Purpose – The purpose of this paper is to explain why visitors click the last several menu items on a web page at a higher rate than middle items in the menu, a menu recency effect. A secondary goal is to test the use of visitor reaction time data as routinely collected by live web servers on the internet and the use of such data in understanding consumer response to direct marketing efforts at the level of an individual web page.Design/methodology/approach – This paper uses a field experiment on a live web site belonging to a medium sized hospitality business, specifically a restaurant. Data, including visitor reaction time, come from the standard web log files routinely collected on all web sites. The sample consisted of more than 40,000 visitors to the web site.Findings – The reasons for the recency effect seem more likely to pertain to short-term memory advantages for later menu items and less likely to relate to eye movements in which the visitor's gaze prematurely jumps to the end of a menu.Originality/value – The first contribution of this paper is to bring to bear a very large sample in order to explain a paradox of search behavior, namely that consumers are more likely to click on the last link in a menu than middle links. Understanding this phenomenon will allow web site managers to better optimize their sites according to their marketing goals. The second contribution is to demonstrate the usefulness of reaction time data in a direct marketing context. Web log servers automatically collect such data. Such data represent an ideal opportunity to leverage the direct connection between marketers and consumers that exists online.