Google's Website Optimizer

Talk about a potentially disruptive technology! Google has released a new product that integrates with AdWords: a landing page testing tool called Website Optimizer to help sites maximize conversion rates. It’s high-powered stuff. More info from Google here. It seems only select advertisers will be invited to participate in the beta.

I’m not a software developer – I’m a marketer who would use this tool, or who would — yes — get paid to manage multivariate tests for clients, using expertise in my company to work on the websites in question in conjunction with the tool. So I’d use it to achieve client results. If I own a business, I want to use the tool, for sure, but I might also want to hire someone to do that job. Either way: great tool, want to use it, but it’s not as if everyone wants to be conducting such tests for a living, and it’s not as if every company has all the expertise in-house. In Google’s “pre-test checklist” – again, very well-written – you are advised to “meet with your marketing team” and “meet with your webmaster” to get all the ducks in a row to run the tests. Hmm, so here, Google has likely not attempted to cover all the possible political permutations involved, but rest assured that for many companies, coordinating these events would be worth a hefty paycheck in itself.

If that’s a “project manager” bias so be it, but that may also explain why this post is a bit of a rant. I don’t think a cool tool is so cool if the company portrays it as so easy to use it’s like falling off a log. That’s just not true! But maybe it’s just the press release wording I don’t like (read on), because it downplays the role of expertise in building the experiments (to say nothing of the underlying businesses). In any case, the FAQ copy on the Website Optimizer pages itself is admirably dry and explanatory, including discussions of “full factorial” multivariate testing.

In Google’s “pre-test checklist” – again, very well-written – you are advised to “meet with your marketing team” and “meet with your webmaster” to get all the ducks in a row to run the tests. Hmm, so here, Google has likely not attempted to cover all the possible political permutations involved, but rest assured that for many companies, coordinating these events would be worth a hefty paycheck in itself.

A short review of Google’s last highly disruptive “price cut” in web analytics:

Not long ago, Google bought Urchin, upgraded and integrated it, and released a high-end web analytics product into the marketplace, charging zero dollars for it.

The fallout went something like this. Web analytics had to get a lot cheaper, even though many customers claimed they wanted more customization that was only available from third-party solutions like Omniture Sitecatalyst and the other industry leaders.

Brave at first, Clicktracks sold only a few months later to J.L. Halsey for a decent, but probably not ideal, selling price. It might be fair to say that was the handwriting-on-the-wall price.

At first there was some resistance to Google’s offering. Agencies and large companies talked responsibly in public about the importance of keeping their data away from the world’s largest technology company.

Just as “jawboning” doesn’t always affect market forces unless the jaw is Alan Greenspan’s, this jawboning at a macro level doesn’t necessarily have strong enough moral force to stop the individual implementer of a marketing project from making their decision based on cost — or possibly even more important — convenience. Given that the access to a full marketing dashboard gets a lot easier for a consultant like me once a fully integrated “full loop o’ Google” system gets installed, it’s very tempting to go in that direction, privacy be damned. Please don’t misquote me: I don’t hate privacy, and I can see how much control advertisers collectively cede to Google by installing Google code on their sites. But on any given day, given the politics and expense of actually getting things done out here (I say “out here” in the sense of the professional marketing playing field, like PGA golfers call the Tour “out here,” and come to think of it why am I typing all this instead of getting out there somewhere in the green grass), there is a strong lunar pull towards Google’s integrated (free) solutions.

So here we go again. Google is now entering the analytics market in a new way, with a multivariate landing page tester called Website Optimizer. This would compete with third-party solutions currently on the market, such as Vertster.

From the standpoint of my group’s consulting efforts, we like Google-made solutions as they work well, are free, give us more we can help clients with, and don’t unduly make the third-party providers or those who have built expensive white elephant, er, proprietary in-house technologies look like the geniuses. In short, we are vendor-neutral so we actually like to use Google-made solutions so we don’t have to integrate lots of disparate ones. That’s why I enjoyed using Google’s dayparting feature in AdWords and began making use of it regularly shortly after it was added.

That being said, giving Google a huge file of what landing page elements, including button sizes, copy lengths, wordings, font sizes, layouts, offers, etc., work best, and user characteristics responding to *different* elements *differently* — well, the mind boggles. Can they possibly use all the information for good? If not, what evil things will they do with it? Will there be another net transfer of wealth from [all other companies in the ecosystem, including private companies in any industry at all] to … [GOOGLE]?

To put it in perspective: Amazon.com built a business around huge amounts of conversion testing, and they pretty much know all there is to know about what works – on their website, for their customers. Google, by comparison, would know about every industry in existence, with more information, too, about what the clicks cost and other clickthrough details. They’d know about the entire loop. Database of intentions? How about a database of intentions and fulfilled desires?

Meanwhile, as trillions of dollars move from us to them, Matt Cutts feels all unclean on my behalf that Yahoo gave me an MP3 player. Just kidding Matt. I’d send you a gift to make up for the wisecrack but I know how you feel about those. Some of your colleagues like cheese plates, by the way, but I’m now too scared to send them.

But I’m probably not really in a generous mood given that Google explicitly notes in their press release that Website Optimizer …

does not require consulting or professional services to implement. By allowing site visitors to determine what content is most useful, as indicated by the highest conversion rate, we are removing the guesswork and trial-and-error experiments that used to be the norm for determining landing page optimization

So, go ahead and do it yourself. Your customers and computers will give you the right answers! Multivariate testing in a box! Agreed, that’s the rough idea, but there is a bit that goes on behind the scenes to get you to the point of testing, let alone to conduct the test, no?

Perhaps all Google means is that the tool works very well, and that testing and not preconceived notions should be the foundation of landing page design. Perhaps this is just a matter of semantics or usability/marketing philosophies, but the wording does sort of sound like a swipe at third parties. “We’ve released a new technology, and only a helpless idiot would need to hire someone to implement it,” is what is sounds like to me. Our clients aren’t helpless, they just don’t run web design and marketing copywriting (and multivariate testing, and media planning, etc.) shops for a living. There is a cost for running sophisticated marketing tests. Google wants to imply – to what end I have no idea – that there is no cost to this (maybe they want props as the “price slashers” for professional services… “we will be welcomed as liberators!”). Forgetting for a moment the main cost – loss of privacy of sales data and consumer behavior to the benefit of the world’s largest online advertising broker – there is also a cost beyond that, the normal costs of running a business and achieving objectives using someone’s expertise.

Although a multivariate testing tool is very good at testing, say, five button designs with five different messages on them, who came up with today’s leading button designs? Who came up with the wording? Who knows that button design is or is not one of the top ten priorities on the page? Top three? Who cares? Who is expert in trends in that area and who would screw it up, even if they had a great tool?

Actually, unless you have extreme click volumes, some of those creative decisions are very important, because you would need 100,000 clicks to test *everything* using brute force. You have to start in a “headstart” position if you’re trying to get answers out of 1,000-10,000 clicks, which would be more the norm. And if those clicks cost $1 each… well, you get the point. You can’t always test all permutations of conversion rates on five button sizes, five button colors, five button designs, and five button wordings on a reasonable number of clicks. There is a real-world environment to consider here and other factors like seasonality and many others can begin to intervene. And in the above example, you’d only have figured out the best button for one particular vertical on one landing page, and nothing else. You could weakly extrapolate the results to other situations. So clearly there’s a need for professionals who can begin with reasonably persuasive, creative, page elements in the first place. And to have them come up with multiple options and to run the tests… will require something akin to professional services.

As Google’s pre-test checklist basically asks: “you’ve already got plenty of creative content to plug into this don’t you?” Well hello. Many companies do not, and expect it to be created out of thin air. Google’s implying that this is trivial and costless is not helping.

Now there’s another factor to consider. Google says that while you’re using Website Optimizer, your AdWords quality scores won’t be affected. But when you select a new landing page, keep in mind that it “might” be affected. It will probably go up, given that Google is bound to think their own tool is doing good things, and they want access to your data so they want you to use the tool, and a reputation for it killing quality scores would not be much incentive to use it.

So does that mean conversion rates are now going to be part of quality score? No? Or that this is a red herring and the use of Website Optimizer actually plays a negligible part in quality score? For the sake of simplicity and agnosticism as to what truly counts as “quality,” I’m assuming the latter. A great user experience shouldn’t necessarily be judged by predefined conversion events.

So if you’re scoring at home or office: I’m like thousands of others in that (1) I’ll probably use this stuff, if they let me; (2) I’m afraid of the larger consequences, especially if Google begins to “organize and make universally accessible — to itself — the world’s confidential marketing information”. At least it’s not causing global warming, that I know of. Google has that covered anyway, with their new solar panels.

Comments

Tag:

Add to Del.icio.us | Digg | Furl