"Do things that don't scale" ... and then cash out before the market notices.

Yes. Seems to be a mostly variable cost business. That means you need to lower your per unit cost. This might be their plan.

Yeah, Groupon did something similar in France, hired hundreds of sales guys to go to every possible shop in every city then fired them all and kept just a handful in each city and a team of permanent ones only contacting by phone in the main office.

I can understand why every new market requires an operations team on the ground, but marketing ought to be centralized.

I could see there being localization challenges, especially when you start talking about multiple countries. Not just languages; cultural nuance, idioms, and political awareness all matter.

This is precisely why you have regional marketing teams. It's all situational, but IME a dedicated team for a region (APAC, EMEA, etc.) is better at executing than a single centralized global team. Typically the operational aspects can be delegated to the regions who know the individual markets best, while things like branding stay central.

Does that particular aspect require full time staff in-situ, though?

Just the legal aspects of regulation would. See Uber's struggles in London.

Where better to get expertise in a region's culture than the region?

By hiring a local business to do it, rather than hiring a set of full-time staff yourself?

For a company that's about the gig economy, they make some strange staffing choices.

I'm not sure. There is a big billboard ad here in Sydney, which seems to be coordinated with a radio and online marketing campaign, around saying how 'safe' Uber is. Could campaigns like this all around the world be centralized into a smaller team?

It's not quite that simple. The needs and responses of a local market can be completely different from one to the next.

Because of differing cultures, we respond very differently to color palettes, design aesthetics, safety concerns, cool factors, etc. And since the entire point of marketing is to sell, then the the brand's visuals, branding strategy, responses to concerns, etc, all have to be completely customized for the local market in order to engage customers.

For example one common difficulty tech companies often encounter while trying to enter Japan was how difficult it was to acquire and retain users compared to anywhere else. Some gave up, some would keep trying, and surprise—it would magically work out when you hire a local ad team, local designers, art directors, etc. Western companies' sleek, advanced designs are easy to use and navigate, but the locals understand that culturally, the Japanese society resists change. They are used to early day yahoo/geocities level of banner ads and flashing gifs on a page so that they can feel certain that what they are using can be authentically catered to the Japanese people and supported. Overly simple designs like a Bootstrap framework was simply too frightening, new and confusing.

Same can be applied to pretty much anything from fashion, beauty, food, retail of any kind, to things like Uber, which specifically has its own complexities to solve while trying to enter a taxi-efficient market like that; a completely different marketing problem to solve than say Australia's safety concern as you mentioned.

TL;DR: Can't easily centralize marketing because local cultures and needs are always different; and you need locals to know what locals will want/engage with.

I'm in Japan at the moment and I was talking to someone the other day about design. There's a contrast that always strikes me. On the one hand there's this very restrained aesthetic in so much of traditional Japanese art and culture. On the other hand, most display advertising and a lot of signage in general looks like something knocked out by a random intern with no design experience at all turned loose on a desktop publishing program and told to go wild.

A small centralized team will have a challenge trying to coordinate billboard, radio, and online ad campaigns in dozens of cities in dozens of countries. There will be different language, culture, legal, and logistical issues in every market.

I think it could be yes. We have the exact same campaign for Uber in Berlin right now (in German language of course). But the essence of the campaign is identical.

That roaring sound you hear is the majority of the world's population wailing about how Silicon Valley companies don't understand that San Francisco is a tiny portion of the wide world.

Localization is a big deal, or you're going to make an epic cultural faux-pas. What if "Uber" means something lewd or offensive in the local language? What if your marketing team cranks out English copy that doesn't translate well, or at all?

It might be centralized, but not entirely automated, at least not yet. Thus each city requires headcount.