Twelve years deciding which respondent data to trust.

I'm Arnold Santiago. I've spent my whole career on the supplier side of market research: the side that finds the respondents, fields the studies, and gets blamed first when the data looks off.

That career has run the full stack. I've been the sole owner of a panel operation, personally running fraud detection and panel-health monitoring across consumer, healthcare, B2B, and international sample. I've project-managed complex quantitative work: MaxDiff, conjoint, A&U, healthcare provider and patient studies, political polling. And I've led accounts for some of the most demanding buyers in the industry, growing a declining $300K account past $1M and managing a $5M book across agencies and brands, including advising Fortune 500 insights teams.

Twelve years of that teaches you one thing above everything else: the gap between what data says and what it means. Separating reliable respondent data from noise wasn't a specialty I picked. It was the job, every day, for over a decade.

Why synthetic data

Along the way I kept hitting the same wall: low-incidence audiences and hard-to-reach segments that traditional sampling served slowly, expensively, or not at all. I once had to fill a research community for a financial services brand when every standard panel came up short. That wall is exactly the gap synthetic data promises to fill, which is why I take it seriously, and exactly why the promises deserve scrutiny.

So I started building synthetic panels myself. Not the vendor-pitch version: the version where you generate the data, test it against real respondents, and publish where it breaks. Four studies in, I've documented agreeable-bias inflation, variance collapse, and over-concentration firsthand, and built the calibration methods that reduce them. The one thing I've never found is a reason to trust any of it unchecked.

Why I'm the independent one

Parada Research doesn't sell synthetic data. That's deliberate. The moment I profit from you adopting it, my read on it is worth less. I sell the judgment: where your synthetic or boosted sample holds, where it's quietly misleading you, and which findings you can put your name on.

I'm currently pursuing an MS in Data Science, because the checking should be at least as rigorous as the generating. The name Parada means "stop" in Tagalog and Spanish: pause and pressure-test before budget rides on it. That's the whole philosophy.