Summer 2009 – Cover Story: Proud of his Ignorance

Having earned an advanced degree studying neural networks and artificial intelligence, Tim Ash is anything but uneducated. Yet he speaks of being proud of what he doesn’t know about e-commerce. In fact, he’s often surprised at what moves consumers to action. But his talent isn’t explaining what your customers might respond to. It’s allowing your customers to tell you themselves.

By Tom Dellner

Click to Enlarge

Chances are, if you’ve attended your fair share of online marketing conferences, you’ve seen a presentation by Tim Ash, president of SiteTuners, a leading testing and optimization company based in San Diego. And if you have, we bet you remember it well.

The personable Ash isn’t you’re typical conference speaker. He’s engaging, blunt, funny and often controversial. He’s not afraid to bust a salsa-dancing move if it helps reinforce a point. He’s also perhaps the industry’s foremost authority on testing and optimization–a rapidly developing and absolutely crucial discipline.

We sat down with Ash for a wide-ranging and fascinating conversation which covered the early days of online marketing, to tweaking sites to the tune of $48 million in added revenue–in a month!–and common e-commerce mistakes which might surprise you.

Online Strategies: When did your interest in e-commerce develop?

Tim Ash: Back in the early to mid-’90s, when companies first thought about getting on the web, they talked about a web “presence” and a web “page.”

Basically, they borrowed the copy used in their print communications, thinking of it as a passive representation or “calling card” on the web.

I was always interested in trying to do more on the Internet. Together with my business partner Robyn Benonsohn, we launched a company called Future Focus (this was in 1995). It was a dotcom incubator. We helped companies raise money and were an acting CTO on their management team. But from the very start, our focus was always on function–on doing something, whether it was selling product or building some sort of web database application.

We’ve always had a focus on e-commerce, on an actionable Internet strategy.

OS: You were something of an e-commerce pioneer. Back then, did you and your peers anticipate the evolution of online marketing as it occurred? Which developments came as a surprise?

Ash: The production quality improvements you could see coming–they are so immediately obvious and striking when you view a screen capture from the early days of e-commerce. The standard in 1995 was a plain page with text links. Then people started throwing in graphics. Next came animated images and we had those flashing smiley faces all over the place. It didn’t take too much imagination to foresee a day when rich media, video and the like would be commonplace.

Unfortunately, what I had hoped to come to pass really hasn’t, and that’s a dramatic improvement in usability. Basically, we see people designing more for dramatic impact. Very few are taking the user into account, at least to the degree they should. There’s still a revolution that needs to occur with respect to how functional, useful and intuitive web interfaces are.

OS: When did your interests turn to testing and optimization?

Ash: As an undergraduate, I studied cognitive science and computer engineering. As a graduate student, I studied neural networks and artificial intelligence. So I’ve always had experimentation and scientific methodology in my blood.

As far as the corporate odyssey, we did a lot of large-scale campaign management in the early days of pay per click. Again and again, we saw that the quality of the landing page was “gating” the size of the campaign. In other words, the efficiency of the landing page determined to a great extent how much money could be devoted to the search campaign (and how much money we earned). We eventually became a superaffiliate and started sending the traffic directly to clients and were paid on results. Again, the scale of our programs was determined by the quality of the publishers’ landing pages.

It was a painful lesson to learn, because it was money out of our pockets, but we saw that the bigger, more systemic problem–and business opportunity–lay in being able to improve the landing page.

We did a quick state-of-the-state analysis and found that the testing tools being used were very rudimentary–allowing only for very small scale tests–and assumed that everything on the page was independent of everything else (which is patently absurd because we know that the elements on the page work together synergistically to affect conversion rates).

So we went back to the drawing board and–over the course of several years–developed our tuning engine, which is at the core of what we now do today at SiteTuners.

OS: How difficult was it to get companies to appreciate the value of testing and optimization?

Ash: It was quite difficult at first. It’s easier to get your head around something more tactical, like controlling your traffic sources. Even something as complicated as pay per click is an understood field–you do your keyword research, you write your copy and you adjust your bids. Testing and optimization, by contrast, requires a lot of skill sets. Because people didn’t have all of the skill sets necessary, the field kind of languished. It’s always been something of an evangelistic sell. We’ll go out there and sort of cry in the wilderness. People hear us out and say, “Wow, that’s interesting. In theory,” and then go back to their offices and do nothing about it.

OS: Today, are people beginning to appreciate the importance landing page optimization?

Ash: Even now, the penetration of landing page testing is very low. A recent study I came across showed that for every $80 spent in driving traffic, one dollar is spent on optimizing the pages on which that traffic lands.

However, there was a real game-changing moment a few years ago when Google launched their free optimization tool, Google Website Optimizer (GWO). It allows people to do both A/B split and multivariate testing. It’s been a huge boon, having Google’s momentum and stamp of approval behind it. It’s a fantastic tool and it’s raised awareness in addition to giving companies the means to do testing.

OS: Does the existence of a free tool from a company as powerful as Google help or hurt your business?

More About Tim Ash

Tim Ash was born in Russia, immigrating to the U.S. with his family in the early 1970s when he was seven years old. He lived in Albany, N.Y., Ann Arbor, Mich., and Cherry Hill, N.J., before moving to California to attend the University of California at San Diego on a regent scholarship. At UCSD, Ash studied cognitive science and computer engineering. As a graduate student (also at UCSD) he studied neural networks and artificial intelligence. Upon graduation, he worked in a number of high-tech roles, all related to programming and machine learning, until he was bitten by the entrepreneurial bug. He started his own company–Future Focus–15 years ago, and is still riding the entrepreneurial roller coaster today. Future Focus was an Internet business incubator, helping to launch early-stage dotcom companies and developing some of the first database-enabled web applications with clients such as CompUSA and American Express. Future Focus went up–and down–with the dotcom wave. As Ash surveyed the wreckage of the company after the dotcom crash, he decided the next venture had to focus on something accountable. He then started Epic Sky eight years ago (now the parent company of SiteTuners) which managed (and still manages) large-scale PPC campaigns. But this discipline has become largely commoditized, in Ash’s view. To remain relevant in a fast-moving business such as e-commerce, he believes vendors need to keep swimming upstream to the more difficult stuff that is less commonly understood. He now sees the battleground as landing page testing, and started SiteTuners five years ago. His career has come full circle, though. Ash has finally found a place where he can match his personal skills–being an entrepreneur, leader and visionary–with the subject matter that he loves: math, statistics and the psychology of human decision-making.

Ash: It absolutely helps. The problem was that people weren’t doing testing. The free tool has removed the biggest impediment keeping companies from doing it. In addition to being free, it’s very easy to get up and running, unlike some other tools.

But just having the tool doesn’t mean you can make effective use of it; if I give you a hammer, it doesn’t mean you can build a house. You still need someone with extensive testing experience, with proven ideas about what to test, how to set up the test and avoid some of the practical “gotchas” regarding the type of traffic sources you’re driving to it and other real-world factors that can impact the accuracy of the results. That’s where we come in.

OS: And you also have your own tuning engine. How does it differ from other testing tools?

Ash: We created it, not because we like building software, but because of a variety of needs. There are several limitations that exit with the first-generation tools out there. The first is test size. Because of the math they’re using, you’re forced to use fewer of your ideas in the test. For example, you can only run, say, three headlines, so you are forced to self-censor and leave what might be winning ideas out of the test. (If you knew which headlines to leave out, you wouldn’t need to be doing testing in the first place.)

Also, context and juxtaposition are important. For example, certain pictures next to certain copy might produce a poor result, but if they were assembled in the right combination, they might have spectacular results.

Our tool permits large-scale tests, considers context and juxtaposition and is fully automated–you push a button and get an answer. You aren’t required to apply statistics or make any judgment calls; you simply get the best-performing version and how it performs statistically against the original.

OS: There are so many areas on a site you could test. When working with a client, how do you know where to start?

Ash: Let’s take the common example of a retailer who sells a catalog of products. Typically, these sites have a homepage, a category page, a search results page, product detail, the shopping cart and the various steps in check-out. So it’s really only a few different page templates.

We’ll assess which are the most flawed and which have the most value. For instance, you might have some people landing on your homepage, but because of your direct-linking efforts, most of your traffic is actually hitting your product detail page. So we would focus our efforts there: from the product detail all the way on down through check-out, because almost all of the traffic has to go through that route. (We find that most check-out pages are dramatically sub-optimal, by the way.) And we will usually start by making local changes on a single page versus global changes to a page shell and layout–that would be a pretty major undertaking as an initial test.

OS: What are some common mistakes that you see?

Ash: Clutter and overload of information–we see this all the time. Too much text on a page. We don’t read on the web; we have the attention span of a lit match and usually hit the back button when we are confronted by too much copy. Inappropriate use of eye-catching graphics is another common mistake, graphics that don’t support the conversion goal. If the goal is to have someone click on a button and you have an animation or a picture of a person on another part of the page, I guarantee that the customer’s attention is not going to be on the button.

Ultimately, we often find that the versions that perform best are relatively stripped-down. It’s a page where the eye can rest, where there’s a clear set of decisions with a limited number of choices that can be made and where people are guided naturally (not by something shouting above the rest of the visual din) to focus on the right parts of the page.

OS: Is there a risk of losing brand image with such a plain page?

Ash: It’s about business objectives. Do you want the visual impression or do you want the conversion? Of course, it doesn’t have to be so cut and dried.

You can have multiple goals. We’ll weigh each goal and place our emphasis accordingly. You may want certain content on one part of the page specifically designed with more of a branding goal in mind–and we can measure its success–and have the rest of the page furthering a conversion goal.

Also, one must not confuse simplicity with cheapness. A page can be optimized, very uncluttered, but with outstanding production quality that enhances the brand image.

OS: Any other common mistakes?

Ash: Giving the customer too many choices on a page. Copy in paragraph form rather than in bullet points with a clear call to action. While there’s strong evidence that certain features like walk-on video spokespeople can be very powerful, as a general rule, rich media with motion should be used with caution. Long forms can be a huge hurdle, too; never ask for more information than necessary to complete a transaction. And finally, you need to tie the content to what happened upstream, what drove the traffic in the first place. You need to meet their expectations and keep the promises you made upstream.

But usually, we find that there’s not one mistake that is the source of the problem. Landing pages are often death by a thousand cuts. It’s tweaking numerous elements in the same direction to achieve a more integrated design that produces significant results.

OS: How significant are the results clients can expect, in terms of conversion improvement and bottom-line results?

Ash: Let’s put it this way. The three main activities of an e-commerce operation are acquisition (through search or display advertising), conversion and retention or maximizing a customer’s lifetime value (through e-mail). The first and last activities are pretty well understood and the best practices are fairly mature. You would be hard-pressed to increase cost-effective traffic by 50 percent or increase the average customer’s lifetime value by 50 percent. But we consistently get double- and even triple-digit increases in conversion, from 20 percent on the low end to well over 100 percent.

It’s easy to calculate the financial impact of a test, as well. You take the percentage by which conversion is improved and apply it to the revenue value that runs through the page. Single tests can have a value of hundreds of thousands or millions of dollars. We had a recent test that ran for three or four weeks and resulted in $48 million in additional revenue to the client.

OS: How do you see testing and optimization evolving over the next few years?

Ash: First, I see it becoming a dedicated practice area within companies, just as SEM is today. People will learn that testing is not optional. If you put up content that has value running through it and you haven’t optimized it, you will be shut out of the game by people that do.

Also, the tools will become more sophisticated. There will be more and more automation until we get to a point where there is no start or end point to a test. You’ll be able to throw in ideas on the fly, which will be implemented–or not–in the way that maximizes conversion. The optimization tool will simply be built in to your platform and run constantly and automatically.

Rich Media and Conversion

Bigger data pipes into our homes and businesses have made high-bandwidth video, animation and interactive applications possible within our web browser. Video producers, animators and game designers are rushing in to take advantage of this seemingly enormous and profitable opportunity. More is better–right? Not necessarily. Undoubtedly, rich media will lead to better, more immersive experiences for some entertainment-related online activities. But is it right for online marketing? Just as the supersizing of our fast-food meals can lead to all sorts of health problems, excess in online marketing can often have deleterious effects, as well. Several factors need to be considered: Expense – Although authoring tools are improving and becoming less expensive, finding the right people to create and modify your rich media content can still be a challenge. There is a high bar in place in terms of people’s expectations of quality. Plan on spending substantially more money to create and future-proof your rich media assets against the emergence of higher-quality standards. Attention diversion – In the landing page optimization testing that we’ve conducted, we carefully try to balance all of the visual elements of the landing page in order to focus awareness and attention on the critical conversion elements. Often, this requires a toning down of visual clutter on the page so the call to action naturally and organically arises out of the relative stillness. Unfortunately, many uses of rich media do not consider this. In fact, they often go in the opposite direction, figuratively and literally screaming for attention, trying to out-compete the rest of the page elements. Motion, contrast, audio and prominent page placement can easily conspire to steal precious attention away from the call to action. Particular care needs to be paid to the auto-playing of video or audio. The use of walk-on video spokespeople tends to have either an extremely positive or negative effect on conversion rate because of its unexpected nature and the use of moving and talking people. It all depends on execution and how it is tied to the conversion goal. Browser and hardware differences – It is notoriously difficult to make even basic standards-compliant HTML code look the same across different browsers and browser versions. All kinds of hacks are required, and still some compromises remain. This problem is multiplied when rich media taxes the bandwidth, CPU, memory and operating system of the computer to a much greater degree. Rich media often requires the latest versions of specialized browser plug-ins, and results in an extremely unfriendly and bewildering experience for those who do not have them installed. Immature design sensibilities – Much of the rich media being created today is similar to the early days of other new media. Think back to some of the synthesizer-heavy rock music of the 1970s or the tacky animated icon smiley faces that were the bane of early web pages. Each new technology was enthusiastically adopted, but it was often used at first in a gleeful “gee, look at what I can do now” method. The often-unasked question with rich media needs to be the following: does this presentation method provide clear and compelling advantages over the use of traditional web design? You may find the answer is no. If so, save yourself the trouble and expense of creating it. Difficulty of tracking and testing – Although there are well-defined standards for measuring and tracking traditional web content, the same is not true of rich media. Different metrics exist to measure engagement and the effect of the rich media. It is also difficult to track events embedded in the Flash or other rich media files, making it problematic to set up landing page tests. The bottom line is that rich media is not a panacea. Much care must be taken when using these powerful new tools. You can be pretty certain that they will have a strong effect on conversion. But the devil is in the details, and you may not know whether the effect will be positive or negative. –Tim Ash