It’s important to note: I’m not a business manager. I’m a developer and also a partner. At Arc90 we all wear a lot of hats, and one of mine is interviewing and hiring new developers. This is what works for our company. We have a specific (smaller) team size, expectation of capability, and company vibe. It might not be the perfect fit for yours.

Overall

1. HR should not be hiring.

The person in charge of hiring should be the most technically talented, passionate, and effectively communicative person on your team. This is never HR. Regarding a particular need, HR only knows what you tell them, and while they may be passionate and communicative, they are never as connected to the role as the people actually involved in making the product.

This almost certainly will change from role to role. Choosing carefully here is absolutely vital. It will color every step of the process from here on out.

The Req

2. Know what you want!

Be clear about what you’re looking for, and be succinct about it. A req with a soup of tech keywords spilled all over it will get you a similarly broad spectrum of applicants, and likely turn away the candidates who are the best fit for your specific needs.

3. You’re pitching them.

Here’s the truth: You want to hire someone that’s out of your league. The people you hire have a direct correlation to the quality of your company’s work, of course you’d want to hire the absolute best person you can afford. Do not be condescending in your req! These are incredibly talented people who have a choice to go to any number of companies. They can walk if they want to. You need to convince them that your company is the best place for them.

3a. But be genuine.

Here is what I know already about the role you are hiring for: You do not want an actual ninja, or an actual rock star. If you did, you’d be posting on shurikens-and-heroin.com rather than the 37signals job board.

You must give a genuine representation of the vibe of your company if you want this person to work for you. If you play up an inaccurate angle, they’ll sniff it out the minute they come in for the interview. You want them to be doing the work you’ve got waiting for them, and be happy about it.

4. Hand write your req, carefully, and proofread it.

For the love of god, make sure your req is well written! Spelling and grammatical errors can immediately sour candidates. If you’re expecting them to put the effort into putting their best foot forward, shouldn’t you? Avoid any Job IDs, ALL CAPS, or unnecessary boilerplate information. This should feel like hand crafted, thoughtful communication.

The Filtering Process

5. Reject poor communicators immediately.

Absolutely vital, this. Proper use of language and being able to quickly and accurately get your point across is the keystone of an effective team.

Additionally, programming is an exercise in semantics. If a person cannot clearly communicate in language, the odds of them being able to effectively communicate the intricacies of technical language are small.

6. Require code samples up front.

This is the fastest way to winnow the pool of candidates, aside from spelling and grammar. Many candidates will just not attach or link to any code samples, which means you can ignore that applicant immediately: They don’t follow directions well. Many other candidates will just have poorly written code. I’m often able to cut about 90% of the remaining candidates just by using code samples as a metric.

There’s also a number of potential bonus points here. Links to github, or other open source projects, can show a commitment to teamwork or open collaboration.

Here’s another tip: We’ve gotten a small number of candidates who have sent in plagiarized code samples over the years. Using something like google code search can help sniff them out, and we’ve outed a number of them that way before.

7. Myth: You’ll only find top tier talent on top tier job boards.

It is absolutely true that you’ll have a higher signal/noise ratio by spending the money and posting on top tier boards like 37signals and Stack Overflow, etc, but that doesn’t mean you can’t find excellent candidates in other places.

For example, when hiring, while we almost always post on some premium boards, we also regularly post on craigslist, and some of our best people have come through it. Your bullshit filter must be incredibly high, and you should include some bozo filtering (requiring code samples is great for this), but that doesn’t mean that talented people don’t occasionally lurk there for one reason or another.

8. Go with your gut, with a bias towards rejection.

Interviewing is expensive. It’s important in this phase to reject early to save yourself time down the line. Trust yourself, as the person most knowledgeable about this role (see point 1), to know what is a good fit and what isn’t, and save both you and the candidate some time if it just doesn’t feel right.

9. Myth: “Smart People are Language-Agnostic”

This is something we’ve wrestled with at Arc90. Often our impulse is to hire people purely because they are effective communicators, and they are great coders, and who cares if they don’t yet know Scala because they can learn it, right?

This is a mistake. The missing piece here is that humans have predispositions, and this particular human may not be predisposed to enjoy Scala as a language. As I’m sure you know, programming in a language you do not enjoy can be a soul crushing experience.

If you find a good candidate who you think is a potentially great fit, but doesn’t know the domain at all, have them spend a few days getting to know it before either of you make a decision. It’s valuable information for both of you.

The Interview

10. You’re still pitching them.

So you’ve filtered your candidates, and found a few that you’d like to talk to in person. Don’t blow it. Don’t forget that this is still an opportunity for you to hire someone smarter than yourself, who will make your team and your company better. Go into every interview thinking of it as an opportunity to find a good match, not an opportunity to grill someone to see if they’re good enough to join the ranks of your hallowed organization.

Additionally: This is really the first impression they’ll have of you. Try to be at least decently dressed for your role, but dressed indicative of it. If you wear a t-shirt every day, at least wear a clean one.

11. Converse, don’t interrogate.

This is not a test.

Well, okay, it is a bit, but it shouldn’t feel like one. This is a time to get to know each other, to assess cultural fit, and to assess communication skills and their ability to speak technically. By now you should have already seen a good deal of their code, at this point you just want to see if they’re sharp with a time limit, and you can talk to them.

12. Be up front about sticking points.

If you’re uncomfortable about an aspect of this candidate and their fit, now is the time to say it. It will be far more uncomfortable if you make them an offer, and they accept, and your fears become reality.

13. Puzzles don’t solve anything.

At this point you should already be familiar enough with the candidates code samples that you can discuss it without having to look at it. For example, you should remember that they used class inheritance in an interesting fashion, and be able to discuss it with them. You’ll learn far more from something like this – real world code with real world reasoning – than you’ll learn from “there’s a bus travelling 50mph with a quarter tank of gas left and how fuel efficient will it have to be to get to Boise, Idaho”?

–

Hopefully all of this has been pretty obvious. When it comes to the hiring process, a lot of it just follows the golden rule. Were you on the other side of the table, how would you want this interview to be conducted so that you’d get maximum value out of the position long term? You would want to be somewhere where you’re providing value, and be happy doing it. Luckily this is exactly what the company wants. The rest is just fit and numbers.

Looking for examples of carefully crafted job reqs? Luckily, there are a few right over here: Arc90 is hiring.

I’ve pitched to friends the tools that we internally use at Arc before, and have been asked the same question by others, here’s the full list of the web applications we currently are loving:

E-mail and collaboration with Google Apps for Business

Email with GMail – our email is hosted by Google. Why? It’s cheap and secure. GMail is also a great email client (conversation view, in-line chat) and lightening fast search.

Asset Collaboration with Google Docs – we frequently collaborate on something by way of a Google Doc, it’s just an easy way to get a bunch of people looking at the same asset – outside of email.

Calendar with Google Calendar – there are some good features, but overall we’re unimpressed with this product.

Group chat with Campfire – our informal ‘rule’ for the usage of Campfire is that any chat that could be useful to more than just one person should be in Campfire (versus an individual chat). Our usage of Campfire is varied, but common usage centers around a few people discussing a topic or for software deployment nights where are 10+ people are collaborating on a specific effort. All chat transcripts are saved and are searchable.

  Group coordination with Yammer – one of the most common questions heard in our 30 person company is "What is [insert name] working on?". We use Yammer to helping raise the level of awareness about what people are specifically working on at any given time. "working from home on our IA Summit presentation, so consider me off today" was my update this morning letting everyone know where I am and whether I’m available. Bonus points: Doug built a bridge from our Yammer account directly to our Arc90 Twitter account, allowing people to pass messages through Yammer directly to Twitter, like so.

Outside communication with Twitter – Twitter is the killer-app-du-jour, but it is a phenomenal way to communicate the goings-on of your company or product(s) with the world at large. If you’re not considering the opportunities created by directly communicating with those outside of your company, be it through Twitter or a blog, you’re missing out.

Idea management and collaboration with Kindling – sure this is a plug, but it’s really an essential tool for Arc90. In fact, I can’t imagine running a business without it (or an equivalent tool). The people working at Arc care tremendously about what’s going on in the company and are motivated to continually try to improve it. Some examples of ideas that came directly out of our use of Kindling: have an internship program, switch to a bottleless water cooler, release Kindling as our first product (), host networking events, have an Arc90 prom (thankfully hasn’t yet happened!) and have a monthly newsletter. We’ve seen ideas big and small, but the increased level of engagement with the company since adopting Kindling internally is undeniable.

Project management with Basecamp – some people at Arc run efforts with Basecamp, but the company hasn’t committed to it as our tool for P.M. It’s a great product though, and also a good one to begin with if you haven’t yet started using web-delivered apps in your business.

If you’re working at a company that doesn’t allow you to adopt tools like these – keep pressing the issue, the market is moving towards this being widely accepted and commonplace. I’m sure that’s very comforting for those currently trapped in SharePoint by their corporate IT department.

We arrived at Mashable’s New Year NYC Networking Eventat 212 Restaurant and Bar last night around 7:30, prepared to spread the good news of our first product launch, Kindling. Much to our surprise, there was a pretty significant line already forming. We
thought, ok – cool, we’ll just wait in line with everybody else and chit chat to pass the time. Who says you can’t wait and network at the same time? Someone started passing out cookies (always a plus). At this point it’s about 10 degrees on 65th Street. After a few minutes, word gets out that the restaurant is full to capacity and no one is allowed to enter until more people leave. Needless to say, this was not good news.

This is when things started to get interesting. We were already with a great group of people that we wanted to hang out with and get to know better, we just needed warmth, space and alcohol. There was clearly only one option in this scenario: host an alternative party.

Jen, our in-house, one-woman, PR-machine took this task very seriously and paired up with someone else (Frederick) who was standing in line. They immediately started scouting the area for bars. The area seemed to lack a nightlife… all they could find was a hair salon called “Blow” (and even if it was a bar, probably wasn’t the kind of place we were looking for, as comically noted by Frederick). Meanwhile, Rich was searching the Internet on his iPhone and had a short list of possibilities, one of which was the Carriage House, a semi-short walk away. We knew this was our place. Jen and Chris rounded up the troops and started corralling everyone downtown. Once the word got out, the masses followed us towards our new prospect of a bar. It was hard to tell at the time, but there appeared to be a steady stream of people up and down the sidewalk. We were almost to the bar when we were split up by a miscalculated green light. This is where we really bonded; it was cold, the walk was longer than we thought, and the Carriage House wasn’t exactly where the Internet promised it would be. Doubt was starting to set in and we feared losing some of the group, but Jen quelled the dissension and we forged on.

When we finally arrived at the Carriage House, it was like walking into heaven. Aside from finding warmth, it was the perfect place. There was empty space in the back of an otherwise busy bar and it just seemed to be waiting for us to arrive. We coordinated with the hostess and told her to direct anyone looking for the Mashable event towards the back.  When we saw the long line of people streaming in, it finally hit us. We were hosting a party!

There were so many amazing people there. Between Twitter and word of mouth, the word had spread and Mashable’s own Adam Hirsch even joined us for awhile. What was particularly surprising was the sheer diversity of reasons that people came to last night’s event. Some were affected by the recession, some were scouting new talent, some were pandering their recent inventions to the Tech World Elite. There were writers and accountants and programmers and bloggers and radio DJs and VCs.  Jeremy showed photos of his 10-month old daughter. Lucius demonstrated a recent iPhone app he’d written. Lauren crossed out the info on her old business cards and scribbled her new info on the backs. Our own Josh challenged people to games of darts and snapped some pictures on the side. In short, it was exactly the way you want a networking to be: connective.

The night was great for a variety of reasons, but mainly because of the people. We could have thrown in the towel and headed home, but we soldiered on, made friends and new connections. But we couldn’t have done it without the goodwill of the others; we were excited to have spontaneously gathered a group of new friends.

To put a perfect end to a perfect story, the Carriage House sent us flowers this morning…seriously, they did. Big thanks to the Carriage House who was not expecting us, but took great care of us.  We had a fantastic time!
 

Two weeks ago, Netflix decided to discontinue one of their features: Profiles. I was annoyed by this, but not enough to make any noise about it, or even tell Netflix. Truth is, I assumed that the decision was final, and that nothing I could say would matter. Well, Netflix proved me wrong, and has once again demonstrated their dedication to customer satisfaction, by reversing their decision in response to customer feedback.

I just found this message in my inbox:

The tone of the message is perfect, just perfect. I’m very impressed, and I sincerely hope that we at Arc90 can do half as well in our customer interactions – particularly if we’re ever as big as Netflix.

1. Select Deposit from the on-screen menu
2. Insert the endorsed check into a slot, just as I would insert a dollar bill into a vending machine
3. When I saw an image of the check on-screen, enter its amount [2008/07/02 UPDATE - It gets even better. I discovered today that it actually OCRs the check amount! After I inserted a handwritten check this morning it magically asked me to confirm whether the amount it read on it was correct. It was! I deposited two other checks and it asked me to enter the check amount, indicating that it couldn't read the amounts or it didn't have a high degree of confidence in its scan, so it didn't waste my time asking me to confirm them. I assume OCR would work nearly 100% of the time on machine-generated checks.]
4. Repeat for any additional checks
5. Press Done
6. Get receipt showing an image of my check(s)

Here’s what I DIDN’T have to do:

1. Fill out a deposit slip, which means I didn’t need to…
2. Find a working pen, which I would need if I could…
3. Remember my checking account number, which I never do, so I used to have to…
4. Wait in line for a teller to give me my checking account number, before I could…
5. Check the under-copies of the deposit slip to make sure nobody leaned hard and smushed their deposit info down into my slip or surreptitiously wrote in their account number to scam me into depositing my check into their account. (And to think some people trust paper.)

Chase saw a process where a machine that is already operating in a high-security and high-knowledge context asked me to write down the account number it already knows on a piece of paper it can’t read and seal it into an envelope it can’t open. What a wonderfully archaic set of steps to get rid of!

For me the customer, life is now clearly simpler, but what about the poor machine? Like all banks, Chase is very comfortable with ATM software and hardware, but surely the addition of machine-reading and imaging each check makes the ATM vastly more complicated and error-prone, yes? Well, no. Every check in the U.S. banking system has the issuing bank’s routing number, account number, and check number printed along the bottom in a standard Magnetic Ink Character Recognition (MICR) format designed for accurate high-speed scanning. Banks have been whipping checks around and reading those numbers all day every day for decades. And the check image I saw onscreen and printed on my receipt? They’ve been doing that for years as well. When’s the last time you got actual cancelled checks back in your statement? What might seem like an incredibly complex and high-risk project for you or me was to Chase essentially sticking together a few very familiar Lego pieces they’d used a million times.

Any time you see a business process that asks for information it either already knows, already should know, or can easily find out for itself, there is an opportunity for vastly improving the user experience. When you further see that you can do so with technologies you already live and breathe, to me it reads like the story of the perfect development project. Let’s review:

• First, there was a clear customer benefit: Customers will spend a lot less time doing tedious and error-prone steps that do not benefit them.
• Second, there was a clear client benefit: Bank staff will spend a lot less time looking up account numbers, opening envelopes and deciphering handwriting on incorrectly filled out deposit slips, and making sure the ATM kiosk is stocked with envelopes and deposit slips. Chase will also spend a lot less money buying and almost immediately disposing of NINETEEN MILLION deposit envelopes a year.
• Lastly, from an IT standpoint, the requirements are clear and the risk is low: Get rid of all that paper by gluing together all these things we already know very well.

Dare I say it, this project sounds fun! Unfortunately, not all IT projects are such no-brainers. It’s not uncommon for a project to be approved with only two of those three stars, or sometimes even one. Maybe it benefits the client and the customer, but it’s going to be a huge and risky endeavor to implement. Maybe it will be a huge pain to implement and actually infuriate the customers, but the cost savings make it worthwhile (“Your call is important to us. Please listen carefully as our menu items have changed”).

In my experience, the one-star and two-star projects tend to come from an incomplete understanding of the business problem, either by IT or frequently from the business itself. Sometimes clients don’t really know what they want. Sometimes they’re too close to the action to think long-term. Sometimes they don’t understand technology at all. And sometimes we in IT just don’t ask, because we’re too busy trying to keep production up and running while watching Twitter feeds of the iPhone 3G announcement. I’m just saying.

No matter how it happens, if you simply treat isolated problems as isolated technology projects, you’ll get paid for your work, and life will go on. However, a deeper relationship with your client lets you do so much more. Understanding your client’s business plans and long-term goals lets you and your client both see the big picture. You’ll be surprised to see how this wider worldview leads to a more holistic approach to these seemingly tactical technical problems (“the ATMs fill up with deposit envelopes by 11:30. Make it hold more.”) When you do that, you’ll see that those one and two star problems can often be rearranged into three star solutions that make everybody happy. Sometimes so happy that they’re inspired to stay up late writing paeans to something as mundane as a deposit-friendly ATM.

On whole I think they’ve gotten the user experience down to about a B+.  Its questionnaire format does a good job of guiding you through the various tax issues and making sure you’re saving as much money as is allowed.  Once completed it can be a little confusing when you need to change something but overall my complaints were minor. Solid but unremarkable software.

Except… Turbotax had a trump up its sleeve.  It turns out they’ve negotiated data sharing agreements with a range of payroll providers, including mine. I was shocked to discover that in about 60 seconds (after entering my provider’s ID and then a figure from my W2 as a “password”) all of my wage information was imported automatically.  No typing, no double checking.  Wow! What a killer feature. Not only did it validate my decision to purchase Turbotax it got me excited enough to come into the office the next day and rave about it.

It struck me that the real work in including this feature wasn’t accomplished at the keyboard but in the board room.  Sure it was probably hard work to format and import the data but the real sweat was in getting all those payroll providers to sign up and share their information. It seems like an obvious move but it took someone at Intuit to not only have the bright idea but to pick up the phone and start calling payroll companies.

I think software folks too often attack problems with a strictly technical mindset.  Can you picture the W2 brainstorming meeting? “What if we have them scan the W2 then OCR the data? Wait! What if they enter only half the fields and we can calculate the rest?” Technical wizardry is great but it’s only one tool in the software palette. It’s easy for developers (and ex-developer manager-types) to default to this way of thinking. It’s in their comfort zone, it doesn’t require unpleasantness like negotiations and contracts, and it offers the chance to write whiz-bang code. For some challenges the technical approach works, but for others it’s sub-optimal or just impossible.

The marketplace has lots of examples of software that wouldn’t have succeeded without this approach: Apple’s iTunes store, OpenTable’s reservations, Seamlessweb’s ordering.  All of these work because people at those companies were willing to step out of the realm of code and do some old-fashioned horse trading. In the context of software as a product, the non-technical approach can mean golden opportunities for companies versatile enough to take it. 

I signed up for JetBlue’s TrueBlue point program today, and I was greeted with this success page:

Thanks for joining. Your TrueBlue number is ###. Unless you’ve got a photographic memory, you might want to print this page for your reference. (You can also cut out the card below for your traveling convenience – or to impress your friends.)

Awesome. Made me smile and I immediately felt good about booking a flight with them (I hadn’t yet, as I wanted to get the number before I booked). They really sealed the deal.

In addition, they had a really concise, clear way to describe the terms of the program:

TrueBlue points live for 1 year. So to earn your free flight, you need to accumulate your 100 points in a consecutive 12 month period. Just to be clear, that means any unused points automatically expire on their first birthday. For example, if you earn 12 points on 12/31/04, those 12 points expire on 12/31/05.

No complicated rules, just a plan English description in a down-to-earth tone.

We’re currently working on our first product offering at Arc90, and crafting the writing within and about a product isn’t easy. It’s good to see that JetBlue took the time to put some effort and creativity into how they speak to their customers.

The first order page :

A great set of options, ranging the gamut from free to $300, from digital downloads to CDs to vinyl to Blue-Ray. I really like that the options are selectable with check boxes, so, for instance I could order the $5 download and the $10 2-CD set, with a single transaction, without having to add them to a “shopping cart” and then “check out”.

If you click the “more info” link under the $5 download option, this graphic pops up:

Again, a great set of options, covering a full range of convenience, quality, and ideology. The rest of the ordering process is straightforward and effective. I didn’t think of getting screen shots, but the screen where you choose which file format you want – which smartly appears during the “pickup” phase, after payment – is clear and – again – effective.

I actually like that the non-free download option has a fixed price of $5 – there’s less to decide than a name-your-own-price scheme such as Radiohead used for In Rainbows, and it means that I know that the musician is good with the price. Plus it’s just a bargain – 36 CD-quality DRM-free tracks for $5!

ReadWriteWeb has some good background about Reznor’s previous attempt at digital music distribution. The Wall Street Journal’s Business Technology blog laughably describes BitTorrent as an “underground filing-trading service.” (And who wrote that post anyway? It’s anonymous! Lame.)

Tip: Downloaded the album, and wondering where to keep that nice PDF booklet? If you use iTunes, you can just drop it in to your library, right next to the tracks themselves. I even set the Artist and Album fields so that the booklet shows up in the track listing.