Skip to content
Luciano Casillas

Luciano Casillas (he/him)

Professional
Data & Insight Analyst at Charter Communications Austin, Texas, US AI Workflow Designer
QR Code

💼 Work Experience

Independent Analytics Consultant

Self-Employed Freelance

Aug 2025 - Present • Austin, TX • Professional Services

Independent Analytics Consultant

Responsibilities

Built portfolio analytics projects (synthetic data) in SQL and Python covering churn, cross-sell, VOC, retention, claims, workforce and call center funnels, with XGBoost/gradient boosting models. Built a 7-stage Python + Claude workflow that scrapes job boards, scores fit and drafts resumes at $0.20 per 40-job run. Automated an Excel/VBA tracking workflow for an 11-person operations team.

Data Insights Analyst

Charter Communications Full Time

Aug 2018 - Jul 2025 • Austin, TX • Telecommunications

Data Insights Analyst, Call Center Analytics

Responsibilities

Built 10+ Tableau dashboards and SQL KPI reporting for four call center business units (15% operational efficiency gain). Built Snowflake, Data Lake, Nexidia and MicroStrategy workflows that cut reporting cycle time 30%. Engineered forecasting and capacity-planning models for Inbound Sales, Retention, Mobile and Repair. Delivered executive trend analysis; raised CSAT 10% with IT, Sales and Operations.

Customer Advisor

Alorica

Feb 2018 - Aug 2018 • Austin, TX • Professional Services

Customer Advisor

Responsibilities

Managed customer cases end to end, resolving escalations. Analyzed dealer and customer data to identify workflow inefficiencies and supported process improvements. Partnered with field and corporate teams on high-profile cases, improving first-time resolution rates.

Rideshare Driver (Independent Contractor)

Uber and Lyft Freelance

Aug 2017 - Dec 2018 • Austin, TX • Transport & Logistics

Independent Rideshare Driver

Responsibilities

Operated as a 1099 independent contractor on the Uber and Lyft platforms, managing own schedule, vehicle and expenses. Delivered direct customer service to riders and resolved issues in real time while meeting platform service-quality standards.

Call Center Supervisor

Centene Corporation (Cenpatico) Full Time

Jan 2013 - Aug 2017 • Austin, TX • Healthcare & Pharmaceuticals

Call Center Supervisor, Managed Care

Responsibilities

Supervised 12-23 representatives on a multi-state government-contract operation, maintaining SLA compliance and QA for every state served. Monitored queues in real time with Avaya and CRM tools. Kept team quality scores above 94% with behavioral coaching and NICE QA audits. Built Excel coaching tools adopted by supervisors organization-wide, and KPI reports for executive leadership.

Customer Service Representative (Temp Placement)

Healthcare Support Staffing / Centene Corporation (Cenpatico) Temporary

Oct 2011 - Jan 2013 • Austin, TX • Healthcare & Pharmaceuticals

Customer Service Representative

Responsibilities

Resolved inquiries, issues and concerns for members and providers by phone, live chat and email in a high-paced contact center. Researched root causes and initiated corrective actions. Documented member and provider information in CRM applications for quality assurance and performance tracking.

Customer Service Representative

Progressive Insurance Full Time

Nov 2008 - Jun 2011 • Austin, TX • Insurance

Customer Service Representative, Insurance

Responsibilities

Handled inbound calls on renewals, coverage, billing, eligibility and policy changes. Educated customers on insurance products, supporting retention and growth. Maintained accuracy across multiple systems while meeting goals for call quality, handle time and first-call resolution.

Technical Team Supervisor

DIRECTV Full Time

Oct 2004 - Sep 2008 • Tulsa, OK • Telecommunications

Technical Team Supervisor

Responsibilities

Led 12-18 billing and technical support agents in a high-volume contact center, holding quality, attendance and schedule adherence targets. Applied behavioral coaching with Excel dashboards to target performance conversations. Sustained team quality scores above 94% through call calibrations and consistent coaching without sacrificing handle time.

Call Center Supervisor (Progressive Roles, Bilingual CSR to National Spanish Call Center Supervisor)

US Cellular Full Time

Jul 1999 - Aug 2004 • Tulsa, OK • Telecommunications

Call Center Supervisor, Retention and Bilingual Operations

Responsibilities

Started as one of 88 bilingual agents; promoted to supervisor within a year. Supervised 18-32 reps and built Excel QA/attendance dashboards. As interim Retention Manager, redesigned the save metric to a percentage-based save rate, lifting churn performance from the low-60s to 84-87% in three quarters. Built the business case for a National Spanish Call Center (64 agents, four teams, top-10 nationally).

Skills & Expertise

SQL 75%
Python 75%
Data Modeling 75%
Tableau 50%
Snowflake 50%
Power BI 50%
Microsoft Excel 50%
Data Analysis 50%
Data Visualization 50%
Forecasting 50%
Business Intelligence 50%

📁 Portfolio & Projects

Telecom Customer Churn: Retention Playbook

Issue: Elevated first-year churn tied to unresolved billing and tech issues in a customer's first 60 days. Findings: Across 225K subscribers, early friction drove 16.7% of churn, putting $32.45M of 24-month revenue (not margin) at risk. Recommendation: Fix early-tenure billing and tech resolution first and focus retention outreach on the highest-risk deciles; at an assumed 20% save rate, about $6.49M is recoverable. Tools: Python, SQL, XGBoost (AUC 0.84, 1.84x top-decile lift), SHAP. Synthetic data, based on my Charter work and rebuilt for a new industry to show skill transfer.

Multiproduct Household Cross-Sell Propensity

Issue: No systematic way to decide which single-product households to target first for multiproduct cross-sell. Findings: Across 150K households, $280M in projected lifetime value sits unconverted; households that started a property quote converted at 38.3% vs. 26.1% baseline, and Independent Agent/Elite/Auto households at 33%. Recommendation: Prioritize started-quote and top-decile households through the Independent Agent channel. Tools: Python, SQL, gradient boosting (AUC 0.69, 2.05x top-decile lift). Synthetic data, based on my Charter work and rebuilt for insurance to show skill transfer.

NovaStar Telecom VOC Survey Analysis

Issue: 120K post-interaction surveys were collected but never used to find which dissatisfied customers would leave, or why. Findings: 40.6% were detractors, about $47M in annual lifetime value at risk; resolution ease, wait time and first-contact resolution drive 69% of model importance (detractors rated resolution ease 6.3 vs. 7.4). Recommendation: Coach to resolution ease and first-contact resolution, and work a ranked outreach list; at an assumed 20% save rate and $25 per contact, about $8.2M net. Tools: Python, SQL (12 queries), gradient boosting (AUC 0.81), Streamlit dashboard. Synthetic data, based on my Charter work and rebuilt for telecom VOC to show skill transfer.

Waypoint P&C Retention Analytics

Issue: Leadership lacked one answer on who is likely to lapse, whether digital quoting yields better customers, and how much cross-sell is missed. Findings: Billing-distressed policies lapsed at 36% vs. 23% (13-point gap); digital quoting renewed at 81.8% vs. 73.0% (about $130 more 36-month CLTV, no selection control); 26,033 auto-only policies (60%) were never quoted renters. Recommendation: Flag billing-distressed policies for early outreach, quote the renters add-on to auto-only customers, and test digital quoting before crediting it. Tools: Python, SQL, cohort analysis, XGBoost/SHAP (modest, AUC 0.61). Synthetic data (60K policies), based on my Charter work and rebuilt for P&C to show skill transfer.

Call Center Scheduling Conversion Analytics

Issue: Operations, QA and Training had no shared agent-level view of what drives a scheduled appointment to convert. Findings: Across 120K leads, callbacks within 15 minutes scheduled at 52.9% vs. 37.7% at 60+ minutes; about $110M in scheduled-but-unconverted order value vs. $62.7M realized; Physician Referral converted at 22.6% vs. 11.5% for BPO outbound; new-hire conversion ramped from 13.9% to 17.5%. Recommendation: Set a 15-minute callback target, favor referral lead sources, and build onboarding around the ramp curve. Tools: Python, SQL, logistic regression, Excel scorecard. Synthetic data, based on my Charter call center work and rebuilt for healthcare to show skill transfer.

CHRT Workforce Intelligence

Issue: Leadership could not see which advisors were at risk of leaving, why, or how much headcount to plan for. Findings: Across 430K records, SLA breaches rose from 11.9% (under 15 open cases) to 26.6% (30+); the 27% of advisors paid below 0.90 of band midpoint left at 19.9% vs. 3.1%; 84 early leavers cost $4.6M to replace. Recommendation: Cap caseloads, review pay for below-band advisors, and use the ranked risk list to guide retention. Tools: Python, SQL, XGBoost with SHAP (about 7x top-decile lift, 5-fold cross-validated). Synthetic data, based on my Charter workforce analytics and rebuilt for financial services to show skill transfer.

Claims Denial Prevention Analytics

Issue: Which denial drivers are real, and which internal assumptions still hold? Findings: Across 200K claims, prior-authorization delays caused $10.4M (35%) of denied dollars; a $4.07M coverage-limit category assumed unrecoverable had a 59% appeal success rate; a two-proportion test showed network status did not drive denials (p=0.44). Recommendation: Attack prior-authorization delays first, appeal coverage-limit denials, and drop the network-status theory. Tools: Python, SQL, hypothesis testing, XGBoost (near-random, AUC 0.51, reported as such). Synthetic data, based on my Charter work and rebuilt for health insurance to show skill transfer.

Trellis Health Plan: Enrollment & 834 EDI Analytics

Issue: High EDI enrollment rejections with no view of which segments drive them or which pending transactions will fail. Findings: Across 15K transactions the rejection rate was 13.0%; Duals (18.5%) and Marketplace (18.2%) rejected at over 2x Medicare (8.7%); Broker Portal was highest by source at 16.4%; 3,567 members (43%) had transactions in more than one line of business. Recommendation: Fix intake for Duals, Marketplace and the Broker Portal, and review multi-line members for coordination-of-benefits exposure. Tools: Python, SQL (DuckDB), logistic regression (AUC 0.637, 1.65x lift). Synthetic data, based on my Charter work and rebuilt for health plans to show skill transfer.

Concord Clinical Reporting Tracker

Issue: One uncaught error in a weekly client campaign report can damage an account, and QA relied on manual spot checks. Findings: Across 351K rows, checks found 2,450 duplicate-key rows (a double-loaded feed) and 9.55% of reports outside the 80-120% pacing band, though overall pacing was within 1.5% of target; Email CTR ran 5.25% vs. 1.25% for Native Display. Recommendation: Run rule-based checks plus a ranked escalation queue (8.28x top-decile lift) so analysts review the riskiest reports first. Tools: Python, SQL, scikit-learn, Streamlit, Plotly. Synthetic data, based on my Charter reporting work and rebuilt for pharma media to show skill transfer.

📜 Certifications

  • Data Science with SQL and Tableau - eCornell University