FAQs - Builty

What is a typical process of building permit filing?

• The permit filing process begins with contacting the local building department to determine the specific requirements for the project. The department will provide a list of necessary documents, which often include construction plans, property surveys, and proof of ownership. After gathering these documents, the applicant must submit the completed application along with any associated fees. The department will then schedule inspections throughout the construction phase to ensure compliance with building codes. Finally, upon successful completion of all inspections, final approval will be granted, allowing the project to be finished.

What is Builty's coverage of permit data? How representative is it of the entire market?

• Builty covers 70% of the top 30 MSAs across the country. The best coverage reflects the most active real estate markets, both urban and suburban, residential and commercial.

How is the data sourced?

• Building permits are issued by over 20,000 local jurisdictions. Builty has developed a system that automatically searches the websites of various cities and counties, downloads the building permits, and extracts useful information from them. It consists of hundreds of automated scrapers that can harvest data from public sources that are not readily accessible by most downloading tools.

How is the data standardized?

• Builty has developed multi-step data processing and machine learning pipelines that classify and deconvolute information otherwise confined within disordered text. The culmination of these efforts is a data product that offers an ordered taxonomy of permit categories, and a standardized repository of a vast multitude of attribute types.

How are Builty category flags populated?

• Oftentimes, building permits are not properly labeled by the original applicantsL for example, a project labeled as re-roofing also includes solar panel installation and electrical system upgrade. To provide more utility, Builty has developed language models in an attempt to correctly classify each permit. The algorithm is trained to generate a TRUE/FALSE outcome across many categories. In cases where the model confidence is below a certain threshold, those categories remain NULL. Due to the nature of construction projects with multiple work types, the categories are not mutually exclusive.

How are permit applications from offline jurisdictions recorded?

• Builty does not currently collect data from jurisdictions that do not offer digital access to permitting systems.

How are permit applications from recorded for online jurisdictions prior to the implementation of a digital platform handled?

• Builty collects records as they are made available through the jurisdiction's online systems. Older records, sometimes labeled “history” or “legacy,” may be included if imported by the jurisdiction, but they do not rely on those as they often lack consistency.

Is there any way to identify or infer rejected permit applications?

• The status_original field is your best resource. However, its values and granularity differ by jurisdiction.

Do the inactive or expired status values include rejected applications?

• Possibly, but this varies. If status_original doesn’t explicitly say “rejected,” then that status is likely not included.

Could permits with missing permit dates indicate rejections, or are they likely just missing data?

• Not reliably. Missing dates can result from incomplete uploads, data entry errors, or system limitations - not necessarily rejections.

Is all content in the DATA column raw, directly web-scraped from each jurisdiction’s online records?

• Yes - it reflects raw online content. No extra metadata is added to it.

Are other fields in the BUILTY database (e.g., CBSA, state, etc) extracted from this column using machine learning or a sort of automated parsing?

• Yes, structured fields may be derived using various automated processes.

When analyzing permit timelines across jurisdictions, which date field should I use—and how should I handle missing file_date, permit_date, and final_date values?

When analyzing dates, it is recommended that you use a combination of file_date, permit_date, and final_date. This is because not every jurisdiction or every permit will have all three dates populated. To get some sense of time, you can coalesce these three fields. There are even cases where all dates are missing, which typically applies to legacy permits and represents a very low count.