Testing in Non-Profits
For profit, business-to-consumer retailers are often the focus for web analytics discussions, but the same best practices and tools can extend to the non-profit world as well. Sue Citro of The Nature Conservancy (TNC.org) walked us through a couple simple ways this ecologically focused non-profit used analysis to help protect the Earth.
Make Your Email Smarter
TNC sends out regular enewsletters that highlight a variety of content. Each link in the newsletter is about a different topic, but not each link has a unique landing page. Using their email software Convio, TNC created simple if/then statements that reordered the content of the landing page to more prominently feature content based on the link clicked.
Let Users Pick Your Words, Not Execs
TNC promotes coral reef conservation with a "Rescue The Reef" program. Alliteration is fine and good, but people weren't associating the program with its point of preservation. After tweaking the landing page to include an image of coral and littering the page with the word "coral", revenue increased 395%.
Online Testing Can Influence Offline Choices
Multivariate testing tells you not only what 'recipe' of variations produced the most optimal results (sales, memberships, etc.), but also which factors boosted or hindered outcomes the most. These data can help you guide creative and messaging choices beyond the site and into areas like direct mail and telesales.
TNC produced some breakthrough results by simply focusing on basic guiding principles:
* Put your audience first
* Test and re-test
* Focus on delivering the right message to the right person at the right time
Horse Or Cart? Media Mix and Search Engine Marketing by Pat Stroh of Impaqt
Search engine marketing should not be a considered a silo, other media influence it and its value is not strictly associated with the click. But how do you measure the base level of search activity, activity associated with other media and the interactive effects on other media?
Impaqt's Pat Stroh suggested a measurement cycle that follows this order: Measurement and data collection > predictive modeling & forecasing > scenario building > testing & optimization. The technical and statistical components are best understood by watching his own video of the presentation. Instead, let me highlight the basic conclusions.
- There is no one universal formula that will let you predict the influence of elements of the media mix on each other. Each client and situation is unique and you have to figure out for yourself the relationship in each new situation.
- Not only is there no guarantee that other media will boost the results of other paid search but it's quite possible it could actually hurt other your search results. If you make considerable investments in relationship to non-search media spends, test your assumptions before you continue to invest.
- In addition to the interplay of elements of the media mix, consider the influence of paid search on retail sales. Plot the two on the same graph and highlight times during which the paid search campaign was paused (more details available in his presentation).