We’re in Amsterdam this week, presenting at a Measureworks conference on web performance and optimization and attending a Tweetup.
Our host, Jeroen, told us yesterday that since the introduction of GPSes in Amsterdam, traffic accidents in the narrow-streeted city have risen significantly. Many people are focused on their instruments, rather than looking around them. This made me think of some issues I’d seen with web advertising recently that would have been hard to detect through instruments alone, and underscored some of the shortcomings of a purely instrument-driven analytics approach.
With much fanfare, Microsoft launched Windows 7. By many accounts, it’s a good operating system, despite the widely derided launch parties they tried to encourage (which, to be fair, did get people talking about the launch.) The launch involved a massive online ad buy, as well as a new online store for the company. Two aspects of this launch caught my attention: The differences between regional stores, and the state of video advertising.
The differences between regional stores
Here’s how Microsoft is selling its wares in the US. There are five steps from landing page to checkout.
Note that you can click on the cropped images here to see a larger view of the page from which it came, if you want to better understand it in context.
First, there’s an invitation to get Windows 7 on the home page. Decent, and since I want it, I’ll click it.
Okay, I’m given some options, but the one I want most is to buy the thing. The other options are stuff like a store locator, but since this is 2009 and I have an Internet connection, I’m most interested in buying it now.
I get to choose which one I want — there are three versions. I can see pricing. Great, I’ll click “Buy now.”
I can choose some options (full or upgrade version, download or shipped version.) The defaults are okay, so I’ll click “Add to Cart”.
The cart contents are confirmed, and I can check out. I have some bonus stuff — a chance to second-guess myself on not getting the backup, which increases order cart size; a chance to buy more than one; and a chance to tell someone else about the purchase. This is good e-commerce behavior. But I’m just going to buy one for now.
That was what I expected, and I have an order, right? Well, not if you’re in Canada. Let’s try the same thing on the Canadian Microsoft site.
And now, Canada:
I start by clicking “Get it now” on the home page. This isn’t exactly the same as the image on the US site, but it’s the same text.
Okay, I’ve got the same three versions from which to choose.
The price is higher, which is too bad because the Canadian dollar is roughly 1:1 with the US dollar right now, so this is a 20% “not paying attention to International currency rates” tax. But that’s nothing new on the Internet.
I also notice that the “buy now” button is called “shop now” in Canada. I assume that’s just the result of A/B testing and that Canadians prefer “shop” to “buy” for some reason. So I click it.
Here’s what I get. Wait, what?
Instead of confirming options and putting it in my cart, I see a screen showing me the names of stores.
Let’s be clear: Store information is not what I wanted. There’s nothing on this page saying, “we’re sorry, in Canada we don’t sell products online; you’ll have to go to a store.” Forget for a minute the fact that Canadians can’t buy Windows software online — astonishing as that may be. This is a completely jarring workflow. After what felt like a normal shopping cart pattern, I was suddenly given a page that made no sense. When this actually happened, I tried the workflow three times before realizing what had transpired.
If you’re a web analyst at Microsoft, would you see this? Likely not — you’d notice a large number of exits from this page, and assume you’d done your job of informing visitors where they could pick up a copy of your new product. Instead, you are experiencing huge abandonment.
Relying on your instruments too much
As the Canadian Microsoft store example shows, sometimes your instruments won’t show you something plain and obvious; you have to use common sense. A simple message on the final page saying, “we don’t sell software online in Canada” would alleviate the misunderstanding and make the page successful. Microsoft could even have a link explaining why — and track traffic to that link, to gauge Canadian interest in online purchases, or solicit comments that might mitigate an outcry from Canadians trying to convince Redmond they’re ready to buy products over the web.
Let’s talk about video
Another part of the advertising blitz around Windows 7 is the videos they’re embedding into CTV’s player website. This is a site on which Canadians can watch recently-aired programs (similar to Hulu in the US and the BBC iPlayer in the UK.)
The CTV website isn’t very sophisticated. The most egregious problem with the site is that it can’t interleave commercials and content cleanly. Where Hulu switches smoothly between a program and an ad, the CTV site actually pops out of full-screen mode each time there’s a commercial, requiring the viewer to re-maximize the player.
Just how bad is this inability to handle interactions with online video? When Microsoft advertises Windows 7 on the site, they run a static image for thirty seconds inviting the viewer to click on an ad at right.
That’s completely broken. It underscores just how nascent online video advertising is today. The only way that CTV can let Microsoft segment its respondents is to force them to pop out of full-screen, then invite them to click on a static image which can be imagemapped.
To do video advertising properly, advertisers will soon demand:
- A referring tag showing where on the screen the visitor clicked (in case there are several visual components)
- The timecode within the ad (in case the ad covers several features.)
- The programming information (what show was it?) and other facts that can be used for demographic targeting
This stuff is available in more advanced analytics and video delivery platforms. But the fact that Microsoft had to waste 30-second video clips on a static image shows just how far we still have to go before video advertising delivers on its promise.
