Test Carrier Cargo Filters: Any, All and Exclusions

To test a carrier cargo filter, write the selection rule in plain language, define how missing evidence is handled, and compare its output with a small set of records whose expected results you have worked out independently. Selecting either of two cargo categories is different from requiring both. An exclusion must apply to the whole intended group, and unavailable cargo data must not quietly become a negative answer.

This guide helps fleet-service researchers check a compound selection rule before using it for a larger prospect list. The output is a repeatable test sheet with expected and observed results. The notation below describes an original, vendor-neutral worksheet; it is not syntax to paste into USDOTData or a claim that a particular search interface supports Boolean operators.

Start with a rule about records, not equipment or demand

Suppose a research team wants to review records marked Fresh Produce or Refrigerated Food, while excluding records marked Household Goods from this particular batch. That is a deliberately narrow selection policy. It does not establish that the excluded carriers lack relevant food operations, or that the included carriers operate equipment the seller can service.

Before testing the combination, preserve what each field actually says. FMCSA's SAFER field help describes Cargo Carried as the type of cargo transported by the carrier or shipper. Its table convention uses an X for an applicable item and no character for an item recorded as not applicable. That is the meaning of a successfully read source table. It does not define what an empty cell in a third-party export means.

For the distinction between a cargo label and a confirmed commercial capability, use the cargo-category research guide. This article addresses a narrower question: does the selection method implement the rule you actually intended?

Keep three statements separate in the test notes:

• Observed fact: a dated source marks a cargo category for a particular USDOT entity.
• Fit hypothesis: that observation makes the entity worth researching for a stated service.
• Confirmed need: separate business evidence establishes a relevant requirement, timing and decision process.

A correctly implemented filter establishes only that the record satisfies the selection rule. It does not move the record through all three stages.

Define the inputs before combining them

Use three input states in the worksheet:

• Y — marked: the chosen category is affirmatively marked in the evidence being tested.
• N — not marked: the category is visibly unmarked in a complete, successfully read source table, or a documented export encoding establishes the equivalent record state.
• ? — unknown: the relevant data was unavailable, omitted, unreadable, ambiguous or not yet checked.

N means not marked in that evidence. It does not mean the carrier cannot perform a particular service. A blank export value should remain unknown until its source and encoding are understood. A successful download containing an empty cargo field is not enough to distinguish “no selected categories” from “categories were not supplied.”

Record the entity identifier, source URL or permitted file reference, source-record date when available and observation time. FMCSA describes the MCS-150 as a way to update an existing company record. Reading a record today does not establish that its cargo observations describe today's available capacity.

For this fictional exercise, define three shorthand inputs. These letters are worksheet variables, not FMCSA category codes:

A = Fresh Produce is marked in the chosen evidence
B = Refrigerated Food is marked in the chosen evidence
X = Household Goods is marked in the chosen evidence

Preserve the actual label alongside the shorthand. If a source uses a different label or coding scheme, document the mapping rather than silently matching similar words.

Write any, all and exclusion rules separately

Any rule: A OR B. A record passes this condition when at least one category is marked. Both may be marked. If neither is marked, it fails. If neither is known to be marked and a needed input is unknown, the answer remains unknown.

All rule: A AND B. Both categories must be marked on the same carrier record. One known N is enough to fail this condition. Otherwise, an unknown input prevents a confirmed pass. Do not combine A from one USDOT entity with B from another entity under the same brand.

Exclusion rule: NOT X. Under the worksheet's definitions, Y becomes N, N becomes Y and ? stays ?. Unknown exclusion evidence is not proof that an exclusion is absent.

The example team's complete rule is:

(A OR B) AND NOT X

In words: “Select a record with either target category marked, provided the exclusion category is known not to be marked.” The parentheses group the two inclusion alternatives before the exclusion is applied.

For clear handoffs, attach a reason to the result:

1. If X is Y, mark excluded by this batch rule.
2. Otherwise, if A OR B is N, mark not selected: neither target marked.
3. Otherwise, if A OR B is Y and X is N, mark selected for research.
4. Otherwise, mark hold: decisive evidence unknown.

This order is a declared reporting policy. It keeps the primary reason consistent when several conditions fail. The prospect-exclusion policy guide covers the broader task of writing campaign-specific reasons and re-entry conditions.

Work through six fictional carrier records

All six records below are fictional teaching cases. Their labels are not real companies or USDOT numbers. Assume each input refers to the same entity and evidence version within its case. Y, N and ? have only the meanings defined above.

Case 1: only the first target is marked

A=Y, B=N, X=N. A OR B is Y; A AND B is N. The complete rule selects this record for research because one target is enough and the exclusion is known not to be marked. Requiring both targets by mistake would lose this intended selection.

Case 2: only the second target is marked

A=N, B=Y, X=N. A OR B is Y; A AND B is N. The complete rule selects this record too. Testing only Case 1 would not reveal an implementation that accidentally ignores B.

Case 3: both targets are marked

A=Y, B=Y, X=N. A OR B and A AND B are both Y. The complete rule selects the record once. A carrier matching two target categories is not two carrier entities.

Case 4: a target match also has the exclusion

A=Y, B=Y, X=Y. Both inclusion rules return Y, but the complete rule excludes this record. The exclusion applies even though both target categories match. That outcome describes the team's chosen batch scope, not an official restriction or a judgment about the carrier.

Case 5: neither target is marked

A=N, B=N, X=N. Both inclusion rules return N. The complete rule does not select this record because neither target is marked. “No exclusion found” is not a positive inclusion reason.

Case 6: cargo evidence is unavailable

A=?, B=?, X=?. Both inclusion rules and the complete rule remain unknown. Place this record on hold and identify the missing evidence. Converting all three unknowns to N would incorrectly report a completed negative review.

The six-case results reconcile as follows:

Case     ANY     ALL     Complete rule disposition
1        Y       N       selected
2        Y       N       selected
3        Y       Y       selected
4        Y       Y       excluded
5        N       N       not selected
6        ?       ?       hold

ANY: 4 yes + 1 no + 1 unknown = 6 records
ALL: 2 yes + 3 no + 1 unknown = 6 records
Complete rule: 3 selected + 1 excluded
              + 1 not selected + 1 hold = 6 records

This is a logic check, not a measured market count or a statistically representative sample. Its value is that each case exposes a different possible implementation mistake.

Catch grouping mistakes before exporting a larger list

Compare the intended rule with this different expression:

Intended: (A OR B) AND NOT X
Different: A OR (B AND NOT X)

The second expression lets every A match through, including Case 4. It applies the exclusion only to the B branch. A screenshot showing the same three selected category names would not prove the two configurations behave the same way.

When testing an actual tool, use its documented controls and record what those controls mean. If it cannot express the intended combination, keep the remaining checks in your existing authorized worksheet. Do not describe a partially filtered result as the completed selection.

If the available workflow requires separate A and B results, reconcile their entity identifiers before counting the combined set. In this exercise, A matches Cases 1, 3 and 4; B matches Cases 2, 3 and 4. Adding 3+3 produces six matching rows, but there are four distinct target-matching entities. After applying X, three remain selected. Case 6 still requires a separate unknown-evidence record; its absence from a positive search does not prove a negative result.

Compare expected results with observed results

Work out expected dispositions before looking at the tool output. Then test the available method against evidence you are permitted to use. The fictional cases above can be evaluated manually; they do not imply permission or a facility to upload synthetic carriers into a production database.

Use one short test card per case:

Rule ID / version / research purpose:
Entity or fictional case ID:
Source / evidence version / observed time:
Exact cargo labels and their Y / N / ? states:
Plain-language inclusion requirement:
Exclusion requirement and grouping:
Expected result / reason:
Actual tool or worksheet result / evidence:
Match or mismatch:
Decisive unknown / responsible reviewer:
Correction / retest result / approval to reuse:

When a result differs, first check whether the evidence and the logic are the same. A refreshed source, truncated export, renamed category, different entity scope or changed missing-value encoding can alter results without an OR/AND error. Preserve the prior rule and input version so the difference is explainable.

Do not infer correct behavior just because the total count matches. An implementation can wrongly include Case 4 and wrongly omit Case 2 while returning the expected total of three. Compare membership and reasons, then reconcile totals.

Use this checklist before expanding the batch

• The plain-language rule identifies any versus all and the entity being selected.
• Exact source labels are mapped without inventing equipment capabilities.
• Known unmarked values are distinct from missing or unreadable evidence.
• The exclusion applies to every intended inclusion branch.
• Expected results were worked out independently of the observed output.
• Single-category, overlapping-category, exclusion and unknown cases were checked.
• Overlapping matches are deduplicated by resolved entity identity.
• Membership and reasons match, as well as totals.
• The rule version, evidence dates and retest triggers are saved.
• Selected records advance to research with unresolved business questions visible.

The lead-list building guide explains how to turn that tested candidate set into a useful research shortlist. Keep its fit and need checks separate from this technical acceptance step.

Frequently asked questions

Does selecting two cargo categories mean either category or both?

That depends on the specific tool's documented behavior. Test an A-only record, a B-only record and a both-marked record. A both-marked record alone cannot distinguish any from all because it passes both rules.

Should a blank cargo export cell be treated as N?

Only if the source's documented encoding establishes that meaning and the relevant data was successfully supplied. Otherwise use unknown. A visibly unmarked category in a complete source table and an omitted export field are different observations.

Does every unknown input force a hold?

No. An unknown matters when it can change the disposition. If A is Y, A OR B is Y even when B is unknown. With X=N, that record can pass this rule. If X is unknown instead, the exclusion remains unresolved and the disposition is hold. A known exclusion can settle the outcome without resolving every inclusion field.

Can I combine two separate searches to reproduce OR?

You can combine their matching entity sets in an authorized worksheet if the source scope, dates and category meanings are comparable, then deduplicate overlaps and apply the exclusion to the combined set. Preserve unknown and omitted evidence separately. Do not claim the combined results cover every eligible carrier unless the underlying search coverage supports that conclusion.

When should I run the cases again?

Retest after changing the rule, category mapping, source format, missing-value handling or the tool behavior on which the selection depends. Keep the previous expected results and explain intentional changes. A larger or smaller result count alone does not show that a new version is more accurate.

Make one selection rule reproducible

Write the rule you intend to use, complete the six-case worksheet and resolve membership discrepancies before expanding the list. The useful result is a selection that another researcher can reproduce, with enough evidence to distinguish a match, a deliberate exclusion and an unanswered question.

By John Hauler, USDOTData's editorial persona. Sources reviewed October 12, 2026. The rule notation, test cases and workflow are original editorial guidance, not an FMCSA determination, a description of current USDOTData filter implementation or real customer results.