The fact that WhatsApp is so popular is no surprise, and yet to become the number one smartphone messaging app in a sea of same-same services is mysterious. Sure, people like to communicate, and will latch onto a service that lets them do it easily and cheaply. But for every WhatsApp, there are ten other flavours that have cuter logos, more features, more platforms, do more, arguably do it better, and are free. So why is the little green speech bubble growing so fast?
It has gone from zero to a billion messages per day between the end of 2009 and the end of 2011. SMS took ten years to do that, and that works on any phone and on any network, as long as you don’t mind wasting airtime (especially if roaming). So clearly people like WhatsApp. We liked it enough to review it on Gearburn, which resulted in an entertaining email discussion with other co-founder, Jan Koum, and hence this conversation with his partner and co-founder, Brian Acton.
So. Why pick WhatsApp, which is a paid-for service, when there are other free mobile messaging systems coming out the wazoo?
It’s free for the first year to get you habituated, and then you pay $1.99 per year (what is a little dodgy is that you don’t see this amongst all the “send messages for free” on the home page, it’s mentioned in the casual chit-chat when you’re busy clicking on the download links).
What makes WhatsApp special? What does it do well? “Mobile real-time messaging,” says Acton. After a minute or two as the conversation continued, he realized he needed to add another quality: “Simple”.
“Simple mobile real-time messaging. People get bogged down in complexity.”
Which is true, but less than half the reason. Simple is not enough – no, it’s the simple but in real-time that is the trick. As soon as you move out of simple and real-time, you’re in the competitive space of a billion other mostly free messaging systems, from Facebook to email to smoke signals. Not quite the telephone’s truly simple real-time chat, but WhatsApp comes pretty close.
Real-time means presence, and presence changes a communication between two people into a conversation. To have a proper chat with someone you need to know they’re hearing you as you speak, and know the time context. Did they get my message now as I sent it? Did they get it three hours later after a nightmare hot sweaty traffic jam swear-fest? Have they not turned on their phone in three days?
A two-trick pony
So WhatsApp is a two-trick pony. Simple. Real-time. Maybe add a third quality: consistent, simple and real-time.
How does WhatsApp achieve this?
WhatsApp is self-funded (with a little $8-m help from Sequoia), but has cash revenues from subscribers and app sales (USD0.99 for iPhone, free for the rest, suck on that, iPhone users), so the company has actual cash to spend. Bucking the typical dotcom trend, Acton and Koum prefer to spend it on engineering, not blow millions on crazy expensive Superbowl commercials [http://techcrunch.com/2011/02/07/tech-super-bowl-ads-didnt-hit-home-salesforce-and-go-daddy-most-disliked/]. And cash revenue means not having to chase the ‘Oh Lord how many of our users do we have to sell to monetize this bloody service?”, where the user is the product.
Says Acton: “We’re able to take our positive cash and reinvest it in our business – do R&D on new platforms, hire great talent, rather than using our cash for marketing and eyeballs. We don’t do any advertising. We build a product that people will want to tell their friends about.”
The engineering priority of the back-end system that Acton is building is speed. It may be simple, but it’s gotta be fast.
“Complexity of the system revolves around scale and volume – when you talk about delivering a billion messages in a day, you have to be very fast at transferring messages with optimal efficiency. Another dimension is price – we’re definitely cost conscious. We could throw hundreds of servers at the problem, but are very aware of the cost of a server.”
WhatsApp is built on a highly distributed but physically centralized system, largely resting on open source foundations. The server platform is based on FreeBSD, with the message delivery components written in Erlang. This is a telecoms-specific programming language originally developed by Ericsson, then open sourced, designed for high concurrency, high throughput. “We’ve pushed it to its limits,” says Acton.
At the moment the serving platform is generally centralized at the Silicon Valley home base. “Our goal is to broaden the distributed nature of it – primarily for failover and business continuity,” says Acton. They’d like to distribute the servers across geographies. “Always in the back go our head we’ve had the [multi country] distributed idea. WhatsApp’s type of activity is naturally partitioned by country code – we can easily take euro dialing codes and put them in a euro cluster.” (Hmm, and cut areas off more easily, like Twitter?
As a startup that has been largely bootstrapping itself, the founders are deeply and habitually thrifty. Moving to a more complex environment in multiple countries shouldn’t be more expensive, reckons Acton. “As long as we stick with commodity hardware and commodity operation environments we can keep costs under control.”
Sweat the little stuff
The plans for the future are refreshingly conservative in this normally ‘innovate at all costs’ environment of feature bloat. “In this calendar year we’re very focused on the basic messaging platform – honing and improving the engagement of the user,” says Acton. “The basic here is providing a good messaging experience. We’re looking at new features – like “share nearby place” where we send location information that includes name, address and more [information gathered from a Web service].”
This focus on simplicity in user interface design comes back to the consistency imperative. When asked if they’d be looking at adding voice calls as a feature, Acton demurred. “We struggle with voice. It’s difficult to do on low-end phones, and we want to create consistent experience. We specifically don’t want to create tiering [where some phones have some features and others not]. We only build in features that all phone users can have. We agonise over changing or adding even a single setting,” he says ruefully.
This point may irritate Nokia S60 users, who, while supported as a platform almost from the start, didn’t get the set of standard (if peculiar) emoticons, a small but incredibly irksome exclusion.
Acton laughs. “Nokia clients now support the full range of emoticons – from the new update released in the last month. That comes back to attention to detail, and being consistent across devices.” According to Acton this seemingly trivial little feature was actually a really tricky coding problem, as these things so often are.
And S40? Arguably the most widely used ‘almost smartphone’ in the world? “Yep, WhatsApp is now on the S40. It’s not quite beta, it’s available in Ovi Store,” he says. “We expect it to be feature-complete in the coming months.”
A version for Windows Phone 7 is at a similar pre-beta stage to Nokia S40. The company is working on a WinPho 8 version, but Acton was not willing to be drawn on the availability of the app, since a launch date for the platform still lies in the realms of speculation. He says that a team of developers has been travelling up to Redmond to work with the Microsoft team, and it’s very obvious that Microsoft is aware of the all-consuming need to have the most popular apps available and ready for download as the new OS launches, probably having seen RIM and HP (Palm Pre, anyone? Anyone? No?) bomb out on this very issue.
Spreading the WhatsApp love to other phone platforms? “We look at other platforms, but the basis of our decision is around reliable messaging. If you can’t support background data we can’t do that,” says Acton.
Features that WhatsApp are actively working on are mostly around groups: group dynamics, group discussion, experimenting with large group settings and broadcasting.
Inter-operability is not for WhatsApp. It is a closed/proprietary system – although arguably open source if it’s based on open source. But it will not exchange traffic with other messaging providers or systems. No Jabber, no WhatsApp-for-BBM.
“Too early to tell,” is his opening gambit, and then rejected utterly. “With a wholly owned system we can iterate and innovate… integrating systems is a lot of complexity,” he says. This can be translating between numbering systems, or mapping user names to cellphone numbers, or that it integrates fantastically with your phone’s address book.
Would WhatsApp release a client for the desktop environment?
“Looking forward; we’re focusing on mobile products. I’m not going to say we’ll never do it… and do it really well.” He hedges a bit more, trails off, and then kicks the turkey of an idea to the curb. “When you start to talk about PCs it can get complicated really quickly.”
Show me the money
How many of the people that are currently happy but free WhatsApp users will start to pony up the two bucks come the end of free time? WhatsApp is not counting its chickens. “It’s too early to tell. Most users have been on the system for less than 12 months, we’ve accumulated so many users only in the last year.” The company is hopeful that people will like the service so much they’ll pay for it – the new challenge being to make the payment system as simple and fuss-free as signing up to the service. “We’re working to make it that to pay has the least barriers, and to improve the payment experience. In the second half of this year we’ll have much more idea of how it’ll work out,” he says.
The company is glittering, it has the smell of sweet, sweet success about it, but the Internet’s highways are paved not with gold, but with the smouldering wrecks of clever startups that went wrong.
An element that many users want to see is encryption of chats – Acton understands the demand from users for it, but has to trade off on the performance hit that it brings. “We’re trying to figure out how to strike the right balance of performance against the need for encryption.”
Some phone users complain of excess battery drain, and the presence function in the app is still pretty spotty – getting this totally right will let WhatsApp really break away. Is the “Last seen today at 21:49” in the status bar when she was actually WhatsApping someone, or just when something woke the apps thread and it waved for the cameras? Does “Typing…” mean typing in your chat, or typing to someone else and ignoring you? The real-time certainty of talking on the phone has still not extended to texting.
Niggles, but millions of users tolerate them well enough. The question is, do they like the little green speech bubble enough to spend those five minutes paying two bucks per year for it?