Stop by our booth if you plan to attend these events, and if you’d like to schedule a meeting with an Optimost representative in advance, please

There was an interesting news item last week that attendance at the Museum of Natural History in New York was up nearly 20% over the past few weeks. The feat is all the more impressive when you consider that the museum is in the unlikely situation of having to compete with sunbathers in Central Park during the recent record-high temperatures in the New York area. The increase was attributed primarily to the success of the movie, "Night at the Museum" in which the fossils, dinosaurs, and other creatures on display come to life after the museum closes.

The buzz about "Night of the Museum" got me thinking about fossils, those fascinating vestiges of previous civilizations that at one time ruled the roost, but no longer have a place in our world. Fossils can be found in museums, countless websites and books, and many hours a day on the Discovery Channel.

But there is one other not-so-obvious place where fossils can be found. If you do a little digging into your website, you might be amazed to find some fossils staring you right in the face - specifically, the forms of your registration process or sales funnel. Just like a trip to your local museum, on your website you're likely to find remnants of a past life that served a purpose at one time but now serve little purpose at all.

So what are some of these fossils? Let's take a tour. A common one we see is a request for a fax number (faxis numberis, in archeological-speak). At one time, there was great hope that the fax machine would be a great marketing vehicle - a much cheaper, more immediate way to reach people than direct mail. And it had the added bonus that your prospect actually had to pay for the paper and printing! Thus, many direct marketers starting asking for a fax number in their prospecting. Of course, this form of marketing never quite panned out, and fax spam has the dubious distinction of possibly being even more reviled than email spam. Yet on many websites we encounter, we still see companies asking for fax numbers, even though they have no intention of ever actually using them. It may seem harmless to ask for it, but it never helps response rates, and usually hurts them.

Another category of fossils to keep an eye out for are unnecessary titles (titelis futilis?). Some websites feel compelled to ask if you are a Mr./Miss/Mrs./Ms., etc. Certainly there was a time when addressing someone with the right title was important, and clearly in some parts of society and cultures it still is. But for many businesses, those titles are rarely used. Inevitably, companies ask for a title as just a clever way to classify the gender of a person without actually asking them. But it can lead to problems. Why make someone take the extra five seconds to decide if they are Mrs. or Ms.? Why upset someone that you forgot to include Dr. in your list of titles? And if you include it, you don't know if they are a man or a woman anyway. So if you do need to know, just asking the prospect directly to select if they are male or female is usually the best way to go.

Another complex fossil group is the requesting of phone numbers (phonis interruptus?). Asking for a phone number almost always hurts response rates. Clearly there are many situations where having it is important, or even a prerequisite for conducting business, but it's still important to think about how you ask for phone numbers and how many you ask for. Certainly in theory, having multiple phone numbers of prospects is better than having just one number, but there are additional risks to this as well. Many sites will ask for a home number and work number. But what if the person doesn't work? Others ask for home numbers and cell phone numbers. For more and more people, those can be one in the same. And while some people like to use their cell phone for all calls, others view it as a private number that is not to be distributed to anyone. If you must have multiple numbers, a simple "Phone Number 1 and 2" or "Primary/Secondary Number" is often the best bet.

Some companies try to justify including these fossilized fields by making the field optional - if it's optional, what damage can it do? But even optional fields can hurt you. They make your form appear longer, which will discourage some visitors. Also, some people will not notice that the fields are optional, and will cause them to freeze in their tracks. You wind wasting their precious attention span as they ruminate on whether they want to give out their work or cell phone number.

While these particular relics may not deliver the same cinematic drama as a resurrected T. Rex, they can quietly rob you of sales. It is precisely due to the inconspicuous nature of these registration form missteps that Web marketers often ignore their potential for damage. So the best way to determine if these fossils are mere artifacts or are actually terrorizing you when you're not looking is to systematically test them and see what impact they are having.

And the beauty of making these fixes is the permanent value they have. Inevitably, the buzz around "Night at the Museum" will fade, and attendance at the museum will return to normal. But the changes you make to your forms will improve your conversion rates for ages and ages to come.

To learn more about ways to improve your forms, we invite you to attend our free webinar, "Optimizing Web Forms", on Wednesday, January 17. Click here to register

LetsTalk is an independent reseller of wireless products and services for consumers and enterprises. Through its consumer site, the company helps its customers research, compare and purchase mobile phones and service plans in a user-friendly environment.

An analysis of their online sales funnel revealed that over 90% of LetsTalk’s visitors were abandoning the sales process prior to completing an order. Although they wanted to test different elements on their site, their rapid price changes and cross-carrier product selection required a dynamic content solution flexible enough to handle a wide variety of inventory and offer combinations.

For example, depending upon the visitor's zip code, current carrier, and current contract status, different combinations of phones needed to be displayed requiring changes to both the content and layout of the page presentation. The testing solution needed to account for all of these possibilities.

LetsTalk turned to Optimost, the only company capable of handling their complex requirements, to help them test, optimize, and improve the performance of their key landing page. A dedicated Optimost project team worked with LetsTalk to develop a test plan which included 10 variables and 74 different values (versions of the variables). Based on this test plan, the Optimost solution engine quickly identified and generated over 1 billion possible permutations of the page. Using its advanced methodology, Optimost was able to conduct multivariable tests on a subset of these possible permutations of the test page. The winning creative resulted in a 7.7% increase in order completion which translated to an increase in annual online revenue of over $2 million.

For more details on this success story and other customer success stories, please see http://www.optimost.com/case_studies.phtml.

How long will a test take to reach statistical significance?
The duration of a currently running test can be estimated by knowing three variables: (a) the conversion rates for each webpage being tested, (b) the number of visitors to the test pages and (c) the desired confidence level to reach statistical significance. For an experiment that has not launched yet, four pieces of information are required: (1) the current or historical conversion rate of the benchmark webpage, (2) the current or historical number of visitors to the benchmark webpage, (3) the desired confidence level to reach statistical significance and (4) the expected lift or margin of improvement relative to the benchmark webpage.

In December’s newsletter, we introduced the concept of "error" and how it relates to the calculation of the confidence interval. The same terms used to define the formula for the confidence interval are used to describe another important mathematical relationship when making statistical comparisons. For example, two web pages, A and B, are both being tested and have been running currently for one week. The current conversion rates for page A and B respectively are 12.5% and 14.1% and both pages see about 300 visitors per day. So in one week, 2,100 visitors have visited page A and 2,100 visitors have visited page B. Currently, the margin of improvement by Page B relative to Page A is 1.6% (or 12.8% lift). The current confidence level of page B’s improvement over A can be calculated by using the following formulas:

N	=	number of visitors to the page being tested
p	=	the conversion rate
SEpool	=	pooled standard error (standard deviation)
z	=	statistic value or score determined by the desired confidence required

In this example, the z-score equals 1.53, which corresponds to 87.4% confidence. (We would need a z-score of 1.96 to equal 95% confidence.)

If we want to calculate how much sample is required for reaching statistical significance (95%), the sample size needed can be calculated from the preceding formulas. The assumption here is that approximately the same number of visitors will be viewing each page version (N = N_B = N_A) and that the z-score value is fixed (z=1.96). Hence, the sample size needed for reaching 95% confidence is as follows:

In this example, the total number needed per page is approximately 3,460 visitors:

Since one week of testing has already occurred, 2,100 visitors have been accrued so only 1,360 additional visitors are needed per page.

Considering each test page has a visitor rate of 300 visitors per day, then 1,360 visitors should be reached in about 4.5 days. Therefore, 4.5 additional days are required in order for the comparison between page A and B’s conversion rates to be statistically significant. Of course, this estimate will change if the number of visitors per day to the test pages change.

Also, it is critical to recognize that these estimates will change if the margin of difference between the two conversion rates change. For instance, if the margin of difference, p_B – p_A, in the third equation decreases, the denominator will get much smaller faster than the numerator given that the denominator’s terms are being squared. As a result, the sample N will get much larger and require more time to gather sample. Intuitively, this makes sense, in order to detect smaller differences, larger sample will be required for a desired confidence. The opposite occurs, when the margin increases where the sample N will get much smaller and hence require less testing time.

The calculations used for estimating remaining time for a given test can also be applied to an experiment that has not launched as long as all pieces of information (conversion rate, number of visitors, confidence level desired and lift) are provided. Sometimes, the one difficult piece of information to estimate is the prediction of lift. Depending on the type of variations of pages being tested (more aggressive variations versus fine tuning), reasonable estimations can still be used to give rough estimates of test length. Historical data on the page being tested or pages similar to it from previous tests are also important in approximating the predicted lift for the new experiment.

Optimost’s customers certainly know that the "performance" of your website is dependent upon the presentation of content, offers, and other messages you give to your visitors. However, serious online brands also need to ensure that the technical infrastructure of their site can support an excellent customer experience. Optimost partner Keynote Systems has a wide range of test and measurement solutions that let you know precisely how your websites, applications, and content will perform on actual browsers, networks, and devices.

Keynote’s core solutions have always been around web performance – the ability to test and measure technical performance (response time, etc.) inside and outside the firewall, all the way to the end user. In addition, Keynote has a suite of customer experience products which enable you to perform behavioral and attitudinal analysis, real-world user testing, online surveys, and competitive and market intelligence. Keynote’s Customer Experience/UX Products supply you with valuable insights that can be used to provide direction and ideas for testing experiments as well as to add additional context to learnings gained from website testing.

By combining Keynote's measurement and experience solutions with Optimost, you can simultaneously optimize the technical, marketing, and sales performance of your web presence.

For more information about Keynote, go to www.keynote.com.

Click here for a complete listing of Optimost Partners.

If you are an Optimost partner and you are interested in either having your company featured in our Partner spotlight or your upcoming events listed in our newsletter, please send an e-mail to .