Testing Your Way to Success

Bob McElwain

Efforts to sell begin with a headline. It may be in an ad or in the title of a page on your site listed in a search engine. Wherever it appears, its purpose is to grab the readers attention, and to compel them to read the ad or listing. The purpose of the content is to cause the reader to take an action such as clicking through to your site.

Once the visitor is on your site, the pages must take up the chore of compelling a further response, such as downloading a program, buying a product, or any of a host of other actions. In what follows, any desirable action can be substituted for the word, "sale."

Apart from improved and/or increased marketing efforts, there are only two ways to increase sales: by improving elements in the paths that lead to your site, and by improving the effectiveness of your site in bringing a sale. The online business person can fine tune these elements far more easily and with greater effectiveness than can be done offline.

The testing of ads before launching a sales campaign is well documented elsewhere. But that a listing in a search engine is also an ad, is often overlooked. The title tag is the headline that causes the reader to read the description tag. And the latter must bring a click to your site, else the "ad" fails.

Since many search engines use these tags to create the listing, improving those currently in place can bring good gains once the pages are resubmitted. While testing in the traditional sense with ads is not practical, the proven methods for creating a great ad apply.

In short, create a attention gabbing headline that compels the reader to read the listing (ad). Create a description that compels the reader to click to your site. One approach is to copy the title and description tags from significant site pages to a text file. Review these "ads" at least once a month. If you did a good job with the initial tags, this is not likely to bring great results. But if the current tags are lacking, results can be spectacular.

Testing The Home Page
Keep this simple. Look only at unique hits and page views (total pages downloaded). Also focus initially on the first screen, for if a visitor scrolls down, you already have a positive response. The task is to make a change, then see what effect it had on page views.

  • If page views increase, assume the change is a plus, and hold the change.
  • If there is a decrease, it's a negative. Reverse the change.
  • If there is little or no change, nothing has been demonstrated. Hold the change or reverse it, as seems best. But make a note and return later to retest.

Throughout, keep accurate records about changes made and the consequences of the change. They are invaluable in providing hints for later changes.

About The Math
Be sure to use ratios when making comparisons. That is, divide the number of page views by the number of unique hits. This erases fluctuations in the hit counts during the testing period.

Deciding upon the length of the testing period is tricky. To some extent, it depends upon time available. If you are getting 1000 hits a day, you may be able to make a change once a day. Once a week works well. For sites generating less than 1000 hits each week, the conclusions may not be as accurate as you would like. This approach is not as effective with sites getting a hundred or fewer hits a day. Still, if the test period is stretched to two weeks, and solid notes are kept, positive improvement can be obtained. In short, smaller visitor counts make it more difficult to be certain a change is positive or negative, but the process does work.

Test Without Violation
Make no other changes in your site during a testing period. Doing so can distort the results and bring bad decisions. If it's just got to be done, reverse the change made and retest later. Let the site run with required changes until you have a good fix on the current ratio of page views to unique hits.

What To Change?
Anything at all. Colors. Backgrounds. The page template. While content is the most important element on the site, all that supports it should be tested. If you like banners, try one. But also try the page without one. And try different locations as well. In the end, however, it is your page content that makes or breaks your site. How much to change is really a judgment call. The first headline on the home page is so important, changing only this one item may be as far as you want to go in one test period.

How Much To Change?
In testing ads, changing only one word may be the limit as the ad becomes polished. On a website, while a word in the headline might bring a change, there is not sufficient time to test so definitively.

How much to change is really a judgment call. The first headline on the home page is so important, changing only this one item may be as far as you want to go in one test period. On other pages, it may be appropriate to try different versions of the first fold. In a "sales letter" a paragraph may be best. There is a lot of guessing to be done. And hunches to be considered. The key is in keeping great notes. An earlier change that made a noticeable positive difference may give you a great idea for the page you are working now.

What To Do While Waiting
Block out a bit of time each day to ponder further changes. The think time is invaluable. Add to the list of possibilities. Recheck your notes in hopes of finding a proposed change related in some way to another that helped. 5 or 10 minutes is enough. It is the continued daily focus on the campaign that pays big dividends. This simple procedure can easily double the effectiveness of your efforts.

More About Testing
There are so many, many things you can test that lead to improving your site, there is no hope of covering them all here. If you have the data, pay close attention to pages upon which visitors click off your site. If you find a common exit page, there is work to be done. Trace paths through your site if possible; they can be very revealing. And if you can, check length of stay. I feel this is more critical than page views. The bottom line, of course, is sales.

Remember: testing converts opinion into fact – In the end, it does not matter what you think about your site. Or about any element on it. It is only opinion. When you test ruthlessly, opinions are replaced with facts: The behavior of your visitors is the only "fact" of importance.