An IT due diligence is conducted under constraints the examiner does not choose: time limited by the timetable of the deal, documents gathered by the seller and restricted access to the target’s teams. It must nevertheless inform an investment decision which, for its part, will be taken on the planned date.
The expression refers to the examination of a company’s information system before its acquisition or an equity investment. French usage renders it as “audit d’acquisition” or “vérifications préalables”. The FranceTerme database (culture.fr/franceterme, entry read on 12 November 2026) gives “audit préalable” as the equivalent of due diligence in the field of finance, with the reference “Journal officiel du 28/12/2006”. The examination looks into five subjects for the investment committee: the state of the application estate and its debt, dependency on individuals or suppliers, security, contracts and licences, and the system’s capacity to be integrated or separated. Each answer can weigh on the price, on the sale agreement or on the budget for the months following the deal.
What state are the application estate and its debt in?
The first document requested is the inventory: applications, versions, hosting, interfaces, the share developed in-house and the share bought in. Its availability is already telling. A target that keeps its IT landscape map up to date passes it on as it is; a target that reconstructs it for the occasion may suggest that its own decisions were taken without it.
The examination then notes what makes future changes more expensive or more fragile: components that no longer receive patches, outdated versions, no automated tests, manual production releases, missing documentation. These forms of technical debt each have their cost. A debt ratio taken from an analysis tool is not used without its calculation conventions, detailed in our article on how to measure and reduce technical debt.
The target may also be running a structural project, a redesign or a change of packaged software, whose completion the business plan assumes. Its real state is established from documents: what has been delivered and verified, the remaining work estimated by those doing it, the suppliers’ commitments.
The answer translates into an upgrade budget, to be included in the acquirer’s business plan. It can provide a price argument when the seller presented the system as ready to support the announced growth.
Does the system rest on a few people or on a supplier?
Dependency first concerns people: a module that only one developer knows how to modify, operations whose procedures are written down nowhere, a founder who remains the de facto architect. The version control history shows who modified what and since when; the on-call rota shows who gets called at night.
It also concerns suppliers. Heavily customised packaged software ties the target to its vendor or integrator. Maintenance entrusted to a provider, as third-party application maintenance, places knowledge of the application with that provider; the contract does or does not provide the conditions for handing it back.
Depending on the answer, the acquirer prepares a dated knowledge transfer, costs the replacement of a supplier or a skill, and examines with its advisers the commitments to request from key people. A dependency with no known remedy can also lead to reducing the scope of the deal, or abandoning it.
Has security been tested, or only described?
A security policy describes an intention. The examination looks for traces of a practice: penetration test reports and the follow-up given to findings, patch application times, review of privileged accounts, the log of past incidents and how they were handled.
Continuity is judged in the same way. The target may display recovery objectives, in the sense of our RTO RPO entry; the evidence expected is the report of the last restoration test, with the time observed. Our page on cyber resilience describes these tests and the mapping of the third parties on which an activity depends.
When the target processes personal data, its GDPR compliance documentation, Regulation (EU) 2016/679, joins the documents requested; its assessment is for the acquirer’s lawyers.
A due diligence does not always give the right to test: what could not be verified is written down as such. The answer sets the measures to carry out before connecting the two companies’ networks and the tests to run as a priority after the deal. How the sale agreement handles a past incident that the seller did not disclose is a question to put to the legal advisers.
Do the contracts and licences withstand the change of control?
The inventory of contracts covers licences, software-as-a-service subscriptions, hosting, maintenance and services. Three readings apply, all to be carried out with a lawyer.
A contract may contain a clause linked to a change of shareholder or to the sale. The scope of a licence (number of users, entities, countries) may no longer fit once the target joins a larger group. Finally, renewal dates and termination windows may fall in the months following the deal; monitoring them is among the practices described in our article on the contractual risk of large projects.
Ownership of the software developed by the target is checked in the same way. Who wrote it: employees, service providers, freelancers? What do their contracts provide regarding rights? Which open source components does the code include, and on what conditions? The IT examination establishes the facts; the legal qualification and the warranties to negotiate in the sale agreement are for the legal advisers.
These readings result in remediation costs to be estimated and a list of points handed to the negotiators.
Can the system join the acquirer’s, or detach from its group?
The question depends on the acquirer’s intention: merge the two systems, let them coexist, or connect only a few flows. In the first case, the examination looks at master data (customers, products, chart of accounts), documented interfaces and the applications’ ability to expose their data. The target architecture and the trajectory are then built according to the approach in our page on IT architecture.
When the target is a subsidiary or a business sold by a group, the examination covers what it shares with the group: network, directory, email, a common ERP, licences taken out by the parent company, pooled teams. Each shared service will have to be rebuilt, or provided by the seller during a transition period whose duration, scope and price are to be settled in the contract, with the acquirer’s advisers.
The answer gives a cost and a timetable for integration or separation. If they are missing from the business plan, the deal is valued without one of its expenses.
What does an IT due diligence hand to the committee, and what does it leave open?
We propose a short format. Each finding carries one of three labels: verified against documents, declared by the target, not verified for lack of access. Each risk is classified according to the action it calls for: calling the deal into question, adjusting the price, requesting a clause, or adding an action to the post-deal plan. Estimates are given as ranges, with their assumptions.
Cross-checking, verifying against documents, reporting options: this method is the one our page on project audit describes for a project or programme, where the sponsor opens access themselves. Here, access limits the conclusion.
The list of points not verified therefore does not expire at signing. It becomes the verification programme for the first weeks, when the target’s teams and environments are finally accessible.