# Matching Music Usage to Musical Works at Scale > How Noctil's MatchIQ engine links public performance and digital usage of recordings to the correct musical works for PROs and MROs — bridging performer to writer and recording to work — with under 2% error and 15–20% less manual review. - Segment: Music Rights Organization - Platform: Noctil Platform - Engine: MatchIQ - Author: Jacob Varghese - Published: 2026-11-01 - Source: https://www.noctil.com/use-cases/matching-music-usage-to-musical-works **In short:** Noctil's MatchIQ engine matches public performance and digital usage, which reports recordings, to the correct musical works in a PRO's or MRO's repertoire. It works even when the work records hold only minimal recording data, such as a title, an artist and an ISRC. In benchmark runs of 100,000 usage lines, MatchIQ keeps false positives and false negatives below 2% on average. Compared with legacy, rule-based matching engines, it also cuts manual review of partial matches by 15–20%, and runs in parallel to finish any volume inside the distribution window. --- _Link every recording played or streamed to the song behind it. For performing rights organizations (PROs), mechanical rights organizations (MROs) and other music CMOs matching public performance and digital usage of recordings to their musical work repertoire._ ## The Situation PROs and MROs collect royalties for songwriters, composers and music publishers. Their usage data comes from radio and TV broadcasters, venues, public performance licensees, digital service providers (DSPs) and partner societies abroad, often millions of lines per distribution period. That usage describes recordings: what was played or streamed. Royalties, though, are owed on the musical work, the underlying composition. Every recording in the usage has to be linked to the work it embodies before the writers and publishers of that work can be paid. ## The Problem Matching recordings to works is harder than matching recordings to recordings, because the two sides describe different things: - **Thin recording data on works.** Work registrations focus on writers, publishers, shares and ISWCs. The recordings linked to a work are often limited to a title, an artist and an ISRC, and many recordings of the work are missing entirely. - **Performer is not writer.** Usage names the performing artist, while works are registered under their writers and composers. A cover, a featured artist or a session performer gives no direct path to the work. - **Recording titles differ from work titles.** Remixes, live versions, radio edits, acoustic versions, translated titles and "feat." credits all change the title in the usage, while the work title stays the same. - **One work, many recordings.** A popular song can have hundreds of recordings across covers, versions and compilations. Only some are linked to the work. - **Missing or wrong ISRCs.** Many usage lines have no ISRC, or one that isn't linked to any work in the repertoire. - **Medleys and samples.** One recording can embody several works, each with its own writers and shares. Unmatched usage leaves royalties unclaimed or sends them to undistributed ("black box") funds. Partial matches pile up for manual review, which puts distribution deadlines and member trust at risk. ## The Solution MatchIQ, the Noctil Platform's matching engine, bridges the gap between the recording described in the usage and the musical work in the repertoire. 1. **Recording-to-work inference.** MatchIQ infers which work a recording embodies from context, such as title, performer, ISRC, duration, release details and known versions. It works even when the work record carries only minimal recording data. 2. **Title normalization.** Version labels (remix, live, edit, acoustic), featured-artist credits, translations and alternate titles are recognized, so the recording title resolves to the work title. 3. **Performer-to-writer bridging.** MatchIQ connects performing artists to the writers and works they record, which allows covers and featured performances to be matched. 4. **Identifier use and repair.** ISRCs, ISWCs and IPI numbers are used wherever they exist. When an ISRC is missing or unlinked, MatchIQ infers the work and can surface the new recording-to-work link to enrich the repertoire. 5. **Multi-work recordings.** Medleys and recordings containing samples can be linked to more than one work. 6. **Configurable thresholds and parallel scale.** Organizations tune confidence thresholds to balance automation against review. Matching runs in parallel to finish any volume within the distribution period, at a low cost per match. ## The Result - **Under 2% false positives and false negatives** on average, in benchmark runs of 100,000 usage lines - **15–20% less manual review** of partial matches, compared with legacy, rule-based matching engines - **Less money in undistributed funds**, because more performance and digital usage is matched to works - **A richer repertoire**, as newly inferred recording-to-work links feed back into the work data - **Matching completed within the distribution period**, at any volume, through parallel processing ## Frequently Asked Questions **How do PROs and MROs match recordings to musical works?** Noctil's MatchIQ infers which musical work a recording embodies from title, performer, ISRC, duration and release context. It links public performance and digital usage to the right work even when the work has little recording data attached. **Can MatchIQ match a cover version to the original musical work?** Yes. MatchIQ connects performers to the writers and works they record, so covers and featured performances resolve to the correct work. **What if the ISRC in the usage isn't linked to any work?** MatchIQ infers the work from context and can surface the new recording-to-work link, so the repertoire becomes richer over time. **Does MatchIQ handle remixes, live versions and translated titles?** Yes. Version labels, featured-artist credits, translations and alternate titles are normalized to the underlying work title. **How accurate is MatchIQ for musical works?** In benchmark runs of 100,000 usage lines, false positives and false negatives average below 2%. Manual review of partial matches is 15–20% lower than with legacy, rule-based engines. **Who is this for?** Performing rights organizations (PROs), mechanical rights organizations (MROs), music CMOs and publishers' administrators who distribute royalties for public performance and digital use of musical works.