jim.shamlin.com

1 - Why You Should Test That

The word "optimization" has become very faddish in recent years, and the author ticks off a long list of acronyms of practices that all seek to attach themselves to the term. Each of them seeks, or attempts to position itself as seeking, a relationship to the only one that really matters: conversion-rate optimization (CRO). That is to say that all the other optimization varieties focus on a very specific task, which is a smaller segment of converting prospects to customers. A we examples:

As such, there's a great deal of activity in "optimizations" that are a piece of the puzzle, but each of which alone is insufficient to achieving the final goal of converting a prospect to a customer or convincing a customer to purchase one more time.

Your Website Is Crucial to Your Business

(EN: The author spends some time talking about the amount of time customers spend online, the amount of money they spend, the influence of the online channel in offline purchasing, and other topics that are big news to anyone who has been on a desert island for the past two decades and does not realize the online channel's significance.)

Your Website Is Underperforming

Chances are, your Web site is underperforming. Even it if seems to be doing well, it can likely be doing better. Even if you've done everything you can think of to make it as good as possible, there are things you haven't thought of yet. There is always room for improvement.

There's a quick aside on conversion rates - the percentage of people who visit a site that complete a desired action (purchase a product, fill out a contact form, subscribe to a newsletter, etc.). Nobody has a 100% conversion rate - so again, everyone can improve.

This translates into real dollars when you look at the cost of advertising. Each visitor to your site costs you something, and if only 5% of visitors buy, then 95% of your advertising dollars are wasted. (EN: The suggestion that this can be fixed by improving the site is premature, though - a badly executed ad campaign can send many visitors to the site who are low-quality leads.)

The author refers to the "halo" effect, which leads people to have a brighter opinion of people they admire in one respect (which is why a talented actor with no academic credentials is regarded as being smart) and also works in reverse (a person who has done something objectionable is regarded as having many other flaws). In that sense the experience customers have with your web site shapes their opinion of the quality of your product and your firm.

The author speaks of load times, which used to be a major concern in the days of modems - and is now back again for mobile devices. The speed of even the latest data networks is horrid, and customers don't care to wait around for you pages to load.

Ease of use is another major issue - regardless of technical performance, if it is difficult for users to get what they want (hard to navigate to content, hard to understand it once they find it, many steps involved in placing an order, lots of questions to answer, etc.) they will reconsider whether they wish to complete the transaction, or leave to find another source.

(EN: It's also worth noting that being about the same as competitors isn't a boon to you. IF the customer decides it's too hard, they will leave. If the second or third site they visit is also difficult, they will be more likely to soldier through that one than go back to the first.)

The author goes into a specific example, rotating offers on Web site home pages. It's a common practice among sites that have multiple things they wish to offer visitors and cannot sort out which one is most important, so their home page becomes a slide show. The author asserts "we have tested rotating offers many times and found them to be a poor way of presenting content"

(EN: Ironically, he then begins to speculate why this might be ... the animation is a distraction, the user doesn't have time to read the one they are interested in before it disappears, navigating back and forth is hard, etc. It's ironic because optimization is about interpreting data rather than guessing at user behavior, which is exactly what he's doing - which is very bad. You may need to speculate to come up with an idea of what to test, but once the results come in, it's important to be objective and avoid speculation in reporting results.)

Web Design for Results (Rather than Aesthetics)

The author is dismissive of "design" in the traditional sense: the desire to create Web sites that are aesthetically pleasing, the kind that win awards for being creative and look great in a portfolio, sometimes even at the behest of a client who wants a site that has visual appeal and assumes that it will get word-of-mouth for having a "cool" web site.

(EN: The author makes generalizations and presents no statistics or examples to back them up. Other sources suggest that visual appeal does matter - people interact more with things that are aesthetically pleasing - but when ease of use is sacrificed in order to do something "cool" it harms the overall results).

On the other side of the coin, there are sites whose design is driven by the technical requirements and limitations of a content management system that is extremely rigid and allows few alterations to its appearance and functionality - in essence, a technology that does a bad job very efficiently. This, too, is an instance of putting emphasis on things that do not matter to the customer and often harm the goal of generating revenue.

The proper approach is to determine what you wish to accomplish by operating a web site, and to ensure that any decision made regarding the site is supportive of, rather than detrimental to, those objectives.

Common Practices Aren't "Best Practices"

The common practices in an industry are seldom based on research or controlled testing - but instead from companies imitating one another, copying what they are seeing without any proof that what their competitors are doing is having any positive benefit - but assuming that they must be getting good results from whatever it is they happen to be doing. This is not merely unscientific, it's blatantly silly.

Compounding the problem, many firms turn to agencies and design firms, who have even less knowledge of their industry or their customer, to show them the way. There is a great deal of bluster and precious little knowledge in these firms, yet their "bogus opinions" seem to be regarded as reliable advice from reliable professionals - until the numbers come in sour, and then the firm hires a different agency to repeat the same cycle.

The author is quick to recognize that he works for an outside agency that firms call for advice - the difference being that his firm doesn't come in with a ready-made solution, but instead offers services to test solutions to determine whether they are effective.

But back on topic, the common practices of an industry may not yield good results at all, and even if they happen to be based on sound reasoning, it doesn't mean they are practicable or applicable given the unique nature of your firm, goals, and market segment. No-one can say for certain what will work for you, not even if it's worked somewhere else. It can be tried, but should be tested.

Whose Opinion Matters?

Traditionally, organizations were ruled with an iron fist from the top-down - such that the opinion of the highest-ranking person ruled out over that of anyone of inferior rank. Then, there was a switch to geeks and prima donnas - people who could convince others that they were knowledgeable experts held sway. Later, there came the age of market research - asking the customer directly. The problem was that there was a gap, and quite a wide one, between what customers say they will do and what they actually do. It was in effect the customers' opinion of themselves.

The testing method takes opinion out of the game entirely. It observes actual human behavior rather than speculating about what actual behavior might be. It uses control groups, the scientific method, and statistical analysis to arrive at an objective outcome.

The Risks and Costs of Website Redesign

Companies that operate Web sites have recognized that when the site is redesigned, the traffic changes - and they generally view this in an optimistic light: more activity on their site means more people are using it, and they presume that the general public is just as excited about their renovation as they are.

However, that's not always true. More activity doesn't always mean more sales, or even more visitors. The author makes a vague suggestion that many major web site redesign projects actually result in a drop in sales and conversions.

If your Web site is of substantial size, a redesign initiative comes at substantial cost - just to make sure that you carry forward without breaking or losing anything important is an effort. Not to mention that anyone with an idea for something new or different wants to bolt it onto the redesign project. As such they can be massive, costly, and disastrous to the business.

Consider the consequences of an arbitrary redesign:

This leads the author to make his case for "evolutionary" site redesign. Making small and incremental changes, testing each along the way. The net result may be a complete redesign of the site, but it happens gradually over a longer period of time, with many checkpoints along the way to ensure that changes are beneficial rather than detrimental. And there are many opportunities to change plans and test new ideas along the way.

Optimization Increases Revenue without Increasing Advertising Spend

Optimization increases the effectiveness of advertising, rather than spending more to reach a broader audience with a less effective message. Applied to a Web site, it creates a more effective landing page (and conversion flow) so that those who are attracted by advertising are more likely to complete the desired action.

Done well, optimization can also increase revenue while decreasing ad spend by saving you the expense of sending out ineffective advertisements. Consider the effect on SEO, in which companies pay by the click. A paid keyword that draws a large number of low-quality prospects (less likely to convert and/or make smaller purchases) represents less revenue from greater expense than a well-targeted keyword that draws a smaller number of higher-quality prospects.

Ideally, you will optimize on both ends - to bring in prospects who are more likely to convert with optimized advertising and then feed them into an optimized flow. A hypothetical example, and a tedious amount of detail, is provided to illustrate how this could work - the net result is a decreased cost per prospect and an increased revenue per prospect.

(EN: The math is pretty basic, but for now seems to be based on the assumption that good leads are in infinite supply - that is, the numbers scale smoothly - when in reality there are diminishing returns on investment and you will invariably hit a plateau or even hard ceiling.)

Conversion-Rate Optimization and Your Business

While the author feels that optimization can help everyone, he does concede that there are certain factors that make it more productive for some than others:

  1. A high level of traffic to the site (not only better returns, but a larger base to experiment with enables results to become statistically significant faster)
  2. Having a well-defined goal such as selling more products, generating more leads, etc. makes results more measurable
  3. Having a genuine desire to accomplish those goals (which seems a given assumption, but it is surprising the number of covert agendas surface themselves)
  4. Having a culture that responds to objective measures of performance rather than subjective.

The author then tosses out some pretty impressive (and likely incredible) statistics for the kinds of results he has achieved, however most of them are extremely vague. He claims to have achieved a 21.3% increase in "ecommerce goals" (what goals?) and a 76.9% improvement in B2B target market (meaning what?). He also notes these numbers sound impressive, but are not achieved overnight or from a single test, but from an ongoing testing campaign involving multiple rounds.

He then runs some more simple math that can be used to justify the expense of optimization: if your Web site turns $100M in business and you can get a 10% improvement, it's a $10M increase in revenue. You can also toss in the lifetime value of those new customers into the pot for an even bigger number.

(EN: Another caveat is that the author is speaking in terms of gross revenue rather than profit. If you can deliver a $10M revenue stream, do not expect that the business will pay $1M to get it. A business that has a 4% profit margin is only getting $400K in profit from $10M in additional revenue, and you have to come in well below that low number to be doing any good at all..)

CRO Works alongside SEO

Because Conversion Rate Optimization and Search Engine Optimization end in the same word, companies often consider it to be an either-or choice. However, the two support one another: SEO gets more people to visit the site from an external source and CRO gets more of those visitors to convert.

There is some interplay between them, as implied before. A bad SEO campaign will flood the site with a large number of low-quality leads, which can undermine the conversion ratio. Moreover, an SEO campaign may wish to load up the page with verbiage to make it appear higher in search engine results, but the vast wall of words that people see when they click through causes them to bounce back out.

(EN: The metrics, however, are fairly easy to separate and show in a matrix - such that you can isolate the effects of each when the audience is split. Day-to-day or week-to-week comparisons remain problematic.)

There are some common interests between them:

(EN: I'm leery of definitive generalizations for CRO. For SEO you can closely consider the computer algorithm that ranks search results and make definitive statements about things that will always work, but for CRO you are dealing with wetware of human psychology, which is harder to analyze and predict.)

It's also noted that SEO only needs to touch the first page in a conversion flow - it generally drops visitors on a landing page where a task is meant to start, and does not seek to drop them in md-flow, so CRO should have domain over all pages but the first, and then negotiate with SEO when there is a conflict of interest on the front page.

(EN: There's also the suggestion to create specific pages for search engine visitors to land on, separate from the pages any normal site visitor would encounter, to separate the traffic and contain any damage from aggressive or ambitious SEO changes.)