Stephen McAteer

Logo

Father of two, husband of one. PhD in mathematical physics. Lead data scientist at the Victorian Auditor-General's Office. Long suffering Essendon supporter.

LinkedIn - GitHub - email

Image source: https://upload.wikimedia.org/wikipedia/commons/thumb/f/f1/Telstra_Telephone_Exchange_in_Lowe_Street%2C_Queanbeyan.jpg/640px-Telstra_Telephone_Exchange_in_Lowe_Street%2C_Queanbeyan.jpg
20 May 2017

Portfolio of work with Telstra

Telstra are a huge organisation with amazing data. Being in a tech industry, they are very open to change and new tools and techniques.

The focus on the customer experience was at the centre of everyting we did - hardly a meeting went by where it wasn’t mentioned.

Below is a summary of some the work I was involved with during by time with there.

Call Centre KPI Correlation

  • Investigation in to the KPIs used to incentivise staff
  • Are the KPIs correlated? Could they be rationalized and simplified without sacrificing effectiveness?
  • Tool: Python, statistical analysis

ADSL Customer Satisfaction

  • Investigation into metro and rural ADSL customer satisfaction
  • Counterintuitive results appeared to be related to customer expectation
  • Tool: Python, statistical analysis

Relationship Between NPS and Cycle-Time

  • Cycle-time is the time it takes from an activation or assurance episode to be completed
  • We established that assurance and activation NPS have different cycle-time-dependence
  • Tools: Pyhton, statistical analysis

Plan Rationalization Studey

  • Do legacy mobile and internet plans have a greater assurance workload associated with them?
  • Is the business case to rationalize the plans them supported?
  • Tools: Python, statistical analysis

Review of Contractor’s “Blended” Trial Analysis

  • An analysis of the impacts of a new operating model was conducted by the contractor porosing the model
  • We conducted a review of the analyis conducted by the contractor
  • Ultimately, we concluded that the trial did not achieve its aims
  • Tool: Python, statistical analysis

Extreme Weather Analysis

  • Can extreme weather events be foreseen and effectively mitigated through actions such as CT relocations?
  • Study still under way when I departed Telstra
  • Involved the use of publicly available data, and historical outage data
  • Tools: Python

Program Effectiveness Evaluation - Network Remediation

  • Analysis of effectiveness of network assurance programs (costing upwards of $200m)
  • Communication of results to senior decision-makers
  • Tools: Python, statistical analysis (black-spot fallacy)

Start-of-Day Appointment Compliance Trial

  • Design and oversight of trial aimed at understanding which measures are effictive in improving start-of-day appointment compliance
  • In a limited-scale trial, we discovered the effectiveness of various measures
  • Findings helped design which measures would be taken forward
  • Tools: trial design, analysis of results

Regional Efficiency

  • Is there variation in the efficiency of various regions at completing certain tasks?
  • Can this understanding be used to drive efficiency going forward?
  • Tools: Python, statistical analysis

Field Demand Variability

  • How large is the demand variability across regions?
  • Is this enough to explain shortfalls in supply?
  • Work was ongoing when I departed Telstra
  • Tools: Python, statistical analysis

Field Service Delivery (FSD) Budget Analysis

  • Root cause analysis of FSD budget outcome
  • Review of demand forecast and how it is applied to budget process
  • Tools: desktop analysis

Fixed-Line Activation Cycle-Time v’s NPS

  • Analysis of the impact of changes in cycle-time on NPS
  • Communication of results to seniors
  • Tool: Pyhton

Effectiveness of Exchange Remediation

  • Analysis of the effectiveness of exchange remediation on reducing assurance demand
  • Results were interesting, but statistically insignificant
  • Tools: Python, statistical anlaysis

Technical presentations:

  • Introduction to Jupyter
  • NPS Parameter Estimation Techniques – overview of methods including normality assumption, bootstrapping and a “trinomial” model
  • Introduced to Shiny (RStudio)

tags: telstra - portfolio