Supporting Users After Software Deployment through Selection-Based Crowdsourced Contextual Help

Abstract

This dissertation introduces LemonAid, a new selection-based crowdsourced contextual help approach, and investigates the larger social and organizational factors of software design, software support, and post-deployment usability practices. Key findings from this dissertation show that that even with the best efforts in user-centered design and usability, not all use cases and nuances in user interaction can be anticipated at design-time. Although users can turn to software support to find help, they often struggle in expressing software problems using the terminology necessary for accessing relevant help. Software teams also struggle in providing one-on-one support and learning about the prevalence of users' reported issues. The LemonAid help approach embeds users' questions and answers directly into the user interface (UI), allowing users to retrieve help by selecting a label, widget, link, image, or another UI element without ever leaving the screen. The key insight that makes LemonAid work is that users tend to select similar UI elements for similar help needs and different UI elements for different help needs. The evaluation of LemonAid's underlying retrieval algorithm showed that LemonAid can retrieve a result for 90% of help requests based on UI selections and, of those, over half have relevant matches in the top 2 results. LemonAid was also evaluated in the field through live deployments on four web sites used by thousands of users to capture the perspectives of end users and software teams. Data from over 1,200 usage logs, 168 exit surveys, and 36 one-on-one interviews indicated that over 70% of users found LemonAid to be helpful, intuitive, and desirable for reuse. Software teams found LemonAid easy to integrate with their sites and found the analytics data aggregated by LemonAid to be a novel way of learning about users' frequently asked questions. The thesis demonstrated in this dissertation is: <italic> A selection-based contextual help system that allows users to find questions and answers from other users and support staff can be helpful, intuitive, and desirable for reuse for end users, and can provide new insights to software teams about frequently asked questions. </italic>