Due diligence has always been an information problem. A buyer must determine, within a compressed timeline and from a curated data room, whether a target company’s reported performance reflects economic reality. The data room may contain 5,000 to 50,000 documents. The general ledger may contain 10 million transactions. The contract portfolio may span 500 agreements across customers, suppliers, employees, landlords and regulators.
The question is not whether technology can help. It is which technology answers which diligence question, what evidence each tool produces and where human judgment remains irreplaceable.
Most discussions of technology in due diligence fall into two traps. The first treats technology as a revolution that will automate analysis and eliminate human error. The second treats it as irrelevant because experienced professionals do not need software to read financial statements. Both are wrong. Technology is a force multiplier that makes experienced professionals faster, more comprehensive and more reliable, without replacing the judgment that diligence ultimately requires.
A diligence team without technology examines what it can reach within the time available. A diligence team with the right technology examines the full population and directs its judgment to the items most likely to affect the transaction. The difference is not the quality of judgment. It is the efficiency of its application.
The Six Categories That Matter
1. Virtual data rooms and document management
The data room is the information architecture of the transaction. The platform determines how documents are organised, how access is controlled, how activity is tracked and how efficiently the diligence team navigates.
Diligence question answered: What has management provided, what is missing and who is accessing what?
Evidence produced: Document indices, access logs, Q&A trails, version histories. The activity log is itself a diagnostic: documents uploaded late, questions answered slowly and areas with incomplete documentation signal where management is least comfortable.
Limitation: The data room organises information. It does not evaluate whether the documents are complete, accurate or internally consistent.
2. ERP and financial data extraction
Tools that extract transaction-level data from the target’s ERP and structure it for analysis. The extraction converts millions of GL entries, sub-ledger records, purchase orders, invoices and journal entries into a dataset that analytical tools can process.
Diligence question answered: What does the underlying transaction data show that the financial statements aggregate away?
Evidence produced: Complete general ledger extracts, sub-ledger detail (AR, AP, inventory), journal entry databases, user logs and transaction metadata. A Rs 500 crore revenue figure in the P&L becomes 200,000 individual transactions, each with a customer, an amount, a date, a user and a description.
Limitation: The extraction moves data. The interpretation, identification of anomalies and assessment of patterns requires analytical methodology and professional judgment.
3. Data analytics and anomaly detection
Tools that analyse extracted transaction data for statistical anomalies, unusual patterns and potential control failures. Journal entry analytics (Benford’s Law, round-number analysis, weekend posting detection, period-end concentration). Receivable and payable analytics (ageing, concentration, related-party indicators). Procurement analytics (vendor substance, duplicate payments, split transactions below approval thresholds).
Diligence question answered: Which transactions, out of millions, should the investigator examine first?
Evidence produced: Ranked exception lists and statistical profiles. A journal entry analytics run on 5 million entries may flag 300 that meet five or more anomaly criteria simultaneously. Those 300, representing 0.006% of the population, are the starting investigation set. Without analytics, the diligence team samples 200 entries selected randomly and may miss every one of the 300.
Limitation: Analytics identifies anomalies. It does not determine whether an anomaly is problematic. A manual journal posted by the CFO at 11pm on a Sunday affecting revenue may be a legitimate close adjustment or a manipulation. The analytics identifies. The investigator determines.
4. AI-powered document review and contract analysis
Natural language processing tools that read, classify, extract and summarise documents at scale. Contract analysis platforms identify key clauses: change-of-control provisions, termination rights, pricing mechanisms, guarantees, indemnities and IP assignments.
Diligence question answered: Across 500 contracts, which contain provisions that affect the transaction, the valuation or the post-close operations?
Evidence produced: Clause extraction tables and risk flags. A 500-contract portfolio screened for change-of-control provisions in hours, producing a list of the 35 contracts whose terms may be triggered by the acquisition. Without AI screening, identifying all 35 requires reading all 500, a task that time pressure may truncate.
Limitation: AI identifies a clause. It cannot assess whether the counterparty will exercise the right, what the revenue impact would be, or whether the risk is mitigable. Commercial judgment and relationship intelligence remain human functions.
5. Entity resolution and related-party identification
Tools that cross-reference entity data (directors, shareholders, addresses, contact information, bank signatories) across the target’s vendor, customer, employee and group-entity databases to identify hidden connections.
Diligence question answered: Are any of the target’s counterparties connected to its promoters, directors or key management?
Evidence produced: Network maps showing connections based on shared attributes. A cross-reference of the vendor master against director databases and employee records may identify 15 vendors sharing a director, address or contact detail with the company. Without the cross-reference, these connections are invisible in any single database.
Limitation: A shared registered agent or PIN code may be coincidental. A shared director who is the promoter’s relative is significant. The tool identifies connections. The investigator assesses meaning.
6. Financial modelling and model audit
Platforms that build, audit and stress-test financial models. Model audit tools examine formula integrity, circular references, hard-coded values, external links and inconsistencies.
Diligence question answered: Is the model mechanically correct, and what happens to the valuation under adverse scenarios?
Evidence produced: Formula audit reports, sensitivity tables, scenario matrices. A model audit may discover that terminal value accounts for 75% of enterprise value, that three hard-coded overrides affect the EBITDA calculation by Rs 8 crore, and that a circular reference in the debt schedule produces different results depending on which computer opens the file.
Limitation: The tool tests whether the model calculates correctly. Whether the inputs are commercially reasonable is a judgment requiring market knowledge, competitive intelligence and forensic scepticism.
The Integration Principle
In Northrop Management Private Limited’s due diligence practice, technology is deployed as an integrated stack where each tool feeds the next.
Data room → ERP extraction → analytics and anomaly detection → AI document review → entity resolution → model audit → forensic investigation of flagged items
The output of each layer becomes the input for the next. No single tool produces a diligence conclusion. The conclusion is the product of technology-enabled analysis and professional judgment applied together.
The financial impact of the technology stack is measurable. A diligence team using the full stack identifies, on average, 30% to 50% more findings than one relying on manual methods alone. More importantly, the technology-identified findings tend to be the ones with the largest transaction impact: the Rs 180 crore of revenue at risk from change-of-control provisions, the Rs 15 crore of vendor payments to connected entities, the Rs 8 crore of model errors affecting EBITDA.
Ashish Chaudhary, frames the technology principle directly: “The question is not which software to use. It is which diligence question each technology helps answer and what evidence it produces. A data-room containing 20,000 documents can be screened for change-of-control provisions, unusual obligations and related-party references before detailed human review begins. That screening does not replace the review. It ensures the review starts with the documents that matter most.”
Questions for the Boardroom
- Does our diligence process deploy analytical technology to examine the full transaction population, or does it rely on sampling that may miss the most consequential anomalies?
- Have we cross-referenced the target’s vendor and customer databases against director, shareholder and employee records to identify hidden connections before approving the transaction?
- Has the target’s contract portfolio been screened for change-of-control and termination provisions, and what is the total revenue contractually at risk upon acquisition?
- Has the financial model been independently audited for formula integrity, hard-coded values and circular references, and what errors were found?
- For every technology-generated flag, has a qualified professional assessed the commercial significance and determined whether the finding affects the transaction price?
Practical Checklist
Before opening the data room: Deploy ERP extraction tools. Request full GL, sub-ledger and user-log exports from the target.
Within 48 hours: Run journal entry analytics (ten-filter triage) and entity-resolution cross-reference across vendor, customer and employee databases.
Within the first week: Complete AI contract review for change-of-control, termination, guarantee and indemnity clauses. Complete model audit.
Throughout the engagement: Feed anomaly findings into the investigation queue. Validate every flag with primary documentation.
Before signing: Confirm that every material flag has been investigated, every model error has been corrected and every finding has been quantified in transaction-impact terms.
Closing Implication
Technology does not replace diligence judgment. It accelerates the identification of the transactions, contracts, connections and model assumptions that judgment should be applied to. The board that insists on technology-enabled diligence gets findings that manual diligence alone would miss. The board that accepts manual-only diligence accepts the risk that the most consequential finding may be in the transaction, the contract or the connection that the team did not reach before the deadline.