For decades, direct marketers have speculated on how direct mail gets read — and responded to. From scanning headlines to shaking envelopes, these theories attempt to identify how to get someone to notice and respond to direct mail. While you can isolate some of these variables with mail and test them, many organizations don't go to the trouble because of the expense and headaches associated with any print-based test, especially if it relates to a specific creative tactic. The result is taking a leap of faith in following these theories.
In the online world, leaps of faith can disappear, allowing the online direct marketer to quickly understand what generates response. By carefully monitoring their Web sites, marketers can confirm speculation about what people look at and what they are attracted to first, and they can do so in minutes, before the recipient even responds, for far less money than testing with print media.
For online marketers, this means collecting behavioral data about what makes people respond to offers. Using this data, online marketers can make fine adjustments to sequences of messages and other creative elements to improve overall response.
However, too many online marketers fail to take advantage of behavior testing to improve Web site performance. Most Web sites are developed, produced and deployed using input from research and/or intuition.
Once live, however, every Web site achieves an average conversion rate — the ratio of purchases or responses to the number of visits on the first page. This number is the best indication of the Web site's ability to sell.
As a result, the conversion rate is possibly the most critical number in any traffic campaign. If conversion rates are low, media will perform poorly. If they're high, you can expand the media mix and roll out programs in larger volume.
So how do you improve the overall conversion rate? There are no magic answers, but there is a methodology that will point you in a clear direction.
The first step is to simply borrow the concept of testing from traditional direct marketing, which involves testing Web site assumptions against each other, nose to nose.
The second step is to identify the points of greatest drop-off to help you focus your testing where it will do the most good.
Third, develop a systems architecture that allows you to easily test site assumptions, read results, and roll out with winning strategies — all in real time.
Testing with Web sites is easier than with traditional direct marketing media because it has little effect on a campaign's economy of scale. Your investment is entirely in strategy, creative and programming. There are not print runs to split, no black plate changes, and no complicated key coding. You simply fan traffic across multiple sites.
For example, you might test the primary headline on the first page of your site against one, two, or 10 alternatives to see what it takes for your market to click to the next page. Not only will you see changes in conversion, you will also gain insight into the exact wording or positioning your market needs. If one of your tests yields an increase in conversion from 10 percent to 15 percent, your cost per response will drop 33 percent.
Identifying the points of greatest drop-off often results in a few spots where the majority of your traffic decides to go elsewhere. In our analysis of drop-off, we find that about 60 percent of drop-off is on the first page of a site and about 15 percent occurs at the pages dealing with the order process. The balance of drop-off is usually spread across all the other pages.
If this is the case with your Web site, this tells you that focusing tests on just the first page might yield significant changes in response. Tests that yield such changes can be minor — moving a button, changing a headline, etc. You don't know until you try.
Developing a systems architecture that makes all this work may not be so easy. The tools for performing sophisticated testing with Web sites are still evolving. There are enterprise marketing automation tools available, such as Annuncio, Rubric and MarketFirst, which provide varying support for testing in online marketing. You may need to develop custom programming for managing traffic, serving up variations on pages, and creating online reports that clearly tell you what's working. This often requires programming. If no one on staff has these skills, outsourcing the production and hosting of your target sites may be necessary.
Even if you perform a limited amount of Web site testing, you can significantly affect results and learn what it is that people need to become your customers. You'll probably discover it's easier to increase campaign performance by focusing on Web site testing rather than banner, mail, print ad or television spot testing. After all, the Web site is your “closer,” which means it will make or break most campaign successes.
Chris Peterson is the president of FusionDM, a San Francisco, CA-based direct marketing agency. Reach him at [email protected].