WFD Match Engine

Match imperfect data with confidence.

Real-world systems rarely describe the same thing in exactly the same way. Match Engine combines evidence across references, descriptions, values, dates and other business data to rank the most likely candidates, so your application can automate clear cases, accelerate review and isolate exceptions.

See how it works Talk to us

Configurable matching intelligence for real-world data.

Source record

Product referenceABC-100

DescriptionFilter cartridge 10in

Quantity2

Unit value£12.50

Candidate 01ABC100

10 inch filter cartridge · Qty 2 · £12.50

Normalised MatchFuzzy Text MatchNumeric Comparison

Candidate 02ABC108

Filter element 10" · Qty 2 · £13.20

Candidate 03ABX100

Filter cartridge · Qty 1 · £12.50

Illustrative scores only. Actual scoring and workflow thresholds are configured and validated for the use case.

Continue

The problem

The same record rarely looks the same everywhere.

Exact matching works when every system shares the same clean identifier. Real businesses rarely have that luxury. Different formatting, terminology, OCR errors and historical identifiers all create uncertainty.

Match Engine is designed for the space between brittle exact matching and overly loose comparison.

System AABC-100

→

System BABC100

DocumentFilter cartridge 10in

→

ERP10 inch filter cartridge

Captured valueO14582

→

Business record014582

SupplierFLT CART 10"

→

Internal10 inch replacement filter cartridge

Multi-evidence matching

One clue can mislead. Several clues can agree.

A single value can be ambiguous. Match Engine lets a strategy combine complementary evidence so the candidate supported consistently across fields can rise above incidental matches.

Reference

Formatting differs

ABC-100
ABC100

Normalised Match

Description

Wording differs

Filter cartridge 10in
10 inch filter cartridge

Fuzzy Text Match

Quantity

Values agree

2
2

Numeric Comparison

Unit value

Values agree

£12.50
£12.50

Numeric Comparison

Candidate 01ABC100

Candidate 02ABC108

Candidate 03ABX100

Start with several plausible candidates.Illustrative scoring sequence

Combined resultCandidate 01 becomes consistently supportedAs independent signals agree, the leading candidate separates from alternatives that were initially almost as plausible.

Workflow outcomes

Don't force uncertainty into a yes or no.

Match Engine returns ranked, confidence-scored outcomes so the surrounding application can respond according to the evidence and the business risk.

Illustrative result92score

Several independent signals strongly support one candidate.

Strong result

Automate

Progress a match automatically when the measured evidence and business risk justify it.

Ambiguous result

Review

Put the strongest candidates in front of a person instead of asking them to search from scratch.

Weak result

Exception

Route uncertain or unmatched records into the appropriate exception process.

The three scores above are illustrative examples, not universal thresholds. Workflow thresholds are use-case specific and should be established against representative data.

Configurable strategies

Define what a good match looks like.

Different matching problems need different evidence. Strategies define the matching policy without hard-coding it into the surrounding workflow.

Choose the fields

Map the source and target values that carry useful evidence: references, descriptions, dates, quantities, values, names and business-specific attributes.

Prepare the values

Normalise formatting, extract meaningful components or prepare numeric and date values before comparison.

Choose the evidence

Apply the public matching methods appropriate to each field.

Exact MatchNormalised MatchPattern / Extraction MatchNumeric ComparisonDate Comparison

Weight what matters

Let distinctive evidence contribute more than common or weak signals where that reflects the business problem.

Set thresholds

Use the scored outcome to decide where automation, review and exception handling should begin.

Change the matching strategy without rebuilding the surrounding application.

Use cases

One matching capability. Many problems.

Match Engine is not tied to one industry, document type or application. Its strategies describe the evidence that matters for each matching problem.

Accounts payable

Invoice to PO matching

Use references, descriptions, quantities, values and other available evidence to rank the purchase-order lines most consistently supported by invoice data.

Catalogue

Product cross-reference

Rank internal products when supplier SKUs, packaging descriptions and terminology differ from the customer's catalogue.

Transformation

Migration & reconciliation

Rank likely equivalent records across old and new platforms when there is no dependable shared identifier.

Master data

Entity matching

Identify potential customer, supplier or other entity equivalents where names, references and historical identifiers differ.

Data quality

Duplicate prevention

Evaluate a proposed record against an appropriate candidate population before another near-duplicate is created.

Document automation

OCR validation

Use business-system candidates as contextual evidence when extracted values are plausible but imperfect.

Integration

Use only the Match Engine capabilities you need.

Core is the matching service. Proxy is an optional companion layer when the solution needs target resolution, centrally managed strategies or job orchestration.

Match Engine Core

Bring the candidates. Match Engine supplies the matching intelligence.

Use the stateless Core API directly when your application already knows which candidate records should be compared.

Source + candidates + strategy

→

Ranked candidates + score information

Independently usable
Stateless matching API
Direct Core calls remain available even where Proxy is deployed

Optional companion

Match Engine Proxy

Add orchestration where the matching layer needs to do more than evaluate caller-supplied candidate records.

Server-side target resolution
Realtime target data from your own database
Controlled or persisted strategy configuration
Asynchronous jobs and Job IDs
Realtime completion notification
REST job state for recovery and polling

Proxy does not replace Core.
Direct Core and Proxy-mediated integration can coexist in the same environment.

Request pathDirect CoreApplication retrieves target data, sends source + target + strategy to Core, then receives the ranked result.

Applicationsource, target and strategy

Proxy optionalresolve / orchestrate

Target sourcerealtime target data

Corematching intelligence

Ranked resultscore + traceability

How it works

From imperfect data to a useful decision.

Supply the source

Send the record or extracted values that need to be identified.

Find candidates

Supply candidate records directly or resolve them through an appropriate Proxy integration.

Apply strategy

Choose fields, preparation, methods, weights and result behaviour.

Combine evidence

Independent signals contribute to the strength of each candidate.

Rank candidates

The most consistently supported records rise to the top.

Drive workflow

Automate, review or escalate according to the consuming application's policy.

Source datastrategyevidenceranked candidatesworkflow decision

Match Engine in context

See where matching fits in a real solution.

This is the same WFD Engine process demonstration used on the Accounts Engine site: an incoming document moves through focused capabilities, Routing Engine orchestrates the flow, Telemetry Engine observes it, and downstream actions can run in parallel.

The WFD Engine ecosystem

Use the Engines you need.

Each WFD Engine is a focused capability that can be used independently. Select an Engine to see the role it can play alongside Match Engine.

Frequently asked

Match Engine, in practical terms.

Does Match Engine require exact matching keys?

No. It is designed for cases where a dependable shared identifier does not exist or where values differ in format, wording or quality.

Does Match Engine make the final decision?

It does not have to. Match Engine provides ranked evidence; the consuming application decides whether to automate, ask for review or treat the item as unmatched.

Do we have to deploy Match Engine Proxy?

No. Core is independently usable. Proxy is optional and adds server-side target resolution, stored strategy and job orchestration capabilities where they are useful.

Can applications still call Core when Proxy is deployed?

Yes. Direct Core and Proxy-mediated integration can coexist in the same environment.

Does Match Engine use AI?

AI-assisted evidence can be included optionally for suitable strategies and approved providers. The platform is not dependent on AI.

Can Match Engine run on-premise?

Yes. It can fit Windows/IIS environments as well as AWS-oriented service architectures.

Proof over promises

Test one real matching problem.

The useful question is not whether a matching engine sounds intelligent. It is whether it improves the result on your data. Bring a representative set of clear matches, ambiguous cases and genuine no-matches, then measure the outcome.

Obvious matchesFormatting differencesOCR noiseAmbiguous casesReal no-matches

Talk to Workflow Doctor