The insights pages explained: analytics, benchmarks, High Court outcomes and defects

LawyerInsight has four statistics pages in the left-hand menu: Outcome Analytics, Benchmarks, High Court Outcomes and Registry Defects. This guide explains what each one answers, how to read it, and where its numbers stop being useful. For a shorter overview see court-wide insights.

Every figure comes from public Supreme Court of India records and describes the past. It is a research aid, not legal advice, and never a prediction for your matter.

The ground rules

Three rules apply to every page in this guide.

1. Court-wide, by case type or by cohort. A cohort is a group of similar past matters, such as one case type from one origin court. Nothing here is a statistic about a person.
2. Sources and method travel with the number. Insight figures sit beside a Method box that says how many matters it is based on, the filing dates covered, the data-as-of date, and a How this is calculated link to the methodology page.
3. Small groups are not shown. Each metric has a minimum sample. Below it you get a count and a plain "too few" message instead of a figure that would mislead.

Outcome Analytics

Open Outcome Analytics (/analytics) for a quick picture of the whole collection. It answers: how much is in the platform, which kinds of matters are most common, how matters end and how many were decided each year.

What you see

How to read it

The three headline numbers are simple totals. They say how big the collection is, not how good or active anyone is.

In the two bar lists, each bar is drawn relative to the longest bar in that list, so the longest always looks full. Compare the printed counts, not the bar lengths, if you need exact differences. The yearly columns work the same way. Recent years can look different from older ones because records arrive after the Court publishes them and because coverage of older matters is still growing.

This page lists counts only. It carries no per-person ranking, and it has no Method box of its own, so use the methodology page for how the platform defines its other figures.

Benchmarks

Benchmarks (/insights/benchmarks) answers: for matters like mine, how long did they take and how did they end?

Choose a cohort

Use the pickers at the top. Each one can be left on "all".

• Case type
• Origin court, the court whose order was challenged
• Subject
• Government side: Government as petitioner, Government as respondent, Government on both sides, or No government party
• Filed from and Filed to, by year

A line under the pickers describes the cohort in words, for example a case type, origin court, subject and filing years, so you always know what the numbers cover.

Time from filing to disposal

This card shows four figures and a bar.

The bar marks the middle half as a shaded band and the median as a line. The note beneath gives the number of matters, how many are still pending, and, where known, how many times disposed matters were listed at the median.

Why two medians? Pending matters have not finished yet. Leaving them out makes the picture look shorter than it is, so the first figure counts them up to their age on the refresh day (a Kaplan-Meier estimate, a standard method for unfinished cases). The first median is therefore usually the longer one.

How disposed matters ended

This card shows the relief share: matters allowed in whole or in part. Below it, one bar per outcome: Dismissed, Allowed, Partly allowed, Withdrawn, Disposed of, Dismissed on delay, Other. Each shows a count and a share of disposed matters.

When the cohort is too small

If too few matters match every choice, the page widens the cohort and says so: "Too few matters matched every choice, so the cohort was widened by dropping" followed by what was dropped. The methodology gives the order: subject first, then government side, then origin court. The widening starts below 30 matters for durations and below 50 disposed matters for outcomes. If even the widened group is too small, you see "too few to show figures" with the count.

On a case page

Open a case and look for Matters like this one. It shows the same cohort figures next to your matter: either "disposed in N days" against the cohort median, or "pending for N days" with the share of similar matters already disposed of by that age. It does not contain party details.

High Court Outcomes

High Court Outcomes (/insights/high-courts) answers: of Supreme Court matters that came from a given High Court or tribunal, how did they end, by subject?

Reading the table

Rows are origin courts, listed alphabetically. That is an ordering aid, not a league table. Columns are All subjects followed by subject groups. Each cell shows two things: the relief share (matters allowed in whole or in part, out of disposed matters) and n =, the number of disposed matters behind it. Shading gets deeper as the share rises.

A second table, Tribunals and commissions, shows one figure per tribunal across all subjects.

Cells with fewer disposed matters than the minimum (50, per the methodology) show "too few", are outlined and are not shaded, so a thin cell never looks like a low figure. You can filter by Subject and by Filed from and Filed to.

Look behind a number

Select any cell and the matters behind it appear below, with case number, title, outcome and dates, paged with Previous and Next. Each case number opens the case. Titles of sensitive matters are withheld.

Relief in a High Court row describes how Supreme Court matters from that court ended. It is not a measure of the quality of that court or of any judge.

Registry Defects

Registry Defects (/insights/defects) answers: what does the Registry most often flag on fresh filings, and how long do those defects take to cure? It is meant for checking a filing before you submit it.

Choose a Case type, or leave it on all. The statistics cover filings from 2014.

The blocks

• Time to cure a defect: Median time to cure, 90 in 100 cured within, Cured within 28 days, and Never cured.
• Most common defects: up to 12 named defects as bars. Each shows the share of filings of this type that got the defect, the count behind it (n), and the median cure time with the 90 in 100 figure. Defects the Registry text does not clearly name are not shown.
• Registry backlog: open defects on pending cases: open defects and the cases they sit on, grouped by how long ago they were notified: 0 to 28 days, 29 to 60, 61 to 90 and over 90.
• Pre-filing checklist, once you pick a case type: the defects most often noted for that type in plain language, with a Print button that prints one page with the source line and method link.

Method points worth knowing

A defect counts as cured only when every notice of that code on the case was cured. Defects notified before 2014 are left out because they were never marked cured. With fewer than 30 cured defects you see counts only. Cure times are estimates: check with the Registry.

The Methodology page

Open /methodology (or any How this is calculated link). Metrics are grouped under Defects, Timing, Outcomes and Profiles, with chips at the top to jump to one. Each entry gives:

• a plain definition and How it is computed
• Who is counted
• the Minimum sample
• Caveats and Sources
• Data as of and a version number

Links such as /methodology#defects go straight to one metric. The definitions are also served by a public methodology endpoint of the platform; the page itself is refreshed daily.

Limits of statistics

1. The past is not your future. A cohort median does not say how long your matter will take.
2. Coverage is partial and growing. Figures use matters whose case status has been collected, newest first, so older years can be under-represented and their durations can lean short.
3. Duplicates are removed, but records can be wrong. Court records contain errors and gaps.
4. Classifications are rules, not judgement. For example, deciding which party is a government body is rule-based and can misclassify.
5. Look at n. A share from 60 matters moves a lot with a few cases.
6. Estimates are labelled. Cure times and likely listing windows are estimates, not deadlines.

What is deliberately not shown

• Outcome, relief or adjournment rates for a named judge or bench. Outcome statistics are court-wide, by bench size or by cohort only. This is a product rule, not a gap in the data.
• Rankings of lawyers or law firms. Under the Advocates Act and Bar Council rules, advocates are not ranked or promoted. Where a profile shows a number, it is a historical tally with its source, such as how many matters a person appeared in.
• Predictions for a single matter.

Common questions

Why is a cell labelled "too few"? Fewer disposed matters than the minimum sample. Showing a rate would mislead.

Why do the two medians differ? One counts pending matters up to their age, the other counts only finished ones.

Can I see how a particular judge decides? No. See the rules above.

Is the order of High Courts a ranking? No, it is alphabetical.

Where do the numbers come from? Public Supreme Court of India records, listed under each metric on the methodology page.

Can I rely on these for advice to a client? Treat them as background. Check the Court's own record for anything you file or cite.

I think a figure is wrong. Write to support@lawyerinsight.in with the page and the figure.