The global ASICS Digital Marketing team works from our office in Shibuya, which gives us access to many great food options, and has recently been equipped with a fancy coffee machine (no nespresso for us).

As we’re preparing to launch e-commerce globally (actually, we’re already selling online in Japan but it’s still a big new push), I’m fleshing out my team. We’re looking for smart and ambitious people who are excited about bringing ASICS and Onitsuka Tiger products to online shoppers, and will help our regional teams achieve their commercial objectives.

At the moment, we’re looking for two roles:

• an e-commerce product lead, who will develop the platform and create new revenue streams, whether new products or new ways of selling
• a testing and optimization lead, who’ll make the platform work harder and increase our commercial performance

We’ll be happy to help the right candidate relocate to Tokyo from elsewhere in Japan or any location overseas. Japanese-language ability isn’t required (we work in English).

At the same time, my European colleagues are looking for a regional head of e-commerce, based in Amsterdam, and we’re always looking for new recruits in Japan, you can also just drop an email to HR.

Would you like to join our team, and participate in changing the face of this company? Review the job descriptions, and send us your resume with a cover letter: tell us what you’d bring to the brands, and how you’d approach the job!

RT @hunterwalk: Read @bhorowitz ‘wartime CEO’ bhorowitz.com/2011/04/15/pea… & then @mhelft tech.fortune.cnn.com/2013/01/03/goo… as bookends to amazing job La …

— Hiten Shah (@hnshah) Janvier 3, 2013

First, the April 2011 assessment by Ben Horowitz on Page’s new brief points out the difference between peacetime and wartime CEOs, which boils down to whether the company’s market is new, expanding, and relatively non-competitive, because its expansion benefits the company anyway, regardless of whether it benefits other players as well — a wartime CEO is focused on outright winning an existing market from the competition.

Second, a January 2013 article in Fortune, which points out that Page has carried out significant reform in the direction of tighter execution control, focused on a more defined plan, and somewhat stepped back from the culture of all-out innovation with less concern for a business model.

During my last visit to China, internet access had been a challenge and this time, I got myself a cute little wifi access point, an affordable airport rental.

Signal is pretty good, and I’m checking my email and using maps (marginally better in China on iOS 6, by the way!) while on the subway. Everything is dramatically slow but functional. FaceTime even works (although I only succeeded in taking an incoming call, but couldn’t place any myself).

However, this is still the internet from within the Great Firewall. And right now, while Gmail access has been restored, Google itself (the search engine) is still blocked, as are Facebook, Twitter (including the URL shortener t.co), and Bitly.

And TinyLetter is in fact using a captcha supplied by Google, hosted on the banned domain www.google.com. Here’s what it’s supposed to look like:

Now this makes sense.

Site and app makers: if you want to be reliable in China, make sure to host your libraries yourselves. You never know when the country’s censors will take down a site you were relying on, and all those libraries and social media widgets will just refuse to load until they timeout, rather than spit out an immediate error.

It starts well, for example, the response comes back very quickly:

@raphal Hello Raphaël, that doesn’t look good! Can you send us your bookingcode in a DM so we can investigate this further, thank you!

— Royal Dutch Airlines (@KLM) Septembre 11, 2012

However, no solution was then given to this problem: after a few follow-up questions, the @KLM team advised me to use my desktop browser. (And since last month, I haven’t tried the mobile site again — I’m not sure the payment form has been fixed.)

I was left frustrated on my primary goal (couldn’t book the upgrade), an irritated that my issue was not understood and dealt with. And I’m not sure KLM picked up on the actual problem in their mobile app, which makes me unlikely to use it again: why would I waste time with a mobile booking that may fail at the time of payment?

Social customer support rule #1: solve your user’s problem, or die trying. When that’s impossible, offer apologies, as you’re letting your user down after creating an expectation.

Social customer support rule #2: make sure to understand what actual issue is underlying that user’s problem, and when it’s a real problem with your product or service, launch an internal project to fix it for everyone.

Social customer support rule #3: follow up, so the user’s trust in your service is restored. This implies some kind of continuity in tracking the issues and the internal projects that relate to them.

Another one, when I was faced with a check-in problem, with the intro text hawking a possibility that wasn’t actually available in the options listed underneath:

@raphal Hello Raphaël, we just tried this and it works! Please only select the option “send boarding pass” and then this should work.

— Royal Dutch Airlines (@KLM) Septembre 14, 2012

I would never have believed I’d see an instance of “we tried it and it works” in the wild! What an amazingly unhelpful answer.

Social consumer support rule #4: afford your users the same standards of respect online as you would offline, although allowances must be made for the 140 character limit, and the informal nature of online interactions.

A few months ago, I was stuck in a traffic jam on the highway to the airport (an accident blocked traffic for a couple of hours). When pinged, @KLM’s answer was: call the airport, which in itself was rather unhelpful. But what’s worse: no phone number was provided. I was back at square one, and vaguely irritated.

Social consumer support rule #5: when offering support, make sure you’re empowered to actually solve the user’s problem, whether googling something for the user, or navigating the internal teams of your organization and roping in the appropriate resource. Act as a concierge, rather than a switchboard.

I think it’s great that KLM is trying hard to be closer to its passengers. However, the rush to use social channels has undermined key success factors for customer service. And in my personal case, the heightened expectations have led to disappointment, and damaged KLM’s brand image as flailing and useless.

What do you think will happen if I click on S? Will I teach my spam filter that it was right in catching this mail, or will I have the email delivered to me? (I suppose the lack of an “N” option is disabled by the administrator, but then does it really need to be in the caption? Never figured out what the red markers were for.)

My point here is: the actual, real-case usability of a web app is complex, fiddly, and can break down easily. Using short-hand for controls along with captions to explain what the short-hand means — this isn’t a sensible trade-off.

I’m a big user of Dropbox, because it allows me to have my files offline (as opposed to Google Docs online-only approach, although Google Drive is supposed to change that), across several computers, easily accessible on my mobile devices, versioned, and shareable (systematically with designated colleagues who also have Dropbox, or as a one-off with special URLs).

My company relies on the dreaded Lotus Notes for email. It’s bad in so many ways, many of them related to how it’s managed for us (the ridiculously low mailfile size limit and mobile support limited to Blackberry), some related to the architectural design of the tool.

Although more of my tools are having difficulty, I’m deliberately picking those two examples: one is a fairly modern digital-native tool, well-written and very stable, while the other is a notoriously clunky holdover computing concept from over a decade ago.

Both fail spectacularly in this context of low-bandwidth, limited internet access. And guess what still does work well?

Read the rest of Surviving low-bandwidth internet (879 words)

As I have noted at greater length elsewhere, I had come to fear that China may be where the 200-odd-year-old carbon-fuelled capital-driven model of economic development runs into an ecological wall. Britain, where it started, and China may be bookends on a period of global expansion that has never been seen before and may never be repeated again.

Guardian correspondent Jonathan Watts on leaving China after 9 years covering the birth of a superpower.

I live in the 11th district of the French-abroad, covering Russia, much of Asia, and Australia. 20 candidates were competing: 8 from established parties (no common candidates were fielded here, either by the parliamentary right nor by the “presidential majority” leftist parties, which is quite common inside France), and 12 others (a very high number, but not uncommon abroad).

While some of my friends voted online (also a first, fraught with technical issues and the worrying requirement to downgrade your Java virtual machine), I know of a couple who did not vote, while they had in the presidential election a couple of weeks ago.

An interesting issue was the emails sent by the candidates to the voters, as authorized by the French state who supplied the data on CDs (first and last name, date of birth, email address, home address).

Obviously, email is cheaper and more effective than paper propaganda (I didn’t receive the official campaign material on time for the presidential election, for example). Outsourcing the sending to the candidates is not a bad idea either (they have to print their own campaign material, although the state mails it to the voters, in one bundle).

But email marketing has rules — some of which have been translated into law, but many of which are simply understood by people, especially in Europe — such as the need for prior approval before “spamming” your recipients.

And this campaign has broken the rule: it wasn’t really possible to opt out of the emails. Some candidates did supply the option, but it was up to them to honor it, and there was no easy way to opt out of the whole election communication.

This is a beautiful real-life example of the pitfalls of old lists and gathering mailing authorizations offline: even if, technically, the candidates had voters’ approval, many experienced the mails as unsolicited and unwelcome.

Politics sometimes takes a page from commercial digital marketing in a very interesting way (the Obama campaign of ’08 and perhaps to a lesser but nonetheless worthy degree, the work done around Hollande this year). However, heavy-handed or tone-deaf tactics can backfire in a way that affects democratic processes. The participation rate in an election isn’t just another conversion goal.

What mattered was the functionality, and the usability of the interfaces, such a vital topic for consumer-facing elements, became a much lower priority. We did give it some thought and worked on it, but it never received anywhere near the same amount of investment or attention.

We often worked on customizing an interface so it was really specialized for the task at hand, i.e. maybe ugly, but extremely clear for its operators because it supports exactly what they’re supposed to do, and maps to their mental model of their work.

Also, because you have so few users and they don’t have a choice whether to use the system or not, you can address the poor usability of an interface through training and documentation. This is increasingly uncommon in consumer-facing services, but remains standard in much of business software, with a few exceptions.

I often rant about Lotus Notes, IBM’s email, calendar, and bureaucracy-support platform, because its interface is riddled with confusing and impractical idiosyncrasies, poorly abstracted or explained technical details, and also because the level of polish of the interface is nowhere near as good as its mass-market competitors. Yet IBM is a huge, and hugely successful business software and consulting company.

SAP, an immensely successful competitor of IBM, is dubbed Systems Against People by some of its users, who find its interface daunting and error-prone, its flexibility severely limited, and find little joy in interacting with it. In both cases, poor implementation may be to blame, but consumer-inspired models are clearly superior.

But both of these companies, and all of their competitors, produce software its users are forced to use.

Having the choice between many options, consumers can afford to be fickle and shop around until they find a service they like. As a result, immense investments are made into creating moments of joy in using those services. Don’t like Facebook’s privacy issues? Switch to Path. Can’t be bothered with Flickr’s focus on the desktop? Jump ship to Instagram. Usability is a very large factor in the success of those services.

Users of business software are not usually consulted in the selection of that business software, the choice is made by the software’s maintainers instead. As long as a program’s goals are met (i.e. can we actually manage our product catalog online with this e-commerce platform?), the CTO’s job is done. The satisfaction of the operator is irrelevant, and therefore nobody bothers to design for it.

I believe there’s a business opportunity here. Of course, training has a hard cost, and while it’s hard to quantify accurately, I am convinced there’s a cost to user frustration, in productivity and good will.

But there’s an even more powerful element: the learning gap. When you divorce the purpose from the gesture, when you increase the distance between your action and its consequences, learning can’t happen. You are still capable of performing the action (with enough training and sense of duty, we’ll perform the most meaningless, most intricate acts), but you’ll never get any better at it, nor will you ever contribute to the field.

Practically: if the tool with which I create a promotion does not offer me a clear view of that promotion’s performance, I’m going to have to go seek that information out myself. I’ll probably do it anyway because it’s my job and I’ve been trained, but my work is clearly less efficient, and it limits opportunities for implicit learning.

If the platform isn’t even trying to learn from its users but just forces itself on them, it means the collective knowledge that is being built up by all the operators carrying out their tasks every day is being left on the table.

And the more esoteric the platform, the fewer chances it has to create joy for its operators. And that’s too bad because joy for the company’s staff is a powerful raw material to foster more effective work and increase consumer satisfaction.

So here are a couple of principles I’ll be pursuing here at ASICS for our e-commerce project over the next few months, as we are choosing a platform:

• first, pick a partner that understands the issue, and is committed to tackling it
• group tools around tasks, and align tasks with overall program goals, to encourage effective activity
• embed performance testing and measurement into operational tools, to make learning constant and easy
• give users the ability to change their environment and make their own tools, but still offer great defaults
• involve operators, observe them, iterate and optimize, just as we would for consumer-facing systems

And I want to call out Demandware for having identified the issue and being in the process of addressing it (kudos!), and Hybris for having started as well. The other guys out there, I think you’re letting your clients leave money on the table.

But while I was able to connect from the company’s internal network (as a test, as it’s obviously not very useful), I didn’t manage to connect from outside. My company Windows machine could, but not my Mac.

I’ve finally tracked down the cause (after a lot of irrelevant debugging attempts, investigating our firewall and doing various DNS lookups): actually, Network Connect writes to /private/etc/hosts and hard-codes there the IP address of the VPN machine, after you successfully connect. And of course, after I’d connected (as a test) to the VPN over the internal network, the IP address it wrote there was the internal IP. This subsequently prevented it from connecting over a public network.

This is a bit of a note to self, in case I run into this problem again and forget the fix. (Of course it would be nice if the Cisco software didn’t do such a dirty thing in the first place, or at least cleaned up after itself, but that might be a lot to ask for.)