Resource Matching: Matching logic and confidence evaluation

Learn how potential matches are identified, evaluated, and scored

This article explains the logic behind Resource Matching and Resource Matching Intelligence. It focuses on how candidate records are retrieved, how identity alignment is evaluated, how Match Confidence is generated, and what safeguards help reduce false positives.

Resource Matching evaluates likelihood, not certainty. Users always remain responsible for reviewing information and making final decisions.

How Resource Matching works

Resource Matching evaluates records in a series of stages.

* Resource Matching Intelligence only.

✅Every Resource Matching view uses the same retrieval and identity-evaluation foundation. Identity information drives matching outcomes, while supporting information provides additional context. Resource Matching Intelligence extends that foundation with multi-signal evaluation, weighted scoring, Match Confidence, semantic evaluation, and result prioritization.

What applies to all Resource Matching views

The following logic applies to Essential, Enhanced, and Resource Matching Intelligence views.

All three views use the same matching foundation. Resource Matching Intelligence adds evaluation depth and confidence-based prioritization, but it does not change the core matching rules.

Candidate retrieval

Resource Matching begins by identifying records that may represent the same individual. Potential candidates are retrieved using trigger-based matching.

A retrieved record is not automatically considered a match. Retrieval only determines which records should be reviewed.

Matching triggers

*Not used for Resource Matching Intelligence.

Soundex matching

Names are often entered differently across suppliers, workflows, and systems. To improve candidate retrieval, Resource Matching uses Soundex-based phonetic matching. Soundex helps identify similar-sounding names even when spelling differs.

Examples

Identity-first evaluation

After candidates are retrieved, Resource Matching evaluates identity alignment. Identity alignment is required before confidence can increase significantly.

Identity information is always evaluated first because it provides the strongest evidence that two records represent the same individual.

Identity information

Examples include:

• First Name
• Last Name
• Middle Name
• Email Address
• Initial Personal ID

Supporting information

Supporting information provides context after identity alignment exists. Examples include:

• Phone numbers
• Address information
• Skills
• Certifications
• Education
• Resume information
• Work experience
• Professional history

Supporting information can strengthen confidence, but it cannot independently establish identity. 

Example

• Matching email addresses provide strong identity evidence.
• Similar job titles alone do not establish identity.
• Similar experience alone does not establish identity.

How data is evaluated

Different types of information require different comparison methods. The comparison method depends on the type of information being evaluated.

All Resource Matching views follow the same evaluation flow:

1. Retrieve candidate records
2. Evaluate identity information
3. Evaluate supporting information
4. Present results for review
5. Require user validation

Users always make the final decision.

User validation

Resource Matching evaluates likelihood, not certainty.

All views require users to:

• Review candidate records
• Compare available information
• Validate identity alignment
• Make the final decision

Resource Matching does not:

• Automatically merge records
• Automatically resolve conflicts
• Automatically determine identity
• Automatically make onboarding decisions

✅ The system provides guidance. Users determine the outcome.

Identity continuity and key records

Resource Matching helps maintain identity continuity across suppliers, assignments, onboarding activities, and reporting processes.

Key identity records

Resource Records and Security User Records

Resource Matching helps maintain accurate relationships between Resource Records and Security User Records. 

Accurate matching helps ensure that:

• Resource identity remains consistent
• User access remains associated with the correct individual
• Identity conflicts are minimized
• Shared-user-record scenarios are avoided

Maintaining these relationships is a core objective of Resource Matching. Accurate matching helps maintain alignment between Resource Records and Security User Records throughout onboarding and access-management processes.

Common terminology

Depending on program configuration or historical terminology, you may also encounter the following terms:

Why these records matter

As individuals move between requisitions, assignments, suppliers, and other workforce processes, identity information is associated with multiple records across the platform.

Resource Matching helps evaluate those records consistently to:

• Reduce duplicate identities
• Maintain accurate workforce information
• Preserve identity continuity across workflows
• Support accurate reporting and tenure tracking

What applies only to Resource Matching Intelligence

Resource Matching Intelligence builds on the standard matching process by adding advanced evaluation features.

The core matching logic remains the same. Resource Matching Intelligence adds:

• Multi-signal evaluation
• Weighted scoring
• Match Confidence generation
• Result prioritization
• Semantic evaluation of professional information

These features help users prioritize review but do not replace user validation.

Resource Matching Intelligence capabilities

Important

• Match Confidence and Weighted Scoring are available only in Resource Matching Intelligence.
• Essential and Enhanced views do not calculate scores or generate confidence levels.

How multi-signal evaluation improves accuracy

Standard Resource Matching presents information for comparison and review.

Resource Matching Intelligence evaluates multiple signals together rather than considering each signal independently.

Why this matters

Individual signals often provide only part of the picture.

For example:

• A matching name may indicate a possible match.
• A matching email address may provide stronger identity evidence.
• Similar experience may reinforce the result.
• Conflicting identifiers may reduce confidence.

When evaluated separately, these signals can be difficult to interpret consistently. Multi-signal evaluation combines them into a broader assessment of identity alignment.

Example signals

Benefits

Multi-signal evaluation helps:

• Reduce reliance on any single field
• Improve evaluation consistency
• Balance strong and weak signals appropriately
• Identify meaningful patterns across records
• Prioritize likely matches for review

Multi-signal evaluation does not change identity-first matching rules. Identity information remains the primary driver of matching outcomes.

How Signal Weighting works

Not all information provides the same level of confidence when determining whether two records represent the same individual.

Resource Matching Intelligence assigns greater weight to signals that are generally stronger indicators of identity and less weight to signals that provide supporting context.

Weighting model

The following weighting model illustrates how identity-focused signals receive greater influence than supporting signals during evaluation.

Why weighting improves accuracy

Without weighting, weaker supporting signals could influence results as much as stronger identity signals.

Weighting improves accuracy by:

• Giving strong identity signals greater influence
• Limiting the influence of weaker supporting signals
• Balancing the combined effect of multiple signals
• Reducing overreliance on professional similarity
• Producing more consistent evaluation outcomes

How weighting affects signal interaction

Resource Matching Intelligence evaluates signals together rather than independently.

Example:

In this scenario:

• Matching identity signals increase confidence.
• Resume similarity reinforces the result.
• The conflicting Initial Personal ID reduces confidence.

The outcome reflects the combined effect of all signals rather than any single signal by itself.

How Match Confidence is generated

Resource Matching Intelligence evaluates signals, applies weighting, generates a composite score, and maps that score to a confidence level.

Evaluation pipeline

Confidence levels

Score thresholds are applied after weighted evaluation and confidence mapping.

Confidence interpretation

The following confidence levels indicate the strength of identity alignment identified through multi-signal evaluation, weighted scoring, and confidence mapping.

What influences Match Confidence

Match Confidence reflects how strongly identity and supporting signals align across records. Incomplete identity fields may limit confidence because fewer identity signals are available for evaluation.

High Confidence

Typically occurs when:

• Multiple identity signals align
• Strong identifiers match
• Supporting information reinforces identity
• Few or no significant conflicts exist

Possible Confidence

Typically occurs when:

• Some identity signals align
• Information is incomplete
• Identity information is missing
• Limited conflicts exist

Low Confidence

Typically occurs when:

• Identity information conflicts
• Reliable identifiers do not align
• Identity evidence is limited
• Similarity is driven primarily by supporting information

Typical examples

How resume similarity affects confidence

Resume similarity is treated as supporting information. 

Resource Matching Intelligence can evaluate:

• Resume content
• Skills
• Certifications
• Professional history
• Experience descriptions

This information may strengthen confidence when identity signals already align.

Example

If two records share:

• Similar names
• Matching identity information
• Similar experience

Resume similarity may reinforce the evaluation and increase confidence. However, if identity information conflicts, resume similarity alone cannot establish a match.

How conflicting signals affect evaluation

Resource Matching Intelligence evaluates signals individually before combining them into an overall assessment.

When signals conflict, confidence may decrease.

Examples of conflicting signals

• Matching names but different identifiers
• Similar resumes but different email addresses
• Similar professional history but different Initial Personal IDs

Conflicting identity information is generally treated as a stronger indicator than supporting similarity.

Example

Although several signals align, the conflicting email reduces confidence because identity-related information carries greater weight than supporting information.

Supporting information vs. identity conflict example

Although the professional information appears highly similar, the conflicting email address significantly reduces confidence because identity information carries greater influence than supporting information.

This example demonstrates why professional similarity alone cannot establish identity.

Result

Conflicting identity information may:

• Lower confidence
• Prevent confidence from reaching a higher range
• Increase the need for manual review

Strong supporting information can increase confidence, but it cannot outweigh significant identity conflicts.

Confidence example

The following example illustrates how multiple identity and supporting signals interact during evaluation and how those combined signals contribute to a Match Confidence outcome.

Example scenario

Evaluation result

In this scenario:

• First names align
• Last names are highly similar
• Email alignment increases confidence
• Supporting information reinforces the result
• Different Initial Personal ID values reduce confidence

The final outcome may still fall within the High Confidence range because several strong identity signals align and outweigh less significant conflicting signals.

Semantic evaluation

Some information cannot be effectively compared through exact matching alone.

Resource Matching Intelligence can evaluate professional information using semantic evaluation.

Why semantic evaluation exists

Traditional text comparison relies heavily on matching words and phrases.

Professional information often varies in wording even when it describes similar experience, skills, or qualifications.

Semantic evaluation helps identify related qualifications, experience, and expertise when different terminology is used. This improves evaluation coverage while preserving identity-first matching rules.

Examples

Although the wording differs, the underlying meaning may be similar.

How semantic evaluation works

Semantic evaluation compares meaning rather than relying solely on exact text matches.

To support this evaluation, Resource Matching Intelligence transforms professional information into comparable representations and evaluates how closely the information aligns.

Similarity calculations are used to identify related experience, skills, and qualifications even when different terminology is used. This allows semantically related concepts to contribute supporting context even when identical keywords are not present.

Information that may use semantic evaluation

• Resume content
• Skills
• Experience descriptions
• Professional history
• Certifications

Technical overview

Resource Matching Intelligence may apply:

• Meaning-based comparison techniques
• Comparable representations of professional information
• Similarity calculations across professional attributes
• Multi-signal evaluation that combines identity and professional signals 

Safeguards that reduce false positives

Resource Matching is designed to prioritize identity accuracy over similarity. Several safeguards help reduce the likelihood of incorrectly identifying two records as the same individual.

Identity-first evaluation

Identity information always carries greater influence than supporting information.

Examples include:

• Name
• Email Address
• Initial Personal ID

Supporting information contributes only after identity alignment exists.

Identity conflicts reduce confidence

Conflicting identity information lowers confidence.

Examples include:

• Different identifiers
• Conflicting contact information
• Misaligned identity fields

These conflicts are treated as more significant than supporting similarities.

Professional similarity has limited influence

Professional information contributes only a small portion of the overall evaluation.

Examples include:

• Skills
• Experience
• Certifications
• Resume content
• Education

Professional similarity can reinforce confidence, but it cannot outweigh significant identity conflicts.

Multiple signals are evaluated independently

Signals are evaluated independently before being combined into an overall assessment.

This helps ensure that:

• Strong identity signals receive appropriate weighting
• Weak signals do not dominate outcomes
• Missing information affects confidence appropriately
• Evaluation behavior remains explainable

User validation remains required

Match Confidence is guidance—not a decision.

Resource Matching Intelligence does not:

• Confirm identity
• Merge records
• Resolve matches automatically
• Override user decisions

Users must always review information and make the final decision.

Result

These safeguards help ensure that:

• Identity remains the primary driver of matching outcomes
• Similar experience alone does not create false matches
• Missing or conflicting identity information lowers confidence
• Matching decisions remain explainable and auditable
• Final control remains with the user

Privacy and governance

Resource Matching is designed to support explainable, auditable, and controlled identity matching.

Data processing principles

• Only necessary information is processed.
• Identity information supports duplicate detection and continuity.
• Supporting information strengthens evaluation.
• Results do not determine qualifications or hiring suitability.
• Human validation remains required.

Built-in safeguards

Key takeaways

Across all Resource Matching views

• Candidate retrieval identifies records for review
• Soundex improves candidate retrieval but does not determine matches
• Identity information is evaluated before supporting information
• Supporting information can strengthen confidence but cannot establish identity
• User validation is always required

Resource Matching Intelligence extends the process

• Multiple signals are evaluated together
• Stronger identity signals receive greater weighting
• Match Confidence reflects overall alignment across evaluated signals
• Semantic evaluation helps compare professional information with different wording
• Results are prioritized to help focus review efforts

Built-in safeguards help reduce false positives

• Identity information always carries greater influence than professional similarity
• Conflicting identity information lowers confidence
• Resume and experience similarity cannot independently establish identity
• No automated matching decisions occur
• Users remain responsible for all final decisions

✅ Resource Matching evaluates likelihood, not certainty. Match Confidence supports decision-making, but user validation always determines the final outcome.