Union campaigns: distinguish petitions, votes and certification
A union campaign can generate a petition, an election headline and several later procedural documents before representation status is established. None of those alone supplies the terms of a collective-bargaining agreement. Research should trace the specific case, the represented unit and the evidence available at each date before estimating the operating exposure.
The deliverable is a US National Labor Relations Board case ledger with separate rows for petition, unit definition, election tally, challenges, objections and certification or another documented route to recognition. This article uses official pages read on October 25, 2026. It does not apply this procedure to railway or airline representation systems or predict a dispute’s legal outcome.
Locate the official process and voting basis
The NLRB Process distinguishes representation petitions, including RC, RD and RM cases, from unfair-labor-practice proceedings. Read its Representation Election Process section. A representation case number should not be replaced by a news article about a different labor proceeding at the same employer.
NLRB’s Conduct Elections page explains unit and voter eligibility arrangements, a majority of votes cast, later objections and certification, and an alternate voluntary-recognition route. These are different steps and pathways. For an actual case, use the dated case documents and applicable procedure rather than assume that a process overview determines every pending objection or challenge.
Begin with the employer’s legal identity, site, case number, petition type and unit description. Record the proposed unit separately from the agreed or directed voting unit. A petition covering one group of employees does not mean all employees of a national company are represented.
Rebuild the main tally
Assume a fictional two-choice representation election with 120 eligible voters. The uncontested valid tally is forty-seven for representation and forty-three against, totalling ninety. Eight additional ballots are challenged and unresolved; assume there are no other ballot categories in this teaching case.
Forty-seven divided by ninety is approximately 52.22%, a majority of the currently counted valid votes. Forty-seven divided by 120 is about 39.17%, a different fraction of eligible people. The latter does not by itself negate the former tally. Do not silently substitute eligible population for the vote denominator used to describe the count.
| Record | Count | Interpretation |
|---|---|---|
| Eligible voters | 120 | Population on the defined eligibility basis |
| Counted for representation | 47 | Current uncontested valid vote |
| Counted against | 43 | Current uncontested valid vote |
| Unresolved challenged ballots | 8 | Not assigned to either side |
| Ballots cast under stated assumptions | 98 | Ninety counted plus eight challenged |
The number cast divided by eligible voters is about 81.67%, under the explicit no-other-ballots assumption. It is not the same as the ninety already counted. Preserve challenged ballots as unresolved evidence rather than count them automatically against the union or discard them from the case history.
Test whether challenges can change the arithmetic
The current margin is four, smaller than eight unresolved ballots. If all eight eventually become valid against votes, the count would be forty-seven to fifty-one. If all become valid for votes, it would be fifty-five to forty-three. The pending population can therefore change the numerical outcome.
If all eight are admitted and x of them favour representation, the hypothetical count is 47 + x versus 51 − x. Two favourable challenged ballots create a forty-nine-to-forty-nine tie; three create fifty to forty-eight. These are arithmetic possibilities, not a prediction of eligibility decisions or a ruling on objections.
Some challenges may instead be excluded, changing the final denominator. Wait for the actual disposition and final tally rather than assign an average probability to each ballot. A model showing the range cannot substitute for the agency’s resolution or certification document.
Follow the unit across documents
Suppose the petition originally described 150 positions but the election documents define a unit with 120 eligible voters under their stated date and criteria. Keep both counts with their document dates. The difference is not automatically thirty job losses; the proposed scope, exclusions or eligibility basis may have changed.
Match the final unit’s job classifications and locations to the issuer’s disclosure. A certified unit of 120 within a company employing two thousand people would cover 6% of that illustrative headcount, if the populations genuinely match. It would not establish company-wide representation, bargaining costs or a uniform wage increase.
Do not aggregate the same unit twice when a later case amends, accretes to or otherwise refers to an existing unit. Obtain the specific relationship before adding employee populations. If the documents cannot be matched, show the overlap as unknown instead of selecting the largest possible represented total.
Separate tally, certification and bargaining
An election headline may describe the initial tally while objections or determinative challenges remain. Record the actual procedural status, date and document. A later certification should be linked to the same unit and case, with any subsequent review or relevant order retained separately.
Representation status also does not supply a signed labor contract. For an operating-cost conclusion, obtain the actual agreement, effective date and covered population in a separate analysis. This tracker does not infer wages, strike dates, bargaining duration or an earnings effect from the vote margin.
Voluntary recognition is another route described by the NLRB. It should have its own recognition evidence and unit scope rather than a fabricated election tally. Do not require every recognized unit to have the same election document, and do not treat an employer’s statement about talks as proof that recognition has occurred.
Test a different evidence boundary
Suppose management says a union won, a news report says workers organized and the public case docket has not yet supplied a certification document. These statements may describe related developments without proving the same stage. Preserve the claim and its author, then state that certification was not located in the reviewed records at the cutoff.
The absence of a located document is not proof that no procedural action occurred. Conversely, a later document cannot be treated as evidence publicly available on the earlier vote date. Store activity date, publication date and research-access date when they differ.
Use the case verification sheet
| Input | Obtain | Check |
|---|---|---|
| Case identity | NLRB case record and filing | Employer, site and petition type |
| Unit | Petition and final election documents | Proposed versus voting/recognized scope |
| Votes | Actual tally and ballot categories | Counted valid, challenged and other ballots |
| Disputes | Challenge and objection dispositions | Arithmetic range not treated as a ruling |
| Representation | Certification or recognition evidence | Same unit and date, current status |
| Commercial effect | Separate signed agreement and company records | Representation not relabelled contract terms |
When sources conflict, compare preliminary versus final tallies, unit boundaries, petition type and publication dates. Do not average counts from different units or treat the latest headline as automatically superseding an agency document. Retain a named unresolved item and the record required to close it.
If eligible counts or ballot dispositions are missing, report only the supported tally and procedural stage. Avoid inventing turnout or a final result. Update after revised unit documents, resolved ballots, objections, certification, recognition or a later agreement. The conclusion should state who is represented according to which evidence, at which stage and as of which date.
Close challenged categories with the actual disposition
For a separate possible outcome in the main case, suppose six of the eight challenges are admitted: four for representation and two against. Two ballots are excluded. The final valid count would be fifty-one to forty-five, or ninety-six valid ballots, while total ballots cast remain ninety-eight under the original assumptions.
This yields 53.125% of final valid votes for representation. Reconcile the two excluded ballots rather than changing turnout retroactively. Even this completed arithmetic does not dispose of a separate objection or replace the actual certification record. Keep vote-count finality and representation status in different fields, and use the source document’s own disposition wording.
Petition type also matters to a portfolio total. A filing seeking to remove an existing representative is not another newly organised unit. Preserve the RC, RD or RM designation and the actual requested action before classifying a development as an increase in represented workers. A count of representation petitions alone does not establish the direction of change.
FAQ
Does 47 out of 120 disprove a majority in the counted vote?
No. In the example the counted valid vote is forty-seven out of ninety, while eight ballots remain unresolved.
Does certification establish the wage terms of a contract?
No. The actual agreement and its effective scope need separate evidence.