Method & provenance

where each field came from, and what it does not tell you · from the workbook Method tab

Provenance on hoverfive levels · nothing is marked in the resting state

Level	What it means	Examples	Sample (hover it)
Public record	Assessor, Registry of Deeds or Corporations Division. Recorded fact.	acquisition date, owner of record, deed book/page, mailing address, principals	Acquired 4 Dec 1998
Source export	The commercial data export. Recorded, second-hand.	RBA, year built, clear height, docks, % leased	300,000 SF
Derived	Computed by a stated rule; the rule is in the hover.	building character, clear height band, racking levels, mill building, SF per business, occupancy, who to call	small-bay / flex
Unverified lead	A Google Business listing or an Apollo stub. A lead, not a fact.	every tenancy row until a user verifies it; decision-maker titles	Stryker Corporation
User entered	Someone at K&S typed it. Named and dated.	notes, tasks, stage, verified tenancies, occupied SF	Verified · Eric R.

Level

What it means

Examples

Sample (hover it)

Public record

Assessor, Registry of Deeds or Corporations Division. Recorded fact.

acquisition date, owner of record, deed book/page, mailing address, principals

Acquired 4 Dec 1998

Source export

The commercial data export. Recorded, second-hand.

RBA, year built, clear height, docks, % leased

300,000 SF

Derived

Computed by a stated rule; the rule is in the hover.

building character, clear height band, racking levels, mill building, SF per business, occupancy, who to call

small-bay / flex

Unverified lead

A Google Business listing or an Apollo stub. A lead, not a fact.

every tenancy row until a user verifies it; decision-maker titles

Stryker Corporation

User entered

Someone at K&S typed it. Named and dated.

notes, tasks, stage, verified tenancies, occupied SF

Verified · Eric R.

Known tradeoff, accepted: hover does not survive a screenshot. A screenshotted tenant panel gives no signal that the businesses are unverified leads. Weighed against on-page clutter; clutter lost.

Absent datathe reason, never a zero

export field blanknot recorded

assessor has no ownernot on assessor

no transfer in registryno transfer on record

rule input missingnot derivable — no clear height

height fails sanity checkwithheld — suspect height

no Google listingno listing at this address

Fieldssource and caveat per field

Field	Source	Watch out for
Property address, city, RBA, year built	Public-records export supplied by Eric: three geographic pulls combined and deduplicated on PropertyID.	All three pulls floor at 50,000 SF. The 10–25k and 25–50k bands are not in this data.
Clear height, loading docks, drive-ins, column spacing, power, sprinklers, construction, cranes, rail, heating, parking, office SF	Same public-records export. Clear height parsed from "36'0\"" to a sortable number; dock and door counts parsed from free text.	Coverage varies: clear height 70%, docks 79%, drive-ins 65%, construction 78%, power 53%, column spacing 30%. A blank means not recorded, NOT zero.
Clear height (ft), clear height band, racking levels	Parsed from the export's Ceiling Ht field ("32'0\"" becomes 32) and banded on the standard industrial breaks. 389 of 558 recorded (70%), rising to 78% on buildings over 250,000 SF. Racking levels estimates (clear height − 3 ft sprinkler allowance) ÷ 5 ft per pallet position.	Two cautions. First, the source field is labelled CEILING height, not clear height. True clear height, to the lowest joist or sprinkler, is typically 1–3 ft lower, so treat these as the optimistic end. Second, eight readings are not clear heights at all (75 ft on a 2025 Auburn build, 60 ft on a two-storey in Holliston, 8 ft on an 1892 Millbury mill): they are building heights or crane bays. Those are flagged suspect and excluded from the band, but the raw number is left visible. Racking levels is an ESTIMATE from height alone; real capacity depends on the sprinkler design and column grid.
Below modern spec	Derived: clear height under 24 ft. Modern distribution wants 32 ft+; below roughly 24 ft a building is functionally dated for that use.	Blank where clear height is unknown OR flagged suspect; do not read blank as "fine". And do not read "yes" as a defect on a mill: a 14 ft mill floor is a different product, not a failed warehouse. Cross-reference the mill flag and building character before treating low clear height as obsolescence: 27 of the 108 buildings under 18 ft are mills, where low clear height is simply what the type is.
Mill building (inferred)	DERIVED from dimensions, not from year built: three or more storeys with clear height of 18 ft or less. By 1960 industrial was single-storey with 20–24 ft clear because forklifts and racking made vertical mills obsolete, so a tall narrow-ceilinged stack is a mill whatever date the assessor recorded. 28 buildings qualify; median recorded year 1910 against 1972 for the rest.	USE THIS INSTEAD OF YEAR BUILT ON OLD STOCK. The assessor's year built records major renovations: 2 Shaker Rd in Shirley reads 1970 but is the 19th-century Phoenix Cotton mill, later Samson Cordage rope works, now Phoenix Park. Only 3 of these 28 were caught by the year-built test alone, which is why they now reach the Adaptive Reuse view by shape as well.
Use type	MassGIS assessor use classification, collapsed into broker categories. NOT from the export: Property Type there reads "Industrial" for all 558 rows with no subtype.	466 of 558 classified; 92 have no parcel match. "Flex" does not exist here: it is a marketing term, not an assessor category.
Occupancy (inferred)	DERIVED, not sourced. Owner-occupied (likely) = the owner is an operating business or a named person AND the building is 100% leased with no availability. Tenant unknown = fully leased but held by a shell entity.	This is an inference about occupancy structure. It is NOT a tenant name. There are no tenant names anywhere in the export.
Businesses at address, business types, tenants	Google Places, September 2026, two independent searches per address, unioned. One geocodes the address and asks what businesses sit within 600 m of that point; the other searches Google's business index for the address by name. Every result from both is discarded unless its street number and street name match the address searched.	SHARING AN ADDRESS IS NOT PROOF OF OCCUPANCY: a business on the same parcel appears here too. 114 addresses (21%) have no listing at all, and that gap is not random: it falls on quiet owner-occupiers with no retail trade, who are exactly the profile this list is built to find. This is a lead, not a rent roll. Verify before quoting it to anyone.
Building character (inferred), SF per business, tenant mix	DERIVED from the tenant list, not sourced. SF per business = RBA divided by the number of businesses found: the best available proxy for unit size. Tenant mix counts how many occupiers are industrial, trades, non-industrial or office. A building where a third or more of the typed occupiers are non-industrial (dance studios, clinics, churches) is called converted; four or more occupiers averaging under 15,000 SF is small-bay.	THIS IS AN INFERENCE ABOUT CHARACTER, NOT A VERIFIED FACT, and it says something different from the assessor's use type, which records the PERMITTED use and lags reality by years. Where they disagree, that disagreement is the interesting part, not an error. Two known limits: one company with several Google pins (Medline lists three) inflates the count, which is why any building above 150,000 SF per listing is forced back to single-occupier; and 114 buildings have no listing, so the column is deliberately BLANK for them rather than guessed.
Occupancy evidence	Derived by comparing each business Google names against the owner of record. "Owner confirmed on site" means the two names match; "occupant identified" means Google names somebody the assessor did not.	"Owner confirmed" is a name match, not a lease. "No Google listing" means nothing was found, not that the building is empty.
Acquisition date, price, deed book/page	MassGIS Standardized Assessors' Parcels, July 2026 vintage, joined on address within municipality.	This is the last RECORDED TRANSFER, not necessarily a sale. 99 rows are nominal $0–$100 transfers and are flagged Unusable.
Owner name (export)	The public-records export. Usually the beneficial parent, e.g. "The Kraft Group".	Not the entity on the deed.
Owner of record (assessor)	MassGIS parcel data. The entity actually holding title, e.g. "TKG 112 Grove Street LLC".	Single-purpose LLCs tell you nothing about who to call; hence the next two fields.
Registered entity, MA entity ID, resident agent, named principals	Massachusetts Secretary of the Commonwealth, Corporations Division. All 250 owner entities looked up by hand.	35 entities returned no people. 11 have no record at all. 4 hit a state-side bug where the result link opens a different company.
Who to call	Derived: real-property signatories first, then managers, then officers. Falls back to the owner name when it is already a person or a callable business.	A name, not a phone number. No contact details are in this system.
Apollo reachable / decision-makers	Apollo People Search, September 2026. Zero-credit search only: it confirms a verified email EXISTS, it does not retrieve it. Decision-maker stubs carry a first name, an obfuscated surname and a full title.	"No" often means the assessor's company name is too mangled to match, not that Apollo lacks the data. Owner-side reachability from the workbook was not carried into the CRM export; owners read as untested here.
Owner mailing city / state	MassGIS parcel data: the address the tax bill goes to.	Often the most useful contact channel in this file, and it cost nothing.
Confidence (parcel match)	Verified = exact address match and the parcel area reconciles against RBA. Sourced = matched but not size-checkable. Unverified = size mismatch. Unusable = nominal transfer. Unknown = no parcel match.	Treat anything below Verified as needing a Registry check before it goes in front of a client.
Years held	2026 minus the acquisition year.	If the LLC changed hands rather than the property, no deed records and tenure reads long on a new relationship. Public record cannot see this.

Field

Source

Watch out for

Property address, city, RBA, year built

Public-records export supplied by Eric: three geographic pulls combined and deduplicated on PropertyID.

All three pulls floor at 50,000 SF. The 10–25k and 25–50k bands are not in this data.

Clear height, loading docks, drive-ins, column spacing, power, sprinklers, construction, cranes, rail, heating, parking, office SF

Same public-records export. Clear height parsed from "36'0\"" to a sortable number; dock and door counts parsed from free text.

Coverage varies: clear height 70%, docks 79%, drive-ins 65%, construction 78%, power 53%, column spacing 30%. A blank means not recorded, NOT zero.

Clear height (ft), clear height band, racking levels

Parsed from the export's Ceiling Ht field ("32'0\"" becomes 32) and banded on the standard industrial breaks. 389 of 558 recorded (70%), rising to 78% on buildings over 250,000 SF. Racking levels estimates (clear height − 3 ft sprinkler allowance) ÷ 5 ft per pallet position.

Two cautions. First, the source field is labelled CEILING height, not clear height. True clear height, to the lowest joist or sprinkler, is typically 1–3 ft lower, so treat these as the optimistic end. Second, eight readings are not clear heights at all (75 ft on a 2025 Auburn build, 60 ft on a two-storey in Holliston, 8 ft on an 1892 Millbury mill): they are building heights or crane bays. Those are flagged suspect and excluded from the band, but the raw number is left visible. Racking levels is an ESTIMATE from height alone; real capacity depends on the sprinkler design and column grid.

Below modern spec

Derived: clear height under 24 ft. Modern distribution wants 32 ft+; below roughly 24 ft a building is functionally dated for that use.

Blank where clear height is unknown OR flagged suspect; do not read blank as "fine". And do not read "yes" as a defect on a mill: a 14 ft mill floor is a different product, not a failed warehouse. Cross-reference the mill flag and building character before treating low clear height as obsolescence: 27 of the 108 buildings under 18 ft are mills, where low clear height is simply what the type is.

Mill building (inferred)

DERIVED from dimensions, not from year built: three or more storeys with clear height of 18 ft or less. By 1960 industrial was single-storey with 20–24 ft clear because forklifts and racking made vertical mills obsolete, so a tall narrow-ceilinged stack is a mill whatever date the assessor recorded. 28 buildings qualify; median recorded year 1910 against 1972 for the rest.

USE THIS INSTEAD OF YEAR BUILT ON OLD STOCK. The assessor's year built records major renovations: 2 Shaker Rd in Shirley reads 1970 but is the 19th-century Phoenix Cotton mill, later Samson Cordage rope works, now Phoenix Park. Only 3 of these 28 were caught by the year-built test alone, which is why they now reach the Adaptive Reuse view by shape as well.

Use type

MassGIS assessor use classification, collapsed into broker categories. NOT from the export: Property Type there reads "Industrial" for all 558 rows with no subtype.

466 of 558 classified; 92 have no parcel match. "Flex" does not exist here: it is a marketing term, not an assessor category.

Occupancy (inferred)

DERIVED, not sourced. Owner-occupied (likely) = the owner is an operating business or a named person AND the building is 100% leased with no availability. Tenant unknown = fully leased but held by a shell entity.

This is an inference about occupancy structure. It is NOT a tenant name. There are no tenant names anywhere in the export.

Businesses at address, business types, tenants

Google Places, September 2026, two independent searches per address, unioned. One geocodes the address and asks what businesses sit within 600 m of that point; the other searches Google's business index for the address by name. Every result from both is discarded unless its street number and street name match the address searched.

SHARING AN ADDRESS IS NOT PROOF OF OCCUPANCY: a business on the same parcel appears here too. 114 addresses (21%) have no listing at all, and that gap is not random: it falls on quiet owner-occupiers with no retail trade, who are exactly the profile this list is built to find. This is a lead, not a rent roll. Verify before quoting it to anyone.

Building character (inferred), SF per business, tenant mix

DERIVED from the tenant list, not sourced. SF per business = RBA divided by the number of businesses found: the best available proxy for unit size. Tenant mix counts how many occupiers are industrial, trades, non-industrial or office. A building where a third or more of the typed occupiers are non-industrial (dance studios, clinics, churches) is called converted; four or more occupiers averaging under 15,000 SF is small-bay.

THIS IS AN INFERENCE ABOUT CHARACTER, NOT A VERIFIED FACT, and it says something different from the assessor's use type, which records the PERMITTED use and lags reality by years. Where they disagree, that disagreement is the interesting part, not an error. Two known limits: one company with several Google pins (Medline lists three) inflates the count, which is why any building above 150,000 SF per listing is forced back to single-occupier; and 114 buildings have no listing, so the column is deliberately BLANK for them rather than guessed.

Occupancy evidence

Derived by comparing each business Google names against the owner of record. "Owner confirmed on site" means the two names match; "occupant identified" means Google names somebody the assessor did not.

"Owner confirmed" is a name match, not a lease. "No Google listing" means nothing was found, not that the building is empty.

Acquisition date, price, deed book/page

MassGIS Standardized Assessors' Parcels, July 2026 vintage, joined on address within municipality.

This is the last RECORDED TRANSFER, not necessarily a sale. 99 rows are nominal $0–$100 transfers and are flagged Unusable.

Owner name (export)

The public-records export. Usually the beneficial parent, e.g. "The Kraft Group".

Not the entity on the deed.

Owner of record (assessor)

MassGIS parcel data. The entity actually holding title, e.g. "TKG 112 Grove Street LLC".

Single-purpose LLCs tell you nothing about who to call; hence the next two fields.

Registered entity, MA entity ID, resident agent, named principals

Massachusetts Secretary of the Commonwealth, Corporations Division. All 250 owner entities looked up by hand.

35 entities returned no people. 11 have no record at all. 4 hit a state-side bug where the result link opens a different company.

Who to call

Derived: real-property signatories first, then managers, then officers. Falls back to the owner name when it is already a person or a callable business.

A name, not a phone number. No contact details are in this system.

Apollo reachable / decision-makers

Apollo People Search, September 2026. Zero-credit search only: it confirms a verified email EXISTS, it does not retrieve it. Decision-maker stubs carry a first name, an obfuscated surname and a full title.

"No" often means the assessor's company name is too mangled to match, not that Apollo lacks the data. Owner-side reachability from the workbook was not carried into the CRM export; owners read as untested here.

Owner mailing city / state

MassGIS parcel data: the address the tax bill goes to.

Often the most useful contact channel in this file, and it cost nothing.

Confidence (parcel match)

Verified = exact address match and the parcel area reconciles against RBA. Sourced = matched but not size-checkable. Unverified = size mismatch. Unusable = nominal transfer. Unknown = no parcel match.

Treat anything below Verified as needing a Registry check before it goes in front of a client.

Years held

2026 minus the acquisition year.

If the LLC changed hands rather than the property, no deed records and tenure reads long on a new relationship. Public record cannot see this.

Known gaps

Size bands

Only 50,000 SF and above. 10–25k and 25–50k need a fresh pull at a 10,000 SF floor. The size_bands table already holds those bands; the importer will fill them when the rows arrive.

Tenants — partial

The source export contains no tenancy field and no tenant names at all. What is here was added by a separate pass against Google Business listings, which names occupiers at roughly four addresses in five. The rest are silent, and they skew toward exactly the quiet owner-occupiers this list is built to find, because a firm with no retail trade has no reason to keep a Google profile. Where Google is silent, the occupancy column still infers STRUCTURE only: a shell entity holding a full building means somebody is in there and nothing here says who. K&S's own knowledge of who is in what will beat this column on the buildings they know.

Devens

18 buildings never matched. It is a MassDevelopment jurisdiction spanning four towns and does not behave like a municipality in parcel data.

Contact details

Names only. Revealing Apollo emails costs 1 credit each and phones 8; the buttons are present and disabled. No email or phone column exists on people or role profiles by design.

Shared addresses

558 buildings sit at 547 addresses. Ten addresses hold more than one building; every Google listing at such an address is linked to each building there, because the source cannot say which one it sits in. Those rows are surfaced, not deduplicated.