In their paper, Am I wasting my time organizing email? A study of email refinding, they looked at 345 long-term users who made over 85,000 refinding actions.
Conclusion? People who create complex folders "indeed rely on thee for retrieval, but these preparatory behaviors are inefficient and do not improve retrieval success."
Search is the curse of the thinking classes. The researchers considered 2 classes of search: prepatory, defining and sorting email into folders; and opportunistic, scrolling, sorting or searching that doesn't require preparation. The prep time can be significant: one study found that people spent 10% of their total email time filing messages assuming the effort would expedite retrieval.
That strategy may have made sense 20 years ago. But modern email clients include important search features such as indexing every word in every email and threading, which automatically structures emails into conversations.
Refinding The research sought to understand actual human behavior "in the wild." So the researchers focused on emailers from many different roles - marketers, engineers, execs, admins - using a Bluemail prototype email system developed at IBM's Almaden research lab by several of the authors of the paper.
Only results from people who had used Bluemail for 30+ days and had used each of the search features at least once were included. They measured daily access behaviors including:
- Sort. Clicking on header fields like sender, date. etc.
- Folder access. Whenever a user opened a folder.
- Tag access. Clicking on a tag.
- Open message. Whenever the user opened a message.
- Operation duration. Subtracting the timestamp of each operation from the timestamp of the next operation.
Results The table shows that opportunistic searches dominate with 87% of accesses. Scrolling - 62% of all accesses - is most popular. Preparatory searches (folder and tag accesses combined) are just 13% of all searches.
Folder accesses took about a minute, over twice as long as scrolls, with both searches and sorts being relatively short (around 15s).
High filers found messages using fewer operations but this did not equate to faster searches, as they took longer searching. Folder-accesses take much longer than searches and sorts.
Do high filers find the target message more often? Nope. High filers were no more successful than low filers.
The Storage Bits take Computers are good for something after all. Who knew?
As masses of data accumulate at an accelerating rate, the limiting factor to utilizing our data is our ability to access it. Search is a key problem of massive data.
The research shows that building human-centered search strategies into info-heavy products like email makes a real difference in how efficiently the wetware at the keyboard functions.
Data-heavy trends are still in their infancy. App designers have a long way to go to improve the interface between the awesome human pattern matching capabilities with the computer's text processing mojo.
And yes, it felt good to have my favorite email strategy validated. I was right to ignore the occasional pangs of guilt over my bulging inbox.
Courteous comments welcome, of course. The paper is available here (pdf).