Design by the People: Testing reveals the most effective web designs
Do you allow designers to dictate the structure and appearance of your website? Are design choices made based on looks, personal taste, or (heaven forbid) the designer’s comfort zone?
Design choices should be made by customers – not with words, but actions.
Most online selling is direct marketing - the consumer can contact and buy from the advertiser directly (in contrast to mass marketing, where the consumer buys from a source other than the advertiser – a TV ad for Nike or Coke is mass marketing, a late-night infomercial with an 800 number is direct marketing).
Notable advertising executive David Ogilvy, founder of the Ogilvy and Mather advertising agency, once spoke to direct marketers of their great advantage over other advertisers – because direct marketing results can be clearly measured, direct marketers know what works and what doesn’t. “You direct response people know what kind of advertising works and what doesn’t work. You know to a dollar. The general advertising people don’t know.” He went on to state that “I expect to see direct response people become an integral part of all agencies.” David Ogilvy lived until 1999, long enough to see the dawn of internet commerce, the greatest opportunity ever for advertisers to learn exactly what works and what doesn’t.
Before online shopping existed, direct marketers using mail, print and television advertising were already using formal testing methods that enabled them to know what copy sells best, what graphics work best, what price and packaging, even what envelope works best. Direct marketing is one of the most common and longstanding applications of predictive analytics. These same test methods can be easily adapted to online applications. The online marketplace offers the best opportunities ever for precise and timely measurement, and many online advertisers make good use of these methods. The most successful online direct marketers invest generously in analytics – Netflix awarded $1 million to the team that created the best algorithm for recommending movies, eHarmony uses patented analytic methods for matchmaking, Amazon.com makes constant incremental improvements to its site design based on test results.
Now, the story that doesn’t make the papers: Despite online advertising’s inherent suitability for testing, many, perhaps most, online advertisers either do not test or do not seriously use test results in decision making.
Here are some real life war-stories shared by advertising executives. (For obvious reasons, their names will not be revealed.)
A digital advertising agency creates banner advertisements for its clients and contracts to place the ads on websites. The agency routinely prepares reports summarizing the results for all ads. But the agency does not remove ads which perform poorly and test alternatives which might improve performance.
A full-service agency prides itself on its testing capability. However, testing is not emphasized in its sales process, nor specifically included in its contracts. When business is tight, analysts are cut.
A retailer hires a web design firm to develop its site. The retailer sets no requirements regarding maintainability or design testing capabilities, so the technology is determined entirely according to the preference of the design firm. On the first occasion that marketers request reasonable improvements, the site must be entirely rebuilt.
These advertisers are wasting money on a grand scale. Whether the reason is ignorance, laziness or sheer ego, the results are the same: less than competitive businesses, poor returns, even lost jobs.
Every element in the design of a website can be tested to determine how it affects visitor response. If you are not familiar with the possibilities for testing in web design, this is the moment to start your education. Customers are constantly sending messages about their preferences, not in words, but by their actions at every moment. You must equip your business to decode and respond to those messages.
What kinds of things can (and should) be tested? Start with copy, on web pages, in email, in conventional mail if you use it. Use the copy that you have now as a baseline (more formally, a “control”) and compare with alternatives to see which results in the most conversions (a sale, a new subscriber, information request). You can test copy on your website, in email, in conventional mail if you use it.
You’ll need the capability to serve alternative versions of web pages or send alternate emails to selected visitors. You must invest in proper analytics here – software that can do real statistical analysis and an analyst who can evaluate the results. Think you can get away without those? Your wiser competitor thanks you.
Copy testing is just a start. Test alternative graphics and the location of graphics on the page. Does that “Help” button really help or just distract consumers from buying? Does a red “Buy Now” button work better than a blue button? Does one customer segment respond to pictures of babies while another prefers kittens?
Sit down for a serious talk with your web team. Is your team able to understand your goals and prepared for the training and procedural changes necessary to reach them? Are you ready, technically, to test alternatives in design? Will you need infrastructure improvements, and if so, what will it take to realize them.
If you have never used testing before, first accept that your business is at serious risk because of your failure to do so. A testing program requires time, money and considerable effort in education for yourself and your colleagues, yet many others have made the same investment and thrived because of it. If you’ve been in the dark or kidding yourself about design testing, open your eyes and your mind now.
Real life examples of web design tests, complete actual test pages and results from Anne Holland Ventures.
Print and television ads test examples from the Gallup and Robinson advertising and market research agency.
Video of David Ogilvy’s talk on the chasm between direct response and general advertising.
©2011 Meta S. Brown
Meta Brown is an analytics consultant, writer and speaker. She has introduced and expanded the use of analytics in offices and factories across the US and Canada. Got a question about promoting analytics? Or on using analytics? Just want to say hello? Email Meta at firstname.lastname@example.org, tweet her @metabrown312 or visit http://www.metabrown.com
Other Posts by Meta S. Brown
The moderated business community for business intelligence, predictive analytics, and data professionals.