Compare
Lekhya vs a manual, sample-based audit.
Not a claim that manual testing is wrong — it is how almost every audit in India is still run. This is what changes, stage by stage, when the testing runs across the full population instead.
| Stage | Manual, sample-based | Lekhya, full population |
|---|---|---|
| Getting a testable population | A junior spends a day cleaning the client's export by hand before testing can start. | The population is validated against the ledger and trial balance as it's ingested; defects surface immediately rather than mid-testing. |
| Selecting what to test | A sample, sized to a judgment call about risk and time available this week. | The full population is tested; rare but material exceptions do not depend on landing in the sample. |
| Performing the test | Vouching and matching performed by hand, at the speed and consistency of whoever is doing it that day. | Matching and reconciliation run as rule-based computation, producing the same result on every run. |
| Documenting an exception | Written up from memory a few days after it was found, with the supporting document re-located from a shared folder. | Raised with its evidence already attached at the moment the exception is found. |
| Review | A reviewer re-performs parts of the testing to trust the conclusion, because the trail back to evidence is thin. | A reviewer opens the finding and the document behind it directly, without re-performing the underlying computation. |
| What the auditor can say | "We tested a sample and found no exceptions in it" — a statement about the sample. | "We tested the population and these are the exceptions in it" — a statement about the population. |
What does not change: materiality, what counts as a misstatement, and whether an exception matters are still the engagement team’s judgment. Full-population testing changes what evidence that judgment is formed on — it does not replace the judgment itself.
Read next
How the split between computed, drafted and auditor-decided actually works.
A full walkthrough of what Lekhya computes deterministically, what a model drafts for review, and what never leaves the auditor’s hands.