Nivara Knows Which Roofer in Dallas Is Doing Ten Jobs a Month, and That Data Is Worth a Lot

The Macro: B2B Sales Data Has a Local Business Blind Spot

ZoomInfo is a $5 billion company. Apollo, Lusha, Clearbit, and a dozen other tools all compete to tell you who the VP of Engineering is at a mid-market SaaS company. If you are selling software to other software companies, the data infrastructure is mature, competitive, and mostly good enough.

Now try to find the owner of a roofing company in Houston. You want their direct phone number, not the office line. You want to know how many jobs they did last month. You want to know if their revenue is growing or shrinking. Try that in ZoomInfo. You will get a generic business listing with a phone number that rings to a receptionist who has never heard of your product.

This is not a small market. There are roughly 33 million small businesses in the United States. The vast majority of them are local, physical businesses. Contractors, restaurants, medical practices, auto shops, salons. They buy software, they buy insurance, they buy equipment, they buy services. The companies selling to them need data, and the existing data providers largely ignore them.

The reason is straightforward. Local business data is messy. There is no LinkedIn profile for a plumber. There is no Crunchbase page for a landscaping company. The data lives in state licensing databases, Yelp reviews, permit filings, and Google Business profiles. Assembling that into something useful requires a different kind of data pipeline than scraping LinkedIn.

The Micro: Ex-Uber Engineers Who Understand Data at Scale

Tejas Agarwal and Pankaj Mishra both worked at Uber, which is relevant here. Uber’s core business depends on understanding millions of local operators (drivers) at a granular level. Tejas built fulfillment metrics for over 6 million drivers. Pankaj worked on Uber for Business and the fulfillment platform. Before Uber, Tejas was at Oracle and studied at UW-Madison. Pankaj went to IIT Kharagpur. They came through YC’s Fall 2025 batch and are based in San Francisco.

The product is built for SDRs who sell to local businesses. The pitch is simple: deep profiles on every business, with licenses, reviews, revenue estimates, employee counts, project volume, and direct contact information for the actual decision-maker. Not the office line. The owner’s mobile number or personal email.

What makes this interesting is the precision targeting. You can filter by trade, geography, revenue, and activity level. “Show me roofers doing 10 or more jobs per month in Texas” is a query that would take a human researcher hours to answer. Nivara claims to answer it in seconds using AI-powered entity resolution across multiple data sources.

The data layer pulls from state licensing databases, review platforms, and web signals. That is a defensible moat if they can keep the data fresh. Stale data is the graveyard of sales intelligence startups. If a business got its license six months ago and Nivara still shows it as active, the product fails.

I like that they are not trying to be everything. The positioning is specific: local commerce businesses, SDR workflows, precision targeting. They are not pretending to compete with ZoomInfo on enterprise data. They are building for the use case ZoomInfo ignores.

The Verdict

Nivara is going after a real gap with the right team background. Selling to local businesses is a $100 billion problem that the existing data providers have mostly punted on. If the data quality holds up, this is the kind of product that SDRs will refuse to give back after a trial.

The risks are predictable. Data freshness is everything. If the license data is six months stale, the product is useless. If the phone numbers bounce 40% of the time, SDRs will churn. The other risk is that ZoomInfo or Apollo eventually decides to build a local business layer. They have not yet, which tells me it is harder than it looks, but “hard” does not mean “impossible” for a company with billions in revenue.

Thirty days from now, I want to see close rates for SDR teams using Nivara versus their old process. That is the number that matters. Not “we have X businesses in our database.” Sixty days, I want to know the data accuracy rate on phone numbers and emails. If it is above 85%, they have something. Below 70%, they have a demo that falls apart in production. Ninety days, the question is vertical expansion. Roofers today, HVAC tomorrow, restaurants next quarter. The playbook writes itself if the underlying data engine works.