Accenture Knowledge Exchange
Overview
Accenture is a giant international company specializing in IT and management consulting. In this project sponsored by Accenture, my team and I explored how we might improve Knowledge Exchange (KX), Accenture's internal platform for sharing expertise among its employees. My redesign of the contribute user flow reduces friction in the content sharing process and encourages more experts to contribute.
Team Members
Hanna Fu - UX Design
Leon Ma - UX Design
Umme Ammara - UX Research
Elaine Xu - UX Research
My Role
Comparative analysis.
Note taking for interviews and evaluations.
Ideation and feedback analysis.
Making high-fidelity prototype of the contribute workflow.
Research Methods
Interview
8 participants
Survey
14 responses
Comparative Analysis
8 products
Social Media Mining
2 platforms
Problem Space
Due to Accenture's stringent practice to protect client data confidentiality, Accenture consultants do not have centralized platform to store project information. Currently, colleagues are still the most effective resource to help new members onboard a project. Still, they need to acquire information from various contacts like clients, colleagues and project managers, leading to a prolonged onboarding process.
Knowledge Exchange (KX) is Accenture's legacy internal platform that allows its employees to share industry knowledge exclusively among themselves. Onboarding team members could utilize KX to acquire part of the needed knowledge.
However, KX has long been known to be unpopular and underutilized. This may be attributed to its poor search functionality and lack of contributions: Few consultants like documenting, and the unfriendly long contributing process, including the necessity to scrub confidential client data from the content, further discourages potential knowledge experts from going through it...
This hinders the flow of knowledge to novice consultants from an otherwise vast pool of talented experts at Accenture.
With that, I ask...
How might we reduce friction in the content sharing process and encourage more experts to contribute?
Design
We conducted design walk-through sessions with Accenture consultants to gather feedback on each iteration of the design and inform the next iteration.
1st Iteration: Sketches
Upload
Checking
Report
Scrubbing
Feedback
#1
AI scrubbing of confidential data, as a substitute to manual scrubbing, would reduce friction in contribute process and thus encourage contribution
#2
The system should display reference to relevant security protocols side by side so the user can conveniently check them while reviewing the uploads.
2nd Iteration: Wireframes
Upload
Checking
Report
Scrubbing
Feedback
#1
The system should allow manual supervision over AI scrubbing to avoid unnecessary scrubbing.
#2
The system should Introduce different security levels for different types of documents to balance security and helpfulness depending on the nature of the uploads.
#3
The system should highlight the specific part of the document related to the security threat detected rather than presenting a summary.
#4
The system should provide better support for action correction, allowing users to delete or replace uploaded documents.
3rd Iteration: Prototype
Uploading
Users start the contribute flow by uploading any document to be included in the contribution. The page gives convenient links to relevant guidelines. Error recovery has been improved: users may delete or replace any uploaded document with the quick action menu.
Security Levels
We introduced security level setting, which will affect how AI scrubs the documents. Since the documents have different levels of confidentiality, the fine-tuned setting ensures confidential data is removed while helpful information is preserved as much as possible for individual documents. Description of each security level is available.
Supervised AI Scrubbing
The review page gives a detailed view of all scrubbing actions suggested by AI, each must be approved by the user. Each suggestion specifies which part of document to scrub and the reasoning. The transparency enhances user trust on AI while maximizes productivity.
Feedback
#1
The "Contribute" button is hard to find. Apply stronger contrast.
#2
It is not clear that each row of the threats reported under the file represents an action item. Add a checkbox to clarify.
#3
The icon for "previous" button does not communicate well. Add label to clarify.
4th Iteration: UI Refinement
Evaluation
Heuristic Evaluation
Heuristic evaluation is conducted with 4 experts. The experts like the clean visual of the redesign, but suggest to give more intuitive labeling of UI elements, warnings on unintended actions and friendlier error recovery.
Usability Testing
We observed 5 users completing the benchmark tasks and asked them to rate the product on System Usability Scale (SUS). The new feature of AI assisted scrubbing received much appreciation. Users note that the redesigned interface is easier to understand than the current version.