Director of Analytics
Led North America analytics strategy across a global staffing enterprise, building a data science Center of Excellence and advising senior leadership on technology and investment priorities.
The Work
As Director of Analytics in ManpowerGroup’s North America Data & Analytics Center of Excellence, Jared defined and owned the analytics strategy across the company’s North American staffing brands. He built and led a data science practice focused on deploying predictive and prescriptive models directly into operational workflows, served as Regional Innovation Captain representing North America on a global innovation committee, and led a team of seven spanning data science, analytics engineering, and business intelligence.
Selected Projects
Automated Match
The Problem: Most solutions to automated candidate matching evaluate document similarity between application materials and job descriptions — does the application check the required skill boxes. This approach frames the wrong objective. It replicates existing recruiter behavior rather than improving on it, and it leaves proprietary outcome data on the table.
The Approach: Rather than scoring skill profile alignment, the model was designed around success criteria — specifically, the likelihood that a match is made and accepted by both the client and the candidate. Framing the problem this way allowed the model to incorporate a richer feature set: skill match outcomes as an input rather than the output, candidate and job attributes like pay, seniority, location, work arrangement, and experience, and external labor market supply and demand signals for the local market. The result was a production model integrated directly into the applicant tracking system, delivering 2× engagement performance versus standard approaches.
Ownership and Leadership: Designed the overall approach and hired a data scientist specifically to execute it. Worked with IT architects on technical architecture and built the infrastructure for the analytics team to host and maintain the models. Coordinated handoff to Enterprise Technology for institutionalization across other brands and regions globally. Oversaw and approved all custom modeling and integration code. Worked with legal to address US and Canada legislative implications of model usage, engaged data privacy and security, and designed the bias testing framework and utility following approval by global governance.
Business Forecast Model
The Problem: The legacy forecasting process asked local market leaders to submit week-by-week or month-by-month production estimates, which were then aggregated into a company forecast. Bottom-up forecasts of this type consistently shade high — people weight improvement scenarios and discount business loss and natural market fluctuation. The result was a forecast that was structurally optimistic and limited in its utility for long-range financial planning.
The Approach: Replaced the bottom-up process with a multi-equation distributed lag model drawing from CRM, front-, middle-, and back-office systems. The model analyzes relationships and changes across multiple outcome variables simultaneously — revenue, cost, profit, total units, new orders and clients — using recent trends, seasonal patterns, holiday effects, and cyclical business behavior as inputs. Forecasts are generated at the client and business unit level then aggregated, producing accuracy more than 10× higher than the prior manual process. Results were deployed into leadership performance dashboards with scenario planning capabilities tied to variables business leaders can actually influence: headcount, production levels, and pricing.
Ownership and Leadership: Originated the project. Built the first iteration solo in 2016–2017 for a single brand and region, owning all ETL, development, deployment, and stakeholder coordination. Expanded scope and model coverage as responsibilities grew. In later iterations, transitioned execution to analysts on the team while continuing to own the data science and statistical modeling personally — using the project deliberately as an upskilling vehicle for analysts moving into data science.
Workforce Success Methodology
A model-driven framework assessing client partnership performance using user-entered attributes. Deployed via web application and analytic dashboards to optimize profitability and workforce outcomes.
Order Priority Model
Designed and deployed a model to allocate resources to the highest-value orders, fully integrated with front-office systems at near real time. Drove $5–10M in annual profit improvement.
Infrastructure & Platform
Led the regional analytics transition from on-premises to a cloud ecosystem (Azure / Snowflake), optimizing system performance, enabling advanced cloud capabilities, and ensuring seamless deployment and administration across environments.