Compare Carrier-List Snapshots and Changes

To compare two trucking-company lists, first confirm that they use the same source, selection rules, row definition and entity identifier. Then separate matched entities, additions, absences and unresolved identities. A larger file does not by itself establish more newly formed carriers, fleet growth or a larger addressable market.

This guide helps sales and research operations explain why a selected carrier population changed between two saved extracts. The output is a reproducible comparison and a research queue. It does not certify carrier status, reconstruct every event between dates or imply that USDOTData supplies a historical-snapshot comparison feature.

Decide which comparison you need

There are two legitimate questions, and they require different outputs:

What changed in our working list? Compare the files actually used, even if the selection rules changed. Explain the rule change beside the result.
What changed under the same selection rule? Apply an equivalent rule to two suitable dated source populations. If the required historical source is unavailable, the comparison cannot be reconstructed from today's data alone.

For example, expanding an Ohio equipment-service prospect list to include Indiana changes the working list intentionally. The increase may be operationally useful: more accounts need location and equipment research. It is not evidence that those accounts opened this week.

Write the question before calculating percentages. A team measuring changes in its own research queue should not label the result a change in the national carrier population. For disagreements about a field within one carrier, use the separate dated-record comparison guide.

Keep a manifest for each source file

Preserve both original files and make working copies. Record the following for each snapshot:

Provider, dataset name and source version, where supplied.
Source snapshot date, extraction time and time zone, kept as separate facts.
Exact filters, including blank/default selections and any exclusions.
What one row represents: entity, contact, operating authority, location or another observation.
Entity key and the field definitions used by the filters.
File scope: complete result, limited preview, page, sample or incremental changes.
Original row count, distinct resolved entity count and unresolved rows.

A file's creation time is not necessarily the date of its source population. FMCSA's Open Data Program distinguishes dataset families, modern and historical schemas, and complete files from daily changes. It describes Company Census data as updated daily from a database that is 24 hours old, rather than real-time data. Check the documentation for the exact file being compared.

The live USDOTData trucking company database provides filters for geography, fleet size, cargo and other recorded characteristics. Keep your chosen settings with any permitted export you already use. This article's manifest and comparison fields are a local research method; they are not a promise that every field is an export column or that earlier versions of a search can be recovered.

Resolve row grain and identity before joining

If the intended comparison is one carrier per row, reduce repeated rows only after determining why they repeat. Three contact rows for one entity are not three new carriers. A carrier with several authority records can also appear several times in an authority-level file.

Preserve the raw identifier and a documented working form. Treat a USDOT number as an identifier, not a quantity. Do not fill a missing number from a similar company name without checking the match. Keep different identifiers separate while investigating relationships.

Use four preparation states:

Resolved entity: the record has an appropriate, usable entity key.
Repeated observation: another row belongs to that resolved entity at the chosen grain; retain its source relationship.
Conflicting identity: supplied identifiers or names disagree and require review.
Unresolved identity: the evidence is insufficient to assign the row safely.

Hold the last two states outside matched-entity arithmetic and report their counts. They cannot silently become additions or absences. An unresolved row in the first file might refer to an entity that appears clearly in the second.

The SAFER Company Snapshot supports individual company searches by USDOT number, MC/MX number or name. Use an appropriate official lookup to investigate a specific identity question; matching an entity still does not establish current commercial fit or buying need.

Test whether the selections are comparable

Compare the manifests before comparing entity sets. Pay attention to changes that can look small in the interface but alter membership:

One file includes two states and the other includes one.
A fleet-size boundary changes from “at least 10” to “more than 10.”
A cargo selection changes from any selected category to all selected categories.
A blank geography is excluded in one extract but retained in another.
One source includes inactive or pending entities while another does not.
A full export is compared with the first page of results.
A changed field definition is mapped to an old column label.

Do not assume a provider uses the example semantics above. Verify the actual filter behavior and documentation. The carrier-data field dictionary helps preserve definitions and missing-value rules.

Assign a comparability result: equivalent enough for the stated task; different but explainable; or unresolved. If only today's source is available, you may test both selection rules on today's source. That isolates the effect of the rules at today's snapshot; it does not reconstruct last week's missing population or prove that the rule effect was identical then.

Build the entity sets and keep meaning separate

After resolving identities, let A be the earlier selected set and B the later selected set. Give every resolved identifier one membership outcome:

Matched: present in A and B.
Added to this selection: present in B but not A.
Absent from this selection: present in A but not B.

These labels describe files. “Added” does not mean newly registered, newly authorized or newly founded. “Absent” does not mean closed, unsafe, unsubscribed or no longer operating. A filter, source coverage, correction, extraction failure or business change may explain the observation.

Check the arithmetic:

Earlier distinct entities = matched + absent
Later distinct entities   = matched + added
Net selection change      = added - absent
Either-file entities      = matched + added + absent

Retain unresolved rows as a separate count, not a fourth set whose distinct entity membership is somehow known. For matched entities, a second field-level comparison can identify changed reported values. Do not combine that field analysis with the membership count: one matched entity can have many changed fields while remaining one entity.

Worked example: a list grows from 50 to 55

Fictional, complete teaching example. The labels below are local example IDs, not real USDOT numbers or a product dataset. Both source files have one resolved entity per row, no duplicates and no unresolved identifiers. The dates are October 2 and October 9, 2026.

The earlier selection used an Ohio principal-office filter. The later selection included Ohio and Indiana. All other selection settings are assumed identical for this example. The complete entity sets are defined compactly:

A: E001 through E050
B: E001 through E045, plus E051 through E060

Every integer in each inclusive range appears once. This gives 45 matched entities, five absent entities (E046–E050) and ten additions (E051–E060). Earlier total: 45 + 5 = 50. Later total: 45 + 10 = 55. The net change is +5, or 10% of the earlier 50-entity selection. The union contains 60 entities.

Now inspect the later file's source geography. In this fictional evidence, E001–E045 and E051–E054 have Ohio principal offices; E055–E060 have Indiana principal offices. Filtering the later file back to Ohio produces 49 entities. Compared with the earlier Ohio selection, that is 45 matched, four added and five absent: a net change of minus one.

The six Indiana records explain the difference between the later 49-entity Ohio subset and the later 55-entity two-state list. They do not establish six new companies. Even the four additions to the comparable Ohio selection need investigation: a new source entry, a changed reported address, a corrected field or another inclusion change could explain them.

An accurate result statement is:

Our working list increased from 50 to 55 resolved entities after adding Indiana. Under the Ohio-only selection available in both files, membership changed from 50 to 49. Four entities entered that selection and five were absent. The underlying business-event causes have not been established.

The observed fact is selection membership at two dates. A fit hypothesis is that some added accounts may suit the service territory. Confirmed buying need is absent until separate business evidence establishes a relevant problem, timing and decision process.

This example supports no national count, market-growth claim, carrier closure finding or equipment-purchase conclusion. Its value is the explanation of a working-list change that a colleague can reproduce.

Turn exceptions into specific research tasks

Assign the next action based on the decision that depends on the difference:

For a new prospect-list member, confirm identity and the service criterion that justified inclusion.
For an absent member, check extraction completeness and filters before investigating the business.
For a matched entity with changed source values, preserve both observations and review the affected field.
For an unresolved identifier, hold the join and assign an identity review.
For a changed source schema, resolve the mapping before presenting a time-series result.

A current official lookup can help resolve a present-tense question. It cannot prove that an entity maintained the same authority or status throughout the interval between two extracts. Preserve any dated history that actually exists and leave an unobserved interval unobserved.

For a sales queue, “needs location confirmation” is a useful task. “New buyer” is an unsupported upgrade. For a shipment decision, route an unresolved status question to the responsible verification owner rather than treating the comparison sheet as booking clearance.

Copyable snapshot-comparison worksheet

Keep one batch block and one exception block per material issue. Short blocks remain readable on a phone and avoid an excessively wide spreadsheet export.

BATCH
Decision and permitted conclusion:
A source / version / snapshot / extracted at:
B source / version / snapshot / extracted at:
Row grain and entity key:
A filters / B filters / material differences:
Full files, pages, previews or deltas:
Identity and normalization rule version:
A original rows / resolved entities / unresolved rows:
B original rows / resolved entities / unresolved rows:
Matched / added / absent:
Arithmetic check and comparability result:
Unsupported claims explicitly withheld:
Reviewer / next review trigger:

EXCEPTION
Entity key or unresolved row reference:
Observed membership or field difference:
Source evidence and dates:
Possible explanation, labeled as hypothesis:
Required corroboration / owner / disposition:

Before releasing the result, have a second reader reproduce the counts from the saved inputs. Verify that the conclusion names the selected population, denominator and date pair. If the files fail a completeness check, use the export-evaluation workflow before doing more interpretation.

Frequently asked questions

Is an absent carrier the same as a closed company?

No. Absence describes a particular selection or extract. First check filters, source scope, pagination and identity matching. A business-closure claim requires separate evidence appropriate to that claim.

What if the two snapshots used different filters?

Report the actual working-list change with the filter difference disclosed. If suitable source data exists, also calculate a comparison under an equivalent rule. Do not present a reconstructed history when only today's source can be re-filtered.

Can duplicate rows inflate additions?

Yes. Contact, authority or repeated observation rows can make a file grow without adding a distinct entity. Establish the row grain, resolve identifiers and preserve duplicate relationships before computing entity membership.

Which date makes an entity “new”?

That depends on the claim. First appearance in your selection is one observation date; registration, authority and business formation are separate events. Label the event actually observed and use the appropriate dated evidence for a stronger claim.

How should rows without usable identifiers be counted?

Report them as unresolved rows outside the resolved-entity comparison. Do not assume each is a distinct addition, and do not merge them by a similar name alone. Recalculate if later identity evidence changes membership.

Can two matching snapshots prove continuous status?

No. They establish observations at two points, subject to each source's dates and limitations. Changes may have occurred between those points. Use relevant history where available and perform current official checks for decisions that require current status.

Make the next comparison easier

Save the manifest when producing the next permitted extract. A few explicit definitions now can prevent a later team from treating a filter change as a market event. Start with one bounded pair and close its identity and scope exceptions before building a longer trend series.

By John Hauler, USDOTData's editorial persona. Sources reviewed October 9, 2026. USDOTData is an independent information service, not FMCSA. The comparison method and fictional example are editorial recommendations.

Compare Carrier-List Snapshots Without Mistaking Selection for Growth