Jamie Roche assesses which elements worked best for competing landing pages.
At the iMedia Summit in May, Lydia Estrada from Ignited Minds suggested that you can't separate the ad experience from the site experience because, in fact, they are both part of the same event
What does this really mean?
Imagine that you just saw a stylish TV ad for Target, which reminded you that you needed to pick up some new items for the kitchen. You head out to shop, but when you reach Target, the store looks like a Woolworth's from the 1970s. It hasn't been updated and feels nothing like the ad that prompted you to go shopping in the first place. You may or may not buy your kitchen items, but you're unlikely to make a special effort to come back.
Many online ads make this very mistake. For a variety of reasons -- most related to the difficulty and cost of changing site content -- the best a marketer can do is to pick an existing site page on which to land the new traffic, rather than creating something new that ties directly to the ad. It's as though they're using a spanking new TV commercial to send consumers to a store that feels totally disconnected from what they just watched.
Rather, if a landing page reinforces the ad, chances for conversion increase.
With this week's test, we've explored the effect of various versions of the landing page on the success of the ad, hoping to improve purchase intent. For the sake of this test, we've defined "conversion" to mean a prospect clicking through to the purchase page.
If you read Monday's column, you'll know we created multiple landing pages with different combinations of three test elements to see which generated the best conversions. There are 81 different ways that these test elements can be combined; we tested 10 for each traffic source.
Test Results:
We looked at purchase intent for three groups: prospects from the Café ad, prospects from the Rasputin ad, and prospects from traffic that came to the site directly (not through advertising). The test ran for two weeks.
Over the course of those two weeks, we saw a 20 percent to 60 percent conversion improvement over the existing page.
In addition to determining which tested version of the landing page works best with which ad, we were also able to identify which specific elements on the landing page are responsible for the success. These results, summarized in a chart for each traffic source, point to a "best possible" landing page for each campaign.
First, let's look at the results for visitors who did not come to the site from an ad. These people either typed in the URL directly or clicked through from another site.
General Traffic Results:
Winner: "Home Page 3: Woman" (as above) was the winner among the tested home pages. It produced a 30.8 percent lift over the default.
Hypothetical best page: Based on the contributions of each of the three test elements, an untested home page featuring the café drawing as the main image is predicted to perform even better than the winner.
| Alternative 1 | Alternative 2 | Alternative 3 | |
| Main Image | Bottles | Woman | Cafe drawing |
| Thumbnails | Man drawing | Photos | Photos with text |
| Benefits | 4 | 2 | 1 |
- The café drawing outperformed bottles and woman as a main image.
- Among the thumbnails, the man drawing beat out the photos with text by a large margin. The photos without text performed significantly worse than the other two thumbnail options.
- Two benefits outperformed four or one.
Rasputin Ad Traffic Results:
Winner: The third version of the woman landing page (to view, please click the Woman image above, and refer to the bottom version) performed best, with a 60.0 percent lift over the default.
Hypothetical best page: An untested version with the man drawing rather than the photos is predicted to perform better, but by an insignificant margin.
| Alternative 1 | Alternative 2 | Alternative 3 | |
| Main Image | Bottles | Woman | Cafe drawing |
| Thumbnails | Man drawing | Photos | Photos with text |
| Benefits | 4 | 2 | 1 |
- The woman performed the best of the main images, followed very closely by the café drawing. The bottles performed significantly worse than the other alternatives.
- The man drawing and photos performed similarly to each other in the thumbnails category.
- All of the benefit copy options performed similarly.
Cafe Ad Traffic Results:
Winner: The second version of the woman landing page (to view, please click the Woman image above, and refer to the middle version) performed best, with a 31.7 percent lift over the default. This image incorporates the taste better benefit of Sugarshots, along with the photos with text.
Hypothetical best page: In the case of the café ad, the winner matched with the predicted winner based on element contribution:
| Alternative 1 | Alternative 2 | Alternative 3 | |
| Main Image | Bottles | Woman | Cafe drawing |
| Thumbnails | Man drawing | Photos | Photos with text |
| Benefits | 4 | 2 | 1 |
- The main image, whether bottles, woman or café drawing, performed similarly.
- Among the thumbnails, the photos with text had a strong, positive impact.
- The four benefits outperformed two or one benefits.
The test results, while directionally suggestive, did not reach statistical significance within the test duration. This is a fairly common problem, given the normal constraints of timelines and traffic. In these situations, marketers can apply their own experience to interpret the data, and then re-test to see if the interpretation stands up.
Summary of Findings:
For general traffic, the element that mattered most was the copy text, and the strongest version included two benefits.
This might be because general traffic, coming directly to the site via a URL, may already know something about Sugarshots. In that case, all they might need to make a purchase decision is a little more information about why it's better, or how to use it.
For the Rasputin traffic, since there were no strong winners among the elements, we can turn to the losers for insight. Neither the photo of the woman nor the drawing of the café moved this group (which, after all, responded to the "edgier" Rasputin ad). Interestingly, the photographs of the bottles negatively influenced them. This suggests that a different packaging strategy might be tested if the Rasputin crowd is the target audience. A follow-up test might try alternative large images more similar in tone to the Rasputin ad.
Finally, the café traffic liked to read. The list of four benefits and the text on the images made a difference to this group. It would be interesting to dig in further to the copy for the café group. It could be that they respond to more benefits, or it could be that one of the benefits that appears in the list of four -- but not in the shorter lists -- is the one they responded to.
So, while statistical confidence was never reached in the above tests, trends and indications have developed. Going forward, a marketer would have a clear idea for the direction of the next test series.