Create a Carrier-Data Field Dictionary: Definitions and Worksheet

A carrier-data field dictionary should say exactly what each field measures, which entity or observation it belongs to, how its date works and which conclusions it cannot support. Keep the source label and raw value alongside any renamed or calculated value. A column called “fleet,” “equipment” or “location” is too ambiguous to use until those meanings are resolved.

This guide gives research operations teams a reusable, ten-field reference based on public FMCSA definitions. The practical question is: can this field support the decision our team wants to make? The worksheet names below are our recommended local names, not promised USDOTData export columns or an API contract.

Define one source mapping before building a universal dictionary

Start with a named source and version. This reference uses the SAFER Company Snapshot field definitions, with the currently linked MCS-150 instructions for reporting context. A vendor export, a historical census file and a current public snapshot can use different labels or structures. Verify each mapping before treating them as equivalent.

FMCSA's Open Data Program distinguishes modern and legacy datasets and describes different row types, including carrier, inspection and violation records. Its dataset guidance also treats the files as snapshots. A dictionary therefore needs the dataset name and record grain, not just a familiar column heading.

Record grain means what one row represents. In the working dictionary below, a value belongs to one identified carrier/entity observation. Save the USDOT identifier and observation time with it. If you expand one carrier's cargo choices into three rows, you have three category observations, not three carriers.

Use these shared rules for every entry:

• Preserve the exact source label, original value and source URL or file reference.
• Store your observation timestamp separately from displayed form, snapshot or field dates. Keep unavailable dates unknown.
• Describe units explicitly: an identifier, text label, address, category, count, calendar date or miles.
• Put normalization and calculations in separate working fields. Keep the rule and dictionary version that produced them.
• Record a missing-value state instead of making one blank stand for several different conditions.

These are editorial research controls. They do not change the meaning of the source or certify a carrier's operations.

A ten-field carrier dictionary you can adapt

The source labels in this section follow SAFER's published field definitions. The suggested local names make those meanings visible in your own worksheet. Each entry inherits the shared entity-observation grain and separate observation time described above.

1. USDOT# → source_usdot_id

Meaning and unit: the identifier assigned to the census record; an identifier, not a measured quantity.

Working rule: keep it as text. Retain the raw value before trimming accidental outer whitespace. Resolve an absent or malformed identifier before joining observations. Do not substitute a company name as if it were the same key.

Date and null rule: record when the identifier was observed; that is not its assignment date. A missing identifier means the intended entity match is unresolved.

Do not infer: authority status, business age or ownership relationships from the number itself. A successful identifier match also does not establish service fit.

2. Legal Name → source_legal_name

Meaning and unit: the company's legal name; text.

Working rule: preserve spelling, punctuation and legal suffixes in the raw field. A separate search-friendly form can standardize case and spaces, but it must remain a matching aid. Do not let it overwrite the official wording you observed.

Date and null rule: retain the observation and any supplied source context. Do not label that context as a name-change date without specific evidence. A blank is unavailable name evidence.

Do not infer: two similarly normalized names are the same entity. Joining on a shortened name can silently combine unrelated businesses.

3. DBA Name → source_dba_name

Meaning and unit: a trade or other identifying name distinct from the legal name; text.

Working rule: keep it separate from the legal name. If your CRM displays a convenient account name, document which source field supplied it and retain both underlying values.

Date and null rule: a blank means no DBA value was observed in this source. It does not prove that the business uses no other names anywhere. Do not manufacture a DBA by copying the legal name into the empty field.

Do not infer: a DBA is a second carrier, a parent company or an independently approved transport provider.

4. Physical Address → source_principal_office_address

Meaning and unit: the location of the company's principal office; a structured address.

Working rule: retain the original address and separate its components only when the parsing is reliable. Keep postal codes as text. If you add coordinates or distance, name the geocoding source, match quality and calculation separately.

Date and null rule: keep the observed address and source context without inventing a move date. An incomplete address remains incomplete; do not fill it from a mailing address without evidence.

Do not infer: a repair yard, truck parking location, dispatch area or service radius. The MCS-150 instructions describe the principal place of business in relation to transport business operations and maintained safety records. They do not supply an inventory of every operating site.

5. Mailing Address → source_mailing_address

Meaning and unit: where the company receives mail; a structured address.

Working rule: preserve it as a separate address purpose, even if its text matches the physical address. If the original form explicitly marks it as the same address, preserve that evidence before resolving the reference in your working copy.

Date and null rule: use “same as physical, explicitly stated” only when the source supports it. A blank exported mailing field is not, by itself, that instruction.

Do not infer: trucks, employees or purchasable service operate at the mailing location. Address equality also does not turn either field into proof of a maintenance site.

6. Carrier Operations → source_carrier_operations

Meaning and unit: the recorded interstate, intrastate hazardous-material or intrastate non-hazardous-material transport category; categorical data.

Working rule: preserve the precise label or documented code. Keep this field separate from Operation Classification and Operating Status. If a provider supplies just “type” or “active,” the mapping is unresolved until its definition is known.

Date and null rule: preserve the observed category and dated source context. Missing category data is unknown scope, not an assumed interstate operation.

Do not infer: current authority for a particular service, lanes served, nationwide coverage or availability for a proposed shipment.

7. Power Units → reported_power_units

Meaning and unit: a count of the specified powered vehicle categories, including owned, term-leased and trip-leased vehicles. SAFER's definition includes trucks, tractors, hazardous-material tank trucks and passenger-vehicle categories.

Working rule: parse a valid count as a nonnegative integer while retaining the raw value. A separately named size band can support account research; record its boundaries. Keep trailers, vehicle ownership and equipment specifications in their own evidence fields.

Date and null rule: attach the displayed reporting context without claiming a field-specific change date. Distinguish an explicit zero from an unavailable count. Neither should disappear into a default “small fleet” bucket.

Do not infer: owned trucks, available vehicles, software seats or a purchase budget. The power-unit prospecting guide develops those business limits.

8. Cargo Carried → reported_cargo_categories

Meaning and unit: the recorded cargo types; a set of category labels rather than a quantity.

Working rule: preserve all selected labels and any available Other description. Sorting the labels for comparison is reasonable; translating Refrigerated Food into “reefer fleet” changes the claim. The MCS-150 cargo instructions, item 24 permit multiple categories and a description for Other.

Date and null rule: SAFER's help defines an X in its tables as applicable and an unmarked item as not applicable. That applies to an intact, understood source table. An absent export column or a table that failed to load is unknown, not an observed unchecked category. Neither state supplies a date when a commodity was last hauled.

Do not infer: trailer count, revenue shares, current contracts or equipment availability. See the cargo-category research worksheet for the follow-up questions.

9. MCS-150 Form Date → source_mcs150_form_date

Meaning and unit: the date from the MCS-150 registration form; a calendar date.

Working rule: preserve the source text and normalize to YYYY-MM-DD only when the source format is understood. Keep a date-only value as a date; adding midnight UTC creates time precision the source did not provide.

Date and null rule: this is a form date. It is not your retrieval time, a truck-acquisition date or a universal effective date for every field on the page. An unparseable date belongs in an exception queue; a missing date stays unknown.

Do not infer: every displayed fact was newly verified on that day. The dated-record comparison guide explains how to retain competing observations.

10. MCS-150 Form Mileage → reported_mcs150_mileage

Meaning and unit: mileage reported on the MCS-150; miles. Item 21 of the current instructions calls for total commercial-motor-vehicle mileage for the prior 12 months, rounded to the nearest 10,000, with instructions for shorter operation and no operation.

Working rule: preserve the mileage and any displayed year or period separately. If a page labels the pair “MCS-150 Mileage (Year),” keep that exact label in your mapping rather than silently dropping the parenthesis. Do not turn a year alone into assumed exact start and end dates.

Date and null rule: mileage needs its reporting-period context. The form date and observation date are different facts. Preserve a reported zero, a missing value and a missing period as distinct states.

Do not infer: current-month utilization, lane mileage or today's fuel demand. Dividing historical mileage by today's power-unit count can mix incompatible periods and scopes.

Use explicit missing-value states

Your research worksheet can use these local states. They are workflow labels, not claimed FMCSA codes:

• Present: the source supplied a usable value, including an explicit numerical zero.
• Not displayed: the source observation contains no value for the field.
• Not supplied by provider: the available export or response omits the field.
• Source unavailable: the relevant record could not be inspected.
• Unparseable: raw text exists, but its structure cannot yet be interpreted reliably.
• Explicit source state: a documented source convention applies, such as an unmarked category in a complete SAFER cargo table.

Keep the source's literal value alongside your local state. “Not displayed” describes your evidence, while “none exists” makes a claim about the business. Those statements are not interchangeable.

Worked example: prevent three plausible but unsupported columns

Fictional example; no actual carrier record or product export is represented. A researcher observes the following on October 4, 2026: Power Units is 12, Cargo Carried includes Refrigerated Food, and Physical Address gives a principal office in Columbus, Ohio. The displayed MCS-150 Form Date is August 20, 2026. No equipment inventory or maintenance location has been confirmed.

A colleague proposes three CRM fields: fleet_owned = 12, equipment = 12 reefers and service_yard = Columbus. Every value looks useful to a refrigeration-service seller, but each changes the meaning of the evidence.

Apply the dictionary instead:

1. Keep reported_power_units = 12, with the source definition. Leave owned vehicles unknown.
2. Keep reported_cargo_categories = Refrigerated Food. Leave refrigeration-equipment type and count unknown.
3. Keep source_principal_office_address = observed Columbus address. Leave maintenance location unknown.
4. Preserve the August 20 form date and October 4 observation date in separate fields. Do not describe either as the date twelve vehicles began operating in Columbus.

The observed facts support a research candidate. The fit hypothesis is that this carrier may operate equipment the seller can service in its territory. Confirmed buying need remains absent. The next useful step is to confirm compatible equipment, its service location and responsibility for maintenance before making a proposal.

This is a semantic correction, not merely a completeness check. All three original CRM cells could be populated, correctly formatted and still make unsupported claims.

Copyable field-definition worksheet

Create one block for each source-to-working-field mapping. Reuse that definition across records; store individual carrier observations separately.

DICTIONARY ENTRY
Dictionary version / review owner:
Source system / dataset / version:
Source URL / documentation reviewed on:
Exact source field label or key:
Definition and source section:
Record grain / entity key:
Unit / value type / allowed categories:
Source date meaning / period meaning:
Missing and explicit-zero conventions:

WORKING MAPPING
Local field name:
Raw value retained where:
Allowed transformation / rule version:
Example raw value -> working value:
Forbidden inference:
Exception handling / review trigger:

Keep a short approval checklist with the dictionary:

1. Can a second reviewer locate the exact source definition?
2. Is the entity and row grain explicit?
3. Are units and dates still attached after import?
4. Can the working value be reproduced from the retained raw value?
5. Do missing values, observed zero and documented source conventions remain distinguishable?
6. Does the field name avoid implying an unverified business fact?

Review the mapping when a provider changes a header, definition, unit, null convention or source. A rename that preserves meaning can be a presentation change. A change from all reported power units to owned tractors is a new definition requiring review of affected filters and calculations. Preserve the earlier dictionary version so old research remains interpretable.

Frequently asked questions

Is the same label the same field across providers?

Not necessarily. “Fleet size” might hide different vehicle categories or ownership scope. Match the source definition, units, dates and grain before mapping two labels to one working field. Mark an unexplained label unresolved.

Where should transformations such as size bands go?

In separate derived fields with a documented rule. Preserve the underlying count, boundary choices and rule version. A colleague should be able to reproduce the band without guessing which source or definition you used.

How should we represent an unavailable source date?

Keep the observation timestamp and explicitly mark the source date unavailable. Do not substitute the export date or infer a filing date from a filename. Limit any conclusion that depends on knowing when the underlying value applied.

What counts as a unit for a text or categorical field?

Describe its value type and scope: legal-name text, postal address or a set of reported cargo categories. “No numerical unit” is still useful documentation. It prevents labels from being treated as quantities or equipment inventories.

Can we replace blanks with zero to simplify reporting?

Only when the exact source specification establishes that meaning for that field and situation. A missing count usually requires its own state. Report unknowns separately; otherwise calculations can turn missing evidence into an apparent zero-sized fleet.

Who should approve a dictionary revision?

Assign a data owner who can verify the source mapping and a user who understands the decision it supports. For a changed meaning, review affected imports, filters and calculations before adopting the revision. This is a recommended team control, not a mandated FMCSA approval process.

Put the dictionary to work

Choose the ten fields your next research decision actually uses and complete one mapping for each. Resolve unclear labels before importing more records. Then apply the export-evaluation guide to assess a particular file using those shared definitions.

By John Hauler, USDOTData's editorial persona. Sources reviewed October 4, 2026. USDOTData is an independent information service, not FMCSA. Local field names, workflow rules and the fictional example are editorial recommendations.