Two children, one score
A 72 can mean a strong reader who writes too little, or a weak reader who guesses well on multiple choice. The mark treats them as the same child. The remedy is not.
P3M.AI, Assessment Intelligence
One paper.
Four reports.
Nobody left guessing.
IMPACT scores every question against a named criterion, then writes what the pattern means — for the child, for the teacher taking the class on Monday, and for the head who has to decide where next year's time goes.
4Report levels
6Subjects covered
K–12Grade coverage
2 wksPaper to reports
12Schools benchmarked
Marks are collected.
Meaning is not.
Most schools already assess more than enough. What arrives afterwards is a column of numbers that nobody has time to interrogate before the next cycle begins.
Sections drift apart quietly
Two sections start a year within three points of each other and end nineteen apart. Nobody notices until the board year, because nothing in the current reporting compares them.
The teacher gets it too late
Results that arrive after the chapter has closed can be filed but not acted on. Timing is the difference between an assessment and an autopsy.
How it works
Every question is
tied to something
you can teach.
IMPACT is built on a three-level architecture. It is the reason a low score points at a page number instead of a mood.
Level 1 · Parameter
The broad ability
The skill area a subject is made of. Six per subject, fixed across the school so Class 6 and Class 8 are comparable.
ExampleReading comprehension
Level 2 · Criterion
The specific skill
What the parameter breaks into. Each criterion carries its own score, so strength and weakness inside one subject stop cancelling each other out.
ExampleInferring motive from an unseen passage
Level 3 · Expectation
The written standard
A sentence describing what a child at this level actually does. The score follows the sentence — which is what makes a rating explainable to a parent.
ExampleExplains why a character acted, using evidence from the text
One assessment.
Four readers.
The same data, written four times, because a parent and a principal do not need the same sentence. Each report answers one question and stops.
Student report
How is this child doing, and what should we do about it?
- Score with percentile in cohort
- Radar across six criteria
- Chapter-wise breakdown
- Plain-English summary
- One suggested next step
Section report
What do I do with this class on Monday?
- Section mean, median and spread
- Band distribution with support line
- Criterion strength across students
- Students grouped by need, not rank
- Re-teach list by chapter
Class report
Is the variation between my sections acceptable?
- Section-by-section comparison
- Criterion × section heatmap
- Growth against baseline
- Chapter variance across sections
- Actions split by scope
School report
Where does next year's time and money go?
- School index with growth test
- Class × subject matrix
- Cohort tracking across years
- Benchmark against other schools
- Priorities by owner
Inside every report
Six things,
every time.
Consistent structure matters more than clever structure. A teacher who has read one IMPACT report can read every other one without relearning where anything sits.
Radar profile
Six criteria at a glance, overlaid with the class or school average.
Percentile
Position within the cohort, shown against the actual distribution.
Chapter relevancy
Every question tagged to a syllabus chapter, so gaps point to pages.
SWOT
Strengths, gaps, openings and risks, each tied to a number from the paper.
Written summary
Three or four sentences a parent can read without translation.
Recommendations
Actions with an owner and a cadence, not adjectives.
Your syllabus.
Your question paper.
IMPACT does not replace what you teach or ask you to adopt someone else's curriculum. Papers are built against your syllabus, in your languages, and tagged to your chapters.
MCQ
Objective items
Auto-scored, criterion-tagged, with distractor analysis that shows which wrong answer children chose and why it was tempting.
SJT
Situational judgement
Scenario items for reasoning, values and decision-making, scored polytomously rather than right or wrong.
OPEN
Written response
Extended answers scored against expectation statements, which is where most schools currently have no visibility at all.
RATE
Likert & multi-rater
Self, peer, teacher and parent instruments for the traits a subject paper cannot reach.
Bilingual papers
English, Bengali and Hindi, with the same criterion tags on both versions so results stay comparable across mediums.
Paper or screen
Runs on printed papers with a marks upload, or on devices. No school needs a device per child to start.
Your calendar
Baseline plus three cycles a year is typical, but the cycle count follows your term plan rather than ours.
Evidence, not impressions
Growth you can
defend in a
board meeting.
Baseline-to-endline movement is tested statistically, not asserted. Every figure in a school report carries its method, its sample and its limitation in writing.
DiD
Difference-in-differences where a comparison group exists.
Effect size
Reported alongside significance, so small real gains are not oversold.
Benchmark
Percentile against other schools running the same instrument.
What a term looks like
From paper to
Monday morning.
A full cycle runs in about two weeks. Most of the school's effort sits in the first week and is administrative rather than academic.
Week 0
Syllabus mapping
We map your chapters to parameters and criteria, and agree the expectation statements with your heads of department. Done once, reused every cycle.
Week 1
Paper and roster
Question paper built and reviewed by your teachers. Rosters uploaded by CSV, one link per section. No manual entry per child.
Week 2
Assessment window
The paper runs in a normal period. Marks uploaded in bulk or captured on screen; open responses scored against expectations.
Week 2 + 3 days
Reports issued
All four report levels arrive together, so the teacher, the coordinator and the head are reading the same finding in the same week.
Handling children's data
Your data
stays yours.
Assessment data on minors deserves a stated position rather than a paragraph in a contract. Ours is written down before a school signs anything.
School owns the record
Student data belongs to the school. Export in full at any time; deletion on request is honoured across backups.
Access by role
Parents see their child. Teachers see their section. Only named leadership roles see the school view.
No child ranked publicly
Reports are written for development. There is no leaderboard, and section reports name students only to their own teacher.
Stated limitations
Every school report carries what the method cannot show. A finding without its caveat is not a finding.
What schools
ask first.
How much extra work is this for my teachers?
One period for the assessment itself, plus roster upload and marks entry, which is bulk CSV rather than child by child. Open-response scoring is the only meaningful addition, and it replaces marking your teachers were already doing rather than adding to it. Syllabus mapping happens once, in Week 0, with your heads of department — after that it is reused every cycle.
Does this work with our board and our syllabus?
Papers are built against your syllabus and tagged to your chapters, so the architecture is board-agnostic. Schools running CBSE, ICSE and state boards use the same parameter–criterion structure with different chapter maps underneath.
Will this be used to rank or appraise our teachers?
Not by default, and we would advise against it. Assessment reports describe students. Observation reports describe teaching practice and are written as development tools with the teacher's own feedback attached. Where a school report shows teaching and student figures together, it states plainly that they are associated rather than causally linked.
We already have an assessment system. Why change?
You probably do not need to. Most schools running IMPACT keep their existing exam calendar and use it alongside — the difference is what comes back afterwards. If your current system already tells a Class 7 teacher which two grammar rules are costing the section most of its marks, IMPACT will not add much.
What happens to our data if we stop?
Full export in an open format, and deletion on request across backups. Nothing is retained for benchmarking without a separate written agreement, and benchmark data is aggregated so no school or child is identifiable in it.
Can we start with one class rather than the whole school?
Yes, and most schools should. A single class across one term is enough to see whether the reports change what teachers actually do, which is the only test that matters. Scaling after that is a mapping exercise, not a fresh implementation.
Bring one class.
One term.
See what your existing assessment data has been unable to tell you, before committing anything larger.
Typically a 30-minute call with your academic head. We bring a sample report built on your own syllabus.