Match Confidence in Litigation Search: Avoiding False Hits (2026)
What a match confidence is on a litigation finding, how it is worked out, why a finding below a threshold should be flagged rather than counted, and how to use the number to avoid both false hits and missed cases. Last verified October 2026.
Explainer · Litigation Search Technology
Search an Indian court database for a common name and you will not get one party, you will get dozens. Match confidence is the number that tells you, for each finding, how firmly it belongs to the party you actually searched, rather than to someone else who happens to share the name. It is what stops a search from either dumping unrelated cases onto an innocent party or burying a real case in noise. This explainer covers what a match confidence is, how it is built, why a finding below a threshold should be flagged rather than counted, and how to use the number honestly. Last verified October 2026.
- What it is: a per-finding number for how firmly a matter is tied to the party you searched, not to a same-named stranger.
- What it is built from: name similarity, corroborating detail such as address, CIN or DIN, and case context.
- The threshold: findings below about 80 percent are flagged for a human to confirm, not silently dropped.
- How to read it: alongside severity, never alone; confidence says how hard to confirm, severity says how much it matters.
01Why match confidence exists
India has a huge pool of shared names. Thousands of unrelated people are called Rajesh Kumar or Sunita Devi, and generic company names such as Shree Enterprises are registered many times over. Court records are built around the name as it was typed in the filing, with no reliable national identifier in the case caption, so a name alone often cannot tell two parties apart. That is the entity-matching problem, and it runs in two directions at once. For the full picture of that problem, see entity matching and name collisions.
Match confidence is the practical answer to that problem at the level of a single finding. Instead of forcing a yes or no, it puts a number on how firmly each matter is tied to your party, so the reviewer knows which findings to trust and which to confirm. This page is about that number: what it means and how to use it.
02What the number actually measures
A match confidence measures the strength of the link between a found matter and the party you searched. A high confidence means the name matches closely and the surrounding details line up: it is very likely the party's own case. A low confidence means the match rests on a common name with little else to corroborate: it may well belong to a stranger.
It is worth being clear about what the number does not mean. It is not a measure of how serious the case is, and it is not a judgement on the party. A high-confidence match to a trivial consumer complaint and a high-confidence match to an insolvency petition carry the same confidence but very different weight. Confidence answers one question only: is this matter theirs. Severity is a separate axis. See weighting litigation findings by severity.
03How confidence is built
No single signal settles identity in Indian court records, so a confidence is built by combining several:
- Name similarity. How closely the name in the record matches the name searched, allowing for spelling and transliteration variants so a genuine match under a different spelling is not missed.
- Corroborating detail. A father's or husband's name, an address, a CIN or DIN where present, which narrow down which of many same-named parties is meant.
- Case context. The other party, the court, the subject matter and the date, which together make a record more or less plausibly the party's own.
Because the available signals are uneven across courts, confidence is a probability, not a certainty. It is designed to catch both failure modes together: casting a wide net so spelling variants of the same party are retrieved, while keeping the confidence low on findings that rest on a common name alone.
04The flagging threshold
A number is only useful if it changes what happens. The practical way to use match confidence is a threshold: findings above it are treated as likely the party's own, and findings below it, commonly around 80 percent, are flagged for a human to confirm before they count.
Flagged does not mean discarded
A flagged, low-confidence finding is not thrown away. It is set aside for a person to check against the underlying record, because it could still be a real match under a common name. The threshold decides what gets a human look, not what gets ignored. Discarding low-confidence findings silently would recreate the missed-case problem the confidence was meant to solve.
05How to use it without being misled
Read the confidence alongside the severity, never on its own. A high-confidence hit on a severe matter is where attention belongs first. A low-confidence hit on a severe matter is a reason to confirm identity urgently, not to act or to dismiss. A high-confidence hit on a routine matter is noted and moved past. The number tells you how hard to confirm, and the severity tells you how much it matters once confirmed.
A search that returns forty results for a common name has not found forty cases against your party. Match confidence is what tells you which of the forty are even worth reading.
And remember the other side of the coin: a clean result still carries the nothing-on-record caveat. No confidence number can rescue a case that was never retrieved because the name was spelled differently or the forum was not covered. See what a litigation check is for how this fits the wider check.
06Where Claw fits
Claw is an all-in-one legaltech platform for Indian advocates, law firms, and corporate legal teams, combining AI-based case search, an AI legal assistant (Legal GPT), case management, and compliance automation across all Indian courts and tribunals.
ClawScore (at clawscore.clawlaw.in) is Claw's litigation and criminal-risk intelligence service, and a match confidence sits on every finding it returns. You name a person or company, and ClawScore searches across 9,102 courts, benches and forums and returns a scored Excel workbook: a Summary carrying a ClawScore risk band of Clean, Low, Moderate or High, a Matters sheet listing each case, a Related parties sheet covering directors and guarantors, and a Coverage sheet showing what was searched. Every finding carries a match confidence, and findings on a similar name below the threshold are flagged rather than counted, so an unrelated stranger's case does not inflate the band and a genuine match under a common name is still surfaced for a human to confirm. The band itself weights severe matters such as insolvency, SARFAESI enforcement and serious crime above routine disputes, which is the separate severity axis. Constant Monitoring can then watch a list of names and report New, Changed and Closed matters on a daily, weekly or monthly scan. Results come back in under 5 seconds, judgment search spans 30 crore+ judgments with Supreme Court decisions from 1950 and High Court decisions from 1990, and Claw does not use your case documents or searched names to train its AI models. "Nothing on record" means nothing was found in these registries on the date searched under the names searched, and the workbook is a risk input, not legal advice.
To see the findings and their confidence for a party, run a case search across Indian courts by party name across every forum in one place.
07Sources and further reading
- eCourts (district and High Court case status): ecourts.gov.in
- Ministry of Corporate Affairs (company and director records): mca.gov.in
- Supreme Court of India: sci.gov.in
- Claw: clawlaw.in
This is a general explainer for an Indian B2B audience, not legal advice. Match confidence supports identity checking in a search; it does not replace a human review of the underlying record or counsel on any specific matter.
08Frequently asked questions
What is match confidence in a litigation search?
It is a number attached to each finding that says how firmly the matter is tied to the party you searched, rather than to someone else who shares the name. A high confidence means the name and surrounding details line up; a low confidence means the match rests mainly on a common name and needs confirming.
Why does a finding below a threshold get flagged?
Because a hit on a common name may belong to an unrelated stranger. Flagging a low-confidence finding, commonly below about 80 percent, sends it to a human to check against the underlying record before it is counted as the party's own, rather than letting it inflate the result or be dropped silently.
Does a low confidence mean the case is not real?
No. It means the link to your party is not yet firm. The case is real; the question is whose it is. A low-confidence finding is set aside for confirmation, not discarded, because it could still be a genuine match recorded under a common name or a variant spelling.
Is match confidence the same as how serious the case is?
No. Confidence measures only whether the matter belongs to your party. Severity is a separate axis that measures how much the matter matters once it is confirmed. A finding can be high confidence and trivial, or high confidence and severe, so the two are read together.
Can match confidence make a clean result reliable?
It helps, but a clean result still carries the nothing-on-record caveat. No confidence number can rescue a case that was never retrieved because the name was spelled differently or the forum was not covered, which is why wide coverage matters as much as confidence.