Offshoring roulette: lessons from outsourcing (2016)

To summarize, the results were literally comically bad. All kinds of hand massaged spacing using nbsp's and weird off by one pixel issues and such—things that to do wrong as they did must have taken an order of magnitude longer than it would have to do them correctly.

So, I went through several rounds of pointing out really obvious issues and asking for revisions. They would apologize profusely and say essentially that the person who had been working on it previously had been a junior developer, but they'll put someone much more senior on it now. Never really made a difference.

But here's the one that really blew my mind. The site color scheme at the time was greens and whites. For some reason though they decided that the heading of the modal box should have a thick orange underline. How did they create this ~4px underline? By embedding an enormous image of a sunset and cropping it to a 4px orange line.

No exaggeration. This was like a 2560x1440px image of a sunset, which they'd cropped to a sliver in the top corner to show an orange line.

I have to assume the developer(s) got sick of my nitpicking and decided to do something so ridiculous I'd just go away. It worked.

Having cost as the reason for going with remote devs is a very bad idea in my opinion, and it should only be seen as an added benefit. You don't want to hire cheap devs because they usually suck, and great developers will cost you quite a bit regardless of where they live.

While it may be difficult to compete with the huge corporations in places like SF, NYC, Seattle, where these competing corporations offer huge salaries, great benefits and have brand recognition, there's quite a lot of value to be had with setting up shop somewhere remote, where there's lower competition and where you can offer above average salaries to attract top talent in the area. Ultimately, every company says that they only hire the best, but unless you're paying top dollar in your area, you're most likely not hiring only the best.

I believe that most good developers will want to be part of a stable company that is actually building amazing things, instead of being part of an outsourcing company where they get passed around from project to project, and their resume will perpetually read "Software Engineer at Outsourcing Company Ltd." - most would trade that in a heartbeat for one of the FAANG companies.

If you want to have the best kind of success with remote devs, don't outsource, invest a bit more and set up an actual office and hire the people directly, people that want to have a career with your company, and not a contract for an unknown period of time, which they likely won't be able to put on their resume.

The biggest problem however, was that no matter how many times I asked or provided documentation/feedback, they would ignore established coding standards and established processes and just do things their way, and when I would ask about this, I would just get shrugs.

Finally, we got saddled with a "Project Manager", which I put in quotes as their contribution to the process was giving daily excuses as to not turn up to our daily progress calls, and the ones he did turn up to, he would visibly fall asleep. When I eventually pulled him up on this his response was that "he wasn't directly paid for but an included part of the cost so you can't complain about what you get."

Admittedly all of the above is rather negative, but I think this highlights the "get what you pay for" aspect of offshore outsourcing.

Definitely agree with the can-do attitude of China. Sometimes I'm pleasantly surprised, other times a vendor thinks they are doing us good and screw us over by completing work before we have agreed to it. This is mostly a temporary issue until you build a relationship with the vendor (in China, relationship is everything) and know each other's expectations.

While I've never worked with an Indian vendor directly, it is interesting to hear about how people there have very narrow knowledge bases. This is something I've noticed with people I've interviewed that are recent immigrants, they know their area, but when asked about the rest of what we would be asking them to do they have zero clue. This is also (lesser so) a case with recent Chinese immigrants. Not great when we are essentially hiring a mechanical equivalent of a full stack engineer.

But if they would pay twice that (still at most half their normal cost) using a smaller higher quality company they could get a much better job done.

It requires more up front work to identify, hire, and retain the remote employees, but you need far fewer developers when you can appeal to the best developers who know what they're doing.

This doesn't work if your company has any attachment to headcount, though. It can be a tough sell to hire 3 remote employees for the price of a cheap team of 10 from an outsourcing firm if your company only looks at the numbers.

I had written some domain models for dealing with users and I needed equivalent models for dealing with notifications. So I asked them to take the work I'd done and use it as a template for the new stuff. I got back code with the method and variable names search and replaced, pretty much exactly like what I expected. What they _hadn't_ bother to modify was the SQL that actually operated on the data (yes, I had to write it out like a cave man). So I had a bunch of notifications domain objects that operated on the USER table.

That experiment terminated early.

What can we, the highly paid locals, can do to justify our high salaries at that hypothetical future point?

(playing starcraft taught me a tremendous amount of respect for East Asian creativity and work ethic)

There are certainly ways to work around most of the problems the author outlines. They are definitely real problems, same as spending 5x for onshore resources. Ideally you should build onshore teams, if you can spend the money. But if you can not, then do not get discouraged by these problems. Employee churns are not that difficult to mitigate, just matter contracting properly with the employees and providing enough consideration in the contracts.

1) 4 hours of time overlap rule - you need to have overlap of 4 hours between teams

2) Requirements need to be much tighter with outsourcing. Often you need a full BA whereas with some devs you can work without a formal spec but some wireframes etc

In our case we worked with two outsourcing firms in Russia, EPAM and Distillery. Our relationship worked very well but we had to approach it correctly.

In our case we treated the outsourcing firms mostly as recruiting firms. They pre-screened candidates and then we interviewed them just as if we were interviewing and on-site employee.

Now, people are people and are imperfect, But the reality was that from both EPAM and Distillery, The candidates that they brought to us were always the highest quality. Something about how they work internally makes it a lot easier to hire through them instead of the normal American way of hiring a software developer.

One more thing: I was part of the team that chose EPAM over some of the other contracting firms. we basically interviewed one of their engineers as if he was going to be a new hire, and we did the same for other firms. The reason why our relationship was so successful was because we we're methodical and carefully chose who we used for outsourcing.

IMO, most of the complaints about outsourcing come from people who didn't screen their contractors very well.

The whole idea of sending a big pile of requirements and then getting something useful back simply doesn’t work. You need direct communication without middlemen.

Like with Avni, if she really had a baby and went to have baby, the relationships were nowhere near for anyone to know it. Which again, to the extend people stay in positions due to liking people there, that sort of "keep me there" relationships were not present.

I suspect this also because we had projects where more western company outsourced their project on us - coders here, analysts, management and decision makers there. The more skilled developer, the sooner they were leaving or were very particular about what exactly they will work on. Even those who stayed (because some perks and good relationships in our local company) dont like the projects they work on. It is more about frustrations and politics then about being proud what you do.

I can only imagine that it would be even worst in outsource only company that aims for cheap and does not have perks. And it is even harder to find good managers willing to work in such setup, they tend to avoid it too.

That's very strange - I would have assumed that the code quality of the Australian outfit would be higher

I have never seen a case where they actually add value; it would always be much easier if we could just get directly connected with the end-customer.

What is something that we can do that would say “these people are worth it, let’s go with them even though they are more expensive”?

Pretty much as described in the lessons.

One important point, setting up the cooperation is critical and you can't hope to get any positive results without considerable effort into setting up and maintaining the cooperation and close vigilance over what is happening.

There is tendency for the quality to fall the moment the team figures out you are not paying attention. Basically, anything goes unless you are constantly showing you are caring and monitoring.

When I was CTO of Andela one of my main claims to fame is pioneering the "month one" curriculum, based on principles of Applied Improv -- taking comedic improvisation education and setting it in a business context. Nigeria is definitely one of those cultures where junior people are culturally ingrained in them not to disagree with their superiors. Extremely important to break them of that practice if they're going to succeed as software consultants.

The other interesting thing I got to see was a shakedown of our company after we poached talent from a connected local company. We had the option (which we took) to pay several years worth of the average per capita income to local government henchmen to be allowed to stay in business.