2: Performance Marketing

This chapter proposes to consider the current state of Web design and the methods by which companies make decisions pertaining to the design of their sites, specifically those that are based on Web analytics.

Data vs. Design

A Web site can be applied to many goals, which results in the notion that a Web site is good for many things. Differences in opinion on the qualities that constitute a "good" site arise from different agendas, all vying to be considered more important than the others.

While some web analysts wish to promote analytics, itself, as a contender, it simply is not. Analytics is the means of measurement - and as such, setting a goal to have good measurements without a sense of what they are intended to measure accomplishes nothing.. it is, in effect, the means to measure progress toward a given goal, and not a goal in itself.

Ultimately, the value of a Web site must be linked to business metrics - and even at that point, there remains a great amount of debate. Is it more important to gain revenue or save costs, hence is acquiring new accounts more important than reducing the cost of servicing existing ones, hence marketing and operations seek to improve different metrics that can sometimes put them at odds with one another.

The author suggests that the analyst should avoid taking sides in such debates - both are valid goals, worthy of consideration, and worthy of measurement - and deciding which strategy to pursue occurs at a much higher level of the organization. Given sufficient resources, it should be feasible to serve many masters (and in the absence of resources, the decision may be rendered by investment - analysis can be a service-for-payment.)

The "Good" Site

It's noted that many sites that are highly effective for their operators never win "awards" or accolades for organizations that recognize the "best" sites. While there is some interest in getting the attention that comes from such recognition, it is ultimately unimportant and a distraction from doing the very things that will make the site effective for the business and its customers.

In short, an aware is an outsider's opinion - and seeking to win their approval equates to serving the wrong masters. Awards tend to be flighty, and go to sites that do things that are unusual or trendy, and do not take into consideration the interests of the people who actually visit the Web site or the firm that pays to operate it.

There are many examples of award-winning Web sites that failed: if you can find an article or a site that lists "winners," chances are some of the links no longer work - and the further in the past an award was conferred, the fewer honorees will still be in business (or will since have dramatically redesigned their site).

That's not to suggest that what's trendy cannot also be effective (some of the award-winners are still around, and still going strong), but merely that if a site happens to satisfy both criteria, that's entirely coincidental.

(EN: There was an example of a design-oriented site that sought to catch attention by use of trendy design and interactive gimmickry, which was very "cool" indeed, but the author suggests it was not effective in accomplishing its goal. Unfortunately, he can provide no numbers to back that assertion, and it may have been highly successful. I tend to doubt that, but without actual numbers, it's not of much value as proof of an assertion.)

In fairness, the author presents an example where obsession with (bad) metrics failed: another company built a site for a branding campaign, and made a goal of getting people to visit the site. A lot of cash was spent, and a lot of people visited, so the site seemed like a success. But few leads were generated.

The author coins a term, "page packing", that describes what happens when a business makes a goal of improving the web analytics reports rather than considering a more meaningful goal. Getting people to view something, or click something, or sign up to access exclusive content, does not result in benefits to the business. This stems from bad assumptions of causal relationships among phenomena. If you observe that people who view 10 or more pages are 50% more likely to make a purchase, you cannot double your revenue by forcing (by trick or incentive) site visitors to view more pages - you just increase page-views (and possibly frustrate buyers who find they have to go through a lot more clicks to get to the checkout).

Balancing Logic and Creativity

(EN: The title of the subsection, and the premise on which it seems to be written, is that 'creativity' is at odds with logic rather than driven by it - which indicates the author doesn't understand design and dismisses it as some sort of wild and unpredictable force of nature. It's a flawed premise, that will lead to flawed conclusions.)

The proposed solution for developing a solution that is both "creative" and effective is testing: to develop a number of designs for any given initiative and test them with a smaller audience, to arrive at an optimal solution for a broader release.

In some instances, this is not feasible because of design and budget constraints, in which case the appropriate approach is to scale back. Instead of making a massive change all at once without testing, roll it out incrementally, with an eye toward the results, and the preparedness to retreat if you appear to be heading in the wrong direction. In addition to observing whether the users are being pushed down the desired behavioral path, also consider attitudinal surveys to determine whether they are happy about it. (EN: A good point, which returns to the earlier notion of immediate versus long-term ROI.)

To appease the creative types, analytics can be presented as a way to liberate rather than constrain their creativity. Being able to test multiple alternatives means that the designer can pursue a "wild" idea and, if it succeeds, have the numbers to support it, which eliminates the conflict between design and business where the choice of which option to pursue is often based on subjective or extraneous criteria. This goes for visual design, copywriting, and any issue where there is no clear choice among alternatives and the result is a battle of wills.

A case study is detailed in which a company invested a tremendous amount of money in getting a celebrity endorsement for one product line, and as such they felt the need to use the celebrity prominently, to the chagrin of other product lines. Analysis was done to demonstrate that featuring the celebrity on the home page indeed increased sale in that one line, but resulted in decreases in all others that were more damaging to overall revenue. While this didn't entirely solve the problem, as the senior executives who backed the endorsement were hard pressed to lose face, they were willing to scale back the celebrity's presence on the Web site and consider when it was appropriate to leverage his star-power.

Another case-study involved a company wanted to do extensive cross-selling on a site by using advertisements across various products. It was t be a large and expensive campaign to do this site-wide, but after A:B testing a few products, it was found that site visitors paid no attention to the cross-sells, and the response rate was negligible. So the decision was made not to move forward and seek a more effective method for promoting other products.