Ask YC: A different kind of scaling question...why don't web dev shops scale?

1. Each project is generally a one-off, custom solution, rather than a reusable technology. You will spend a lot of time building features specific to one customer, in exact way that customer demands, rather than building generally applicable solutions. You can't say "No, that's a feature only 1% of the public will use!" because they're paying you to implement it, no matter how ridiculous the request is.

2. It is hard to win projects. There is a long, complex, competitive process that requires account people to find leads, make presentations, and respond to RFPs. This is a fulltime job for several people.

3. Companies want to work with firms of similar size. You'll get asked about the number of employees, yearly revenue, and other details, and if they don't match up with what the client is looking for, they will not work with you.

4. Because each solution is customized, you can't specialize in one skill area. Let's say you are the expert in developing Rails apps. Most web dev projects are big integrations where your Rails expertise is just one piece of the puzzle. You may have to integrate with some legacy custom system, or 3rd party software for e-commerce, CRM, email marketing, accounting, etc. You will almost always have to scramble to learn something new or find a consultant who does. This is very nonscalable as you are always forced to pick up non-core skills or spend your time as a virtual HR department and find people that do.

5. The custom nature of the work results in a lack of focus. You are spending all your time serving your clients' needs, being dragged along by their requests, instead of implementing a focused vision to a specific problem.

6. You may think that you can turn your consulting business into a product business. You can, if it is small enough, say, 2-4 people. Once you get addicted to the upfront revenue of projects, it becomes nearly impossible to make the investment to build a product, as this both costs money and detracts from your client work. You will not put in the 6-12 months of no-revenue work needed to create a new product from scratch.

7. Web dev shops don't just do development. The biggest are basically full-service marketing agencies.. You need to have a full complement of professionals, not just developers. These include web strategiests, information architects, designers, HTML coders, project managers, account managers, salespeople, usability experts, branding experts, etc.

8. Your recurring revenue comes from running operations and maintenance on existing websites. This can be a good source of revenue but it is exceptionally boring work that no good developer wants to work on.

9. The true leaders of a service biz like this are sales and account managers. To win projects you have to spend your time building relationships, not technology. The key skills in a consulting business are soft skills such as sales, hiring, marketing, and project management. Tech people are a cost center in a consulting business, not a revenue generator.

10. In a consulting business, you almost always charge by the hour. Even huge, multi-million dollar projects are broken down into line items by man-hours, and savvy clients will insist on seeing the breakdown to make sure you're not "making too much". But hours != value. In a product business, you can charge whatever the market will bear regardless of how long something took you to do, but in a service business, if something takes you an hour to do and you charge $200 an hour, then you're getting only $200, even if it might generate $10,000 in value. This results in a disincentive for people to work efficiently, exercise judgement as to the use of their time, and engage in out-of-the-box problem solving.

11. Unlike a startup, money cannot scale a consulting firm. With a startup, more money can be used to hire more developers to build your product, which will hopefully gain traction in the marketplace, resulting in growth that is faster than linear. In a consulting firm, the most you can do with extra money is hire additional sales people. They then need to scare up some projects, and repeat this indefinitely. This is a slow, linear growth path.

12. In consulting, projects are self-contained You can't build A and then build B on top of A that will take you to the next level of profitability, then build C on top of B You're just building A, followed by A, and A, A, A, indefinitely. The most that you can take from project to project is general domain expertise, contacts, and reputation.

They don't scale because they're services businesses, and services businesses are harder to scale to large number of clients, since you're selling hours rather than bytes, and hours can't be duplicated for free infinitely like bytes can.

Big 4 accounting? H&R Block? I see your point, but there are companies who have made it work. And even for product-based businesses, only an extremely small percentage sell goods that can be duplicated for free infinitely. Amazon is doing okay.

Please note the use of the word "harder", rather than "impossible".

It's possible to build a company the size of Accenture or Infosys... but it's damn hard and requires a LOT of skill, talent and luck for a very long period.

Noted. And I do agree with your original point; look around your neighborhood at 10 different small local businesses. Dry-cleaning, carpet-cleaning, etc. Services. Products are sold by Wal-Mart and Home Depot and Gap. There's no question that, for whatever reason, service is hard to scale.

One reason could be the revenue model, its hard to expand without recurring revenue streams. Many places, like Robert-Half, don't have coders on staff, so there's no capital expenditures when the market is slow. A large firm would still have to pay people. Now, an example would be IBM, they're a huge development shop if you think about it, but they have many contacts and work on long projects that consistently bring in revenue. Its hard to get to that level, especially when your clients are diverse and hire you for one-off projects.

In your example about law firms, I think its inaccurate to say there are no large nationwide firms. Its more accurate to say there are few large nationwide plaintiff's firms (although lawyers have other restrictions we won't get into). Plaintiff firm's clients don't usually come back for a second round, which makes it hard to employ many people who might just sit around doing nothing. Business law firms, on the other hand, do have large nationwide presences, because their work is often on retainer or consistent.

I think you're right that the market is against the structure of a large firm. Its just much more economically useful to be small or be part of another company that also does other things, because the revenue stream is inconsistent and project based.

For some reason, I started thinking about the barber shop/salon type model. Where you pay X to rent a booth to do you work in, but has shared resources. This idea quickly starts to look like a tech incubator, but different.

Would it be a independant Hacker Incubator? Supplying physical space, mail boxes, phone lines, shared secretary pool, printers, and hopefully some amount of mentorship; but, centered around the independent spirit of the developers.

(Sorry, I think I rambled away from the topic.)

No, absolutely, that would be stage 1 of the let-us-share-an-office plan. And I think that would be awesome (http://news.ycombinator.com/item?id=232235). This post was more "how far could we take it?", but the model you brought up would definitely work.

And, I might, add, strippers work on this model. They pay a floor fee to the club and keep the tips. (Don't ask me how I know.) If it's good enough for strippers, by golly, it's good enough for me! (This also brings to mind the concept of specialization, although I'm not sure if I want to elaborate.)

In my many years consulting, I've seen two types of clients:

1. Corporate - pays the market rate but will not assume the legal risk of taking on individual developers; will only work with another corporation or agency.

2. Mom & pop - will work with individual developers but only on a fixed bid; expectations are about a quarter of realistic budget.

So my choices as a freelancer are: work 8-hour days and get paid my full rate through a recruiter (yes, they skim off the top but who cares, my rates are what they are), or work long days + weekends on freelance projects and get paid pennies on the dollar (both in terms of lower hourly rates and never getting paid for all the hours you work).

Add to that the tax benefit from maintaining a home office, and you realize that there is simply no reason not to work through an agency/recruiter. They take care of finding the good (i.e. paying) clients, billing, etc. A developer "coop" is simply neither here nor there... at least from my experience here in New York.

Don't forget that companies like US Web / CKS, Razorfish, and Organic grew quite large during the dot-com boom, when there was plenty of Web dev and design work to go around. (Sure, these companies had major downfalls, but I'd argue that that wasn't due to a failure to scale up, but rather management problems and a sudden dearth of clients due to the bust.)

Likewise, if you consider that outsourcing firms like Infosys and Satyam are likely doing some Web development work in addition to more traditional "IT" work, those firms have scaled quite well. Infosys has over a billion dollars in revenue each year.

Maybe the moral is that shops that only focus on Web applications will have a harder time finding new and larger clients than firms that can provide services that go deeper than the Web interface -- into CRM, ERP, and so on. If you don't like to do that sort of work, then you stay small and run a boutique shop.

they do scale.

but in order to scale, the dev shop has to be willing to take on different sized projects. and in doing this, they evolve from a dev shop to a consulting firm. this is because, to be able to ensure everyone has work, turning down small-scale one-person types of jobs becomes more and more difficult.

for example, the shop would be much more willing to send the java guy in on contract to fix issues or do some short-term work on an existing system if the alternative is just having him sit on his hands. if he sits on his hands for weeks while he waits for work, he's either not getting paid, or the company isn't getting paid for having him work for them, which is a net loss. and he's also probably surfing dice while he waits.

IBM, Logica CMG and Accenture would be examples of big dev shops. Not solely web.

I see your point though. Maybe the niche is too small and generally too cyclical.

Viget is pretty large, in my eyes.

http://www.viget.com/about/team/

Well, we've got 32 employees right now, but I wouldn't call us "large," really. Large to me is ThoughtWorks, or Sapient.

Your staff look like an Abercrombie & Fitch store...

That's interesting, how is the smoothstart project coming along?

p.s. I'm creating it's competitor.