How a Tax Firm's BD Team Won 4 Clients in Its First Month

The client is a national property-tax consulting firm with about 100 consultants and close to 1,000 clients, one of the largest professional-services rollups in tax consultancy, backed by a Dallas-based middle-market private equity fund. It finds new business in public appraisal rolls, and in Texas those rolls are spread across 254 county appraisal districts, each in its own format. Only four publish electronically. The BD team screened one county at a time against a single-county value threshold. The firm's floor is roughly $5M in business personal property, so a company holding $800K in each of ten counties never showed up anywhere, even though it was an $8M account. Consultant-of-record data is partly redacted, so there was also no clean way to tell which large accounts were unrepresented or had recently switched advisors. The sponsor's first attempt at a fix, running county rollups through ChatGPT by hand, timed out at scale.

Ingestion pipelines for the four counties that publish electronically (Harris, Dallas, Travis and Hockley), rebuilding each certified roll from the district's exports and normalizing every county into one schema. Open-records petitions, filed by ModernEdge, extended coverage to about 80 public counties. An entity-resolution engine on Claude Sonnet that links company and advisor-firm name variants across counties, so the same taxpayer appears as one account. It runs above 95% precision on the top 100 companies and advisor firms, and anything below the confidence threshold goes to a review queue that the firm's staff clears by hand. A ranked prospect list that flags likely-unrepresented and recently-switched accounts. Clean CSV exports go to the BD team, and a person decides what enters HubSpot.

The work ran as a co-build with the sponsor partner, the firm's CFO, the BD lead and the firm's tech lead, with written progress on Mondays and Wednesdays and a working session every Friday. The team started with the four counties that publish electronically, because the other 250 require an open-records petition with a ten-business-day response window each. ModernEdge filed those petitions itself, so the firm never set a precedent with a county it might later appeal against. Ingestion pipelines rebuilt each roll and normalized it into one schema, then the entity-resolution layer on Claude Sonnet collapsed name variants into single accounts. The four-county pilot alone covered several hundred thousand companies, and tens of thousands of them appeared in more than one county. A review queue sits between the matches and the BD team, so every account is checked before outreach. The pilot was scoped at four to six weeks and BD was working live lists about two months after kickoff. The team then extended by a month for the records petitions and statewide enrichment, added 240 hours, and kept the price where it was.

Firms whose prospects only show their true size once records are joined across jurisdictions: property-tax consultants, lien and title services, utility auditors and government-contract sales teams. It pays when that join is the whole competitive edge. If the records you prospect from are already clean and in one place, an off-the-shelf data vendor such as ZoomInfo or Apollo is the better buy. It also needs a human review step between the data and the outreach, because a rep calling the wrong company after two names collide is worse than having no system.