Solution Brief
The Customer
Launched in June 2007, Slicethepie is a web-based financing platform for the music industry, which enables unsigned artists to raise money from fans and investors to professionally record and release new albums.
The Problem
Slicethepie uses site users (or Scouts) to review and rate tracks. The Scouts were submitting up to 19,000 reviews per day and it became evident that a small minority of the 70,000 Scouts were using techniques to abuse the system, including copying and pasting the same or similar reviews or submitting reviews that were not music related. This resulted in approx 1% of reviews being sub-standard/not genuine.
Slicethepie manually checked every review and this was hugely time consuming, ineffective and did little to support the integrity of the site. It was anticipated that this problem would only escalate as the user base and number of scout reviews increased (up to 1.5 million reviews to date).
The Solution
The solution was completed in two phases:
- Phase 1: Create an automated off-line system to highlight any substandard reviews and identify, via sophisticated forensic linguistic analysis, users operating multiple accounts.
- Phase 2: Integrate the CFL Software’s review checking algorithms into the Scout Room review submission process so that the reviews were checked in real time and only accepted if they passed a range of sophisticated linguistic tests.
The Benefits
- Review checking workload/cost reduced by 88% from 8 hours per day to just 1.
- 10 fold increase in identification of “bogus” reviews from 5-10% to close to 100%.
- Bogus reviews rejected pre rather than post submission
- Number of accounts being locked reduced by almost 90% (despite increased user numbers).
- Quality of reviews is visibly better and Slicethepie have received a lot of very positive feedback from both Scouts and artists.
- Credibility of the Scout Room (and hence the site) has been greatly enhanced.
- Review administration now efficient, cost effective and hugely scalable
- The software is language independent, working effectively regardless of user nationality.