Mining Word of Mouth Communities in 3G Mobile Networks
0 likes411 views
This document summarizes a presentation about mining word-of-mouth (WOM) communities in 3G mobile networks. It discusses analyzing call records to identify WOM communities through visualizations and train-of-thought analysis. Identifying influencers within communities and precision targeting of marketing allows enterprises to maximize budgets and fine-tune customer relationship management. Future research opportunities include automating the visualization process and investigating communities' subject matters through online capabilities. The conclusion is that identifying previously unknown WOM communities through mobile data analysis can provide new insights and competitive advantages for marketing.
Mining Word of Mouth Communities in 3G Mobile Networks
- 1. BSI Mining Word of Mouth Communities in 3G Mobile Networks Marketing 2.0 Conference, Hamburg 2005
- 2. BSI Join the conversation MARKETING 2.0 CONFERENCE www.marketing2conference.com
- 3. Suresh Sood, School of Marketing, University of Technology, Sydney Mining WOM Communities in G3 Mobile Networks October 7, 2005 Send Correspondence to: [email protected]
- 4. AGENDA • The Problem • Content Specific WOM communities • Backgrounder (data set) • Visualisation of calling data • Train of Thought Analysis • Seeing the WOM communities • Influencers • Steps to Achieving Enterprise Success • Marketing to WOM Communities • Future Research Considerations
- 5. The Problem How do you identify the natural WOM communities in mobile (3G) networks ?
- 6. Content Specific WOM Communities • Buying a new golf club or computer ? • Complaining about customer service • Friends or experts • Advice or just information sharing – Fishing hole – School fete preparation – Fund raising – School reunion – Soccer – Buying a new car – Retirement planning • Storytelling
- 7. Backgrounder – Data Set Service Video A-B Call Activation Cancellation Date/time Fault code Calling number Customer Dialed number City Call Duration Post code Date of Birth State Martial status … Gender Customer Type Consumer Commercial • April – July 2003 • Key cities Sydney-Brisbane-Melbourne-Brisbane-Perth • 65,536 calls + • Key fields encrypted and/or reformatted
- 8. Visualisation of A-B video calling data Dialing Numbers (A) Called Numbers (B)
- 9. Node Link Visualisation of A-B video calling data
- 10. Train of Thought Analysis • A bottom-up approach using Nodes & Links • Origins in intelligence and “bad guy” investigations • Perceptual process of discovery to uncover structure • Distinguish patterns, structure, relationships and anomalies • Knowledge is colour coded • Marketing Analyst can spot the WOM communities • Not sure why but where does this lead • Harnesses the power of the human mind Marketing Marketing Marketing Data Information Knowledge
- 11. WOM communities in A-B party call records
- 12. WOM communities in A-B party call records
- 13. WOM communities in A-B party call records
- 14. WOM communities vary over time
- 15. Influencers People with Roles – broad and wide social – Information/knowledge networks sources and dissemination – frequently communicate – Social pressure in creating group norms – credible influencers – Social support in trying and – Interested in discovering and using new things. telling other about relevant new ideas Tipping Point (Gladwell 2000)
- 16. Can you see what I see ? Potential Influencers
- 17. The Product Adoption Curve and Influencers • Different than Early Adopters • High Influence in Product Adoption % of Population Adopting Time
- 18. Steps to achieving enterprise success • Identify key communities & influencers • Maximize budgets e.g. preview advertising, special offers can be made or tested with influencers first at much lower costs than comparable advertising experiments. • Precision targeting can be achieved around the communities and influencers • Fine tune CRM around the communities and influencers.
- 19. Marketing to Communities 1. Identify WOM communities 2. Assign profitability – Customer probability of buying service = f { product, influence in community } - Customer network value is based on influencing sales to other customers 3. Efficient and intelligent marketing to communities using viral techniques
- 20. Future research considerations Issue Action Investigating WOM communities can be very Exploration of automating the visualisation expensive by virtue of the labour required to process as well as other techniques e.g. identify them using visualisation alone. CART decision tree technique for classification of a dataset. Privacy concerns prevent the ability to readily Simultaneously provide blogging capability correlate WOM communities with the subject to share stories using 3G service of matter of the information being transferred customer experiences, luxury brands & between parties. country destinations. Identification of influencers and other key Follow up interviews with individuals members of WOM communities identified as influencers.
- 21. Conclusions • New insights from mobile calling data by identifying patterns of WOM communities not previously known can be identified • Opportunity to maximise marketing expenditure, increase precision of targeting and create competitive advantage
- 22. When we dream alone, it is only a dream. When we dream together, it is no longer a dream but the beginning of reality. Adapted from a Brazilian proverb