Context
When a government wage-subsidy program was introduced in response to a sudden economic shock, employers needed fast, accurate guidance on eligibility and claim value — and the rules themselves were evolving as the program rolled out. Delays or errors in interpretation carried real financial consequences for businesses trying to keep employees on payroll.
Challenge
The regulation changed faster than typical advisory processes could keep up with, source payroll and revenue data varied widely in quality across clients, and the analysis had to support real decisions — not just a compliance opinion after the fact. There was no time for a slow, sequential process of interpret-then-model-then-advise; those steps had to run in parallel.
My Role
I contributed to the analytics and scenario-modeling work supporting this program, translating evolving regulatory guidance into decision-ready models used in client and internal advisory conversations.
Approach
We built scenario models that let advisors and clients test different assumptions — revenue decline thresholds, employee counts, qualifying periods — against the current version of the rules, and update those models quickly as guidance changed rather than rebuilding from scratch each time. A big part of the work was upstream data quality: payroll and revenue data arriving in inconsistent formats had to be cleaned and validated before any scenario was trustworthy.
Because the stakes were high and the rules were genuinely ambiguous in places, we kept a tight feedback loop with subject-matter specialists interpreting the regulation, so the models reflected the latest defensible interpretation rather than a stale one.
Outcome
The analysis supported client decisions connected to approximately CAD $100 million in wage-subsidy outcomes across the engagements involved — work I contributed to as part of a larger team, not a result I claim sole credit for. The scenario-modeling approach also became a reusable pattern for later rounds of the same program as rules continued to change.
When regulation is moving in real time, the bottleneck usually isn't analytical horsepower — it's data quality and the feedback loop between technical modeling and subject-matter interpretation. Fixing those two things is what makes fast, high-stakes decisions defensible.