MassPermits Get leads →
HomeResearch › Methodology

Methodology

MassPermits publishes analysis without a byline, so the only thing worth trusting is the method. This page describes it in enough detail that a stranger could re-run any figure on the site and get the same answer.

What this dataset is

37,361 building-permit records from 31 Massachusetts municipal sources, each one publishing its own records. The dataset behind this build was assembled 2026-08-18 and spans 2026-02-19 to 2026-08-18.

Where each row comes from

Every record originates with the municipality that issued it, published by that town in its own format - spreadsheets, monthly PDFs, and open-data portals. The full list, with record counts and date spans, is on the coverage page.

What we do not have

20 of 31 sources publish no declared project value. 11 publish a category label instead of a description of the work. 2 publish coordinates. Each absence rules out a class of question, and the affected sources are excluded from those questions by name rather than silently averaged in.

How work is classified

The trade on every page of this site - Roofing, HVAC, Electrical and the rest - is derived by us from what the town published, not a field the town provides. Where a town publishes a category label we use it; otherwise the description text is matched against rules. Classification error is therefore our error, not the municipality's, and each research article prints the exact rule it used.

The four rules every research page follows

Only within-source ratios are compared across sources.

Coverage differs enormously from town to town, so comparing raw counts between towns measures our pipeline rather than Massachusetts.

A source is excluded from any question its own records cannot answer.

Some towns file a whole category of work under a fixed label instead of a description. Searching their text finds one side of a comparison and not the other, which produces a confident and completely false number. Those sources are named and excluded, not ranked.

No article publishes an owner name, an unmasked house number, or any named individual.

The contractor column is never article subject matter at any level of aggregation, because most of its values are private people.

Every figure is recomputed at build time from the live data.

If a figure moves outside the range its sentence was written for, the page stops rendering rather than publish a claim its own data no longer supports.

Privacy

Public pages and feeds mask the house number of every address and never publish an owner name. The contractor field is withheld from every public page - most of its values are private individuals rather than firms, so it is treated as personal data and is never the subject of an article. Full details are supplied only to paying subscribers, under the terms they agree to.

Corrections

Corrections are appended, never silently edited, and a corrected article keeps its URL. The corrections page also records analyses that were completed and then rejected before publication, with the reason.

Licence

Figures and tables on this site are published under CC BY-NC 4.0. Attribution to MassPermits, non-commercial reuse.

How to check us

Each research article links a JSON file containing every number on the page plus the rules and thresholds used to produce it. Beyond that: the open-data catalog, the JSON feed, and RSS, with per-town CSV, JSON and XML linked from every town page.

Get the full Massachusetts permit list

Exact addresses, contractor & owner details, and a sortable file — plus a free sample so you can see the quality first.

See a free sample →

One-time lead pack $49 · Weekly feed $99/mo