I enjoyed Gary Wolf’s story in Wired about Craigslist’s unusual approach to product development. The piece centers on the ways in which Craigslist differs from other internet powerhouses. It has a tiny staff – 30, as compared to 16,000 for eBay – despite the fact that it generates more traffic than Amazon or eBay. It has an ugly, old-school, AJAX-free interface, and few of the web 2.0 community-building features you’d expect from a site rooted in local communities. And the team behind Craigslist seems actively disinterested in making money, passing up countless paths towards more revenue.
Wolf pins some of this unusual philosophy on Craig Newmark, playing up his celebrated weirdness. I can see why this is an irresistable angle for a story, but I know Craig, have had a couple of excellent conversations with him, and think he’s far closer to geek-normal than Wolf’s story implies. (In fairness, the folks who start and run high-tech companies are rarely geeks… they’re folks who can manage geeks. Craig probably has more in common, personality-wise, with top eBay programmers than with Pierre Omidyar or Meg Whitman.)
The key insight in the story, I thought, was the idea that Craigslist is a company based around customer support. Craig famously answers thousands of user emails every day. (I shared a Wifi-enabled business-class flight to Korea with him once, and can vouch for the fact that he’s both phenomenally devoted and a very fast typer…) But Wolf makes the point that the company’s product development philosophy is based around the idea that Craigslist builds what people ask for, and not what engineers or marketers might want to build.
This is a damned radical idea and pretty far from how most companies operate. Most companies (tech companies, at the very least, and perhaps any manufacturing company) are dominated either by engineers or marketers. Engineers tend to design and build products that other engineers would like, even if there’s no market for them… see HP’s mid-70s wristwatch/calculator, which retailed for $795 for the perfect example. There are countless ways that marketers can screw up product development, but the classics involve proposing products that have already built and are already popular, making incremental tweaks to existing products, or proposing products that are unrealistic with current technologies.
There aren’t a lot of companies led by customer service people, largely because customer service is generally viewed as the deepest dungeon within a corporate hierarchy, the dimly lit place that the ambitious folks are all trying to climb out of. Lots of customer service work sucks, especially if you’re crippled by “customer relationship management” systems and 3-ring binders that narrowly circumscribe what support peple can do to help customers.
Smart customer service people are the most important asset to a business – they’re the feedback mechanism that lets you know what’s working and what’s failing. After Tripod merged with Lycos, I ended up with a weird hybrid of responsibilities, leading our business development, research and development and customer service departments. It looked weird on an organizational chart, but it worked really well. Bizdev guys want to make deals, and R&D folks want to play with cool tech, and they both try to add new features to products, not always well-concieved ones. But customer service folks want things to work so that customers will stop complaining, and this can be turned into a tight feedback loop that keeps your bizdev and R&D folks from going too far off the reservation.
Is Craigslist the prototypical customer-service driven company? Probably not. Most people’s complaints about Craigslist focus on spam and fraud. A company more closely led by customer service would probably have much bigger teams policing boards, trying to make these issues less visible to the community. Craigslist’s success – and pathology – probably reflect a deep corporate commitment to staying small and efficient as well as to being customer-driven. But the resistance to feature-creep, the old-school design and the quirkiness are all classic characteristics of customer-service driven business. Props to Wolf for putting his finger on an interesting truth, and to Craigslist for taking such an atypical approach to building an online business.
This idea of producing what people ask for, not what the company experts think is appropriate (the Apple model?) is a great one to consider w/r/t development agencies.
Instead of pushing money and projects out the door, the major donors would do well to ask average citizens in the developing world what they want and need. That would be a good use of ICT *for* development.
Ethan – thank you for your generous reflections on this story. I think you are right about the rarity of customer driven businesses. Something I didn’t write about in the main piece, but that you will probably appreciate, is what it means for programmers to be “driven.” What people ask them to do, with a tone of confidence and entitlement, is often more difficult than it seems; maybe it is even impossible to do without unleashing consequences that haven’t been considered by the person asking. As you say, usually this person is an executive of some type. They may be highly skilled; they may even be right in what they are asking; still, a programmer’s first response may well be a kind of skeptical silence, neither a yes nor a no, along with a posture that conveys an uncertainty about whether, if the request is judiciously ignored, it will go away. In haunting the craigslist user forums, and watching the stream of demands and complaints and requests go unanswered, I had to laugh sometimes. Sure, most of them probably should go unanswered. But it was like the craigslist engineers had traded the annoying managers of a standard company for _thousands_ of annoying managers…