How T-Mobile uses Analyst Agent for skill-based compensation

When a job doesn’t fit the mold, comp questions get harder.

For Ardavan Khosravi, Compensation Market Research Manager, at T-Mobile, the hard comp questions never stop. How should we price this role? Where does it fit? What are these skills worth in the market?

Job descriptions were “creatively written,” packed with emerging skills and hybrid responsibilities that didn’t map neatly to the company’s job architecture.

“Sometimes a JD wouldn’t fit the mold at all,” Ardavan says. “I had to figure out which skills mattered and whether the market paid a premium for them.”

Ardavan fields a steady stream of questions from analysts and business partners, and nearly all of them come down to translating a job description into T-Mobile’s structure and understanding what the skills inside that description are worth in the market.

This translation step often took hours and the answers were hard to defend.

Ardavan started with a generic GPT tool to extract skills, then moved into manual research. Googling unfamiliar terms, scanning job postings, comparing roles across the industry, and piecing together insights from sources that didn’t line up.

The output worked in a pinch, but it wasn’t something he could stand behind.

“There was always noise in the data,” he says. “I knew the methodology wasn’t strong enough for high-stakes decisions.”

Everything changed when the comp team adopted Analyst Agent.

Now Ardavan drops a job description into the agent and gets a clear readout in seconds to share with business partners, eliminating more than 25 hours of manual research each month.

“I finally have a place to start,” he says. “And I trust the data because it’s built on the same Compa dataset our team already relies on.”

Instead of patching together insights, Ardavan responds quickly with answers he can defend.

Analyst Agent eliminates manual research, improves pricing and leveling decisions, and delivers market answers that hold up with Finance, business partners, and the board.

“The difference is trust,” Ardavan says. “Analyst Agent uses the same Compa data our team already relies on. When I respond to the business, I know the answer is grounded in data, not guesswork.”

And that trust changes how the team operates.

“Having an agent built on Compa’s data gives me confidence when questions come in,” Ardavan says. “It’s what we all hoped agents would deliver: better, faster, more precise answers we can use for high-stakes decisions. We’re no longer reactive; we’re empowered.”