev tools available today, I took over running research and development for all our products at Borland. These products ranged from software development tools (Delphi, etc.) to Quattro Pro, a spreadsheet, to databases like Paradox and dBase.When I took over, I was tasked with cutting expenses. I surveyed all the teams sizes, including contractors, vs complexity / size of code. One team stood out: Quattro Pro was getting the job done with about half as many as the rest of the teams. 10 developers wrote its large code base (we had to build all the features Excel had in the day and then some for Windows).I downsized every team to match Quattro Pro’s team size. Despite widespread predictions to the contrary, not a single schedule slipped at all. In fact, I would submit the bigger teams would’ve missed their targets had they not downsized.In the end, the limiting factor for developer productivity is still communication and you can’t get more than about 10 to communicate effectively. The Mythical Man-Month hasn’t really ever changed.That’s why I can’t imagine 50 developers being a good idea, especially outsourced and offshored which only makes communication worse.BTW, many of the principles that made that original Quattro Pro team so successful are the foundations of what we call Agile and Scrum programming today. James Coplien from Bell Labs found we had the highest productivity of any team he had measured using these techniques:https://www.scrumalliance.o…BTW, once you realize you can only use 10 people effectively, you work a lot harder to make sure they are the best 10. And you also quickly realize the corollary:Good software is not that expensive to develop. It just doesn’t cost that much to hire so few people. Bad software, OTOH, is extremely expensive when you consider it can sink your boat totally.
Good to hear.Neither of those places are doing a very good job of letting the world know though.They can fix this if they want to of course.
It’s all upside.Don’t forget the obvious which is start to bring some conferences to town. Smaller ones with high visibility.For example bringing RadicalXchange to Detroit last year was just smart. Having Vitalik Buterin keynoting it didn’t go unnoticed.
Thanks Charlie.I’m not feeling shock.Who’s number six?
If I was a place 2.5 hours from the NY market I would focus on the food and light manufacturing startups.Labor, life style, taxes, and more are so much better.Know someone who is involved in PA based co packing/HPP project of some scale that is serving the Tri-State market.See a lot of it happening in the Kingston area though they still need to deal with the craziness of NY State.
Sure, let’s talk.