How to Build a Contract Repository (Step by Step)

Most Indian teams do not lose contracts to fire or flood. They lose them to a shared drive with three "final" folders, a departed employee's inbox, and a WhatsApp attachment nobody re-saved. A contract repository fixes that: a searchable, controlled store of executed agreements, built so any authorised person can find the right contract, the right version, in under a minute. It is not a full contract lifecycle management (CLM) system, and this guide (published by Adira, which sells CLM software, so we have a stake in you eventually needing more) is written to get you a working repository first, on a budget if needed.

You do not need procurement approval or a six-month project to start. You need a plan for six things: gather everything into one place, make scanned paper searchable, agree what data you track per contract, set folders and permissions, capture renewal dates before they capture you, and control versions so "final_v3" never wins an argument in court.

What a repository actually is, and is not

A repository stores documents and lets you find them by metadata, not by remembering which folder you put them in three years ago. It does not draft, redline, route for approval, or auto-populate a dashboard of what every contract obligates you to do; those are workflow and obligation-management jobs a mature CLM bundles on top. If you are deciding whether "repository" is even the right word for what you need, this explainer on what a contract repository is draws the line clearly.

Step 1: Gather everything, then de-duplicate

Pull every contract from every place it currently lives: shared drives, individual laptops, email attachments, the finance team's vendor folder, and physical files in a cabinet. Do this first, since your schema and naming convention should be shaped by what you actually have, not a template built for cleaner habits. Expect duplicates: the same MSA saved by three people under three names. Sort by counterparty and contract type, open near-duplicates side by side, and keep the one with a visible signature block and the latest amendment attached. Discard superseded drafts once the executed copy is confirmed safe.

Step 2: OCR your scans

A scanned PDF that is really just a photograph is dead weight: you cannot search it or pull text from it for a metadata field. Run every scanned contract through OCR (optical character recognition) so the text becomes searchable, and check the output against the original, since OCR reliably mangles tables and small print. This walkthrough on turning a scanned contract into searchable text covers free and paid OCR tools and where each fails. Once OCR'd, you can also highlight and markup the clauses that matter for free in Weave, without uploading the file anywhere, a useful last check before filing it away.

Step 3: Define the metadata schema before you file a single contract

Metadata is what turns a folder of PDFs into a repository: the fields you record about every contract so you can search, filter, and get alerted without opening each file. Define this schema once, in writing, before bulk-filing, since retrofitting it onto thousands of contracts later is far more painful than doing it up front.

Field	Why it matters	Example
Parties	Counterparty and your own contracting entity, if you have more than one	Acme Logistics Pvt Ltd / Your Co (Delhi)
Contract type	Lets you filter by MSA, NDA, lease, employment, vendor	Master Services Agreement
Value	Contract value or annual spend, for prioritising effort	INR 42,00,000 per annum
Effective date	When obligations start; anchor for term calculations	15 Jan 2026
Expiry date	When the term ends, or would end without renewal	14 Jan 2027
Renewal-notice date	The real deadline, calculated from the notice clause, not expiry	30 Nov 2026 (45 days before expiry)
Owner	The named person responsible, not "the team"	Priya Menon, Procurement
Governing law	Which state's law and courts apply	Laws of India; courts of Bengaluru

The renewal-notice date is the field most repositories get wrong, because it is easy to record the expiry date and assume that is the deadline. It usually is not: if a clause requires 45 days' written notice to stop an auto-renewal, your real deadline is 45 days before expiry, and a repository that only stores the expiry date lets that window pass silently. See how to track contract obligations and renewals for the fuller mechanics.

Step 4: Folder structure, naming, and access permissions

Pick one location, not five, and a naming convention every contributor follows: counterparty name, contract type, and effective date, in that order, so a folder sorts itself by filename alone (for example, AcmeLogistics_MSA_2026-01-15.pdf). Organise by counterparty or business function, not both, or people will file the same contract in different places depending on how they think about it that day.

Access permissions matter early. A vendor MSA with pricing terms should not be visible to everyone, and an employment contract certainly should not be. Set at least two tiers: a broad one that sees a contract exists and its non-sensitive metadata, and a narrower one, legal, finance, and the named owner, that can open the document and its commercial terms. If your tool lacks field-level restrictions, restrict at the folder level instead of leaving everything open by default.

Step 5: Capture renewal alerts as you file, not later

Do not treat renewal tracking as a separate project. As each contract is filed, calculate and record its renewal-notice date (not the expiry date) and set a reminder with real lead time: sixty days out for a decision, fifteen days out as a final check. Clean metadata with no reminder attached still lets deadlines pass. See tracking contract obligations and renewals for a full template.

Step 6: Version control, so "final_v3" never has to win an argument

Decide, in writing, what counts as the master copy: the fully executed, stamped version, not the last redline sent before signature. Every other version, drafts, redlines, unsigned copies, is a historical record and should be labelled as such (a /drafts subfolder works), never left sitting next to the master under a name that could be mistaken for it. When a contract is amended, file the amendment as a linked, separately dated document rather than editing the original PDF, and keep a simple version log at the top of the folder: date, what changed, who approved.

The India-specific part almost every repository gets wrong

Two statutes decide which version of a contract you must store, and one Supreme Court ruling decides whether your scanned copy holds up.

Store the stamped, executed version, not just any signed copy. Section 35 of the Indian Stamp Act, 1899 is blunt about what happens if you do not:

"No instrument chargeable with duty shall be admitted in evidence for any purpose by any person having by law or consent of parties authority to receive evidence, or shall be acted upon, registered or authenticated by any such person or by any public officer, unless such instrument is duly stamped." Source: Section 35, Indian Stamp Act, 1899, Indian Kanoon

In practice, an unstamped or under-stamped agreement can be refused as evidence until the deficient duty, and often a penalty, is paid. A repository holding only an emailed signature page, with no confirmation the instrument was stamped under your state's schedule, may be storing a document that cannot be enforced as-is. Make "confirm stamping" a checklist item before a contract is marked filed.

Track registration where relevant, separately from stamping. Stamping and registration are two different requirements, and a repository schema should not conflate them. Section 17(1) of the Registration Act, 1908 makes registration compulsory for certain categories of document, including:

"leases of immovable property from year to year, or for any term exceeding one year, or reserving a yearly rent" Source: Section 17, Registration Act, 1908, Indian Kanoon

A one-year-plus lease, or an instrument transferring an interest in immovable property worth Rs 100 or more, generally needs registration with the local Sub-Registrar, not just stamping and signature, to have full legal effect. Add a "registration required (Y/N)" field for any contract touching immovable property.

Digital storage is legally recognised, but only as a copy of a properly created original. Section 7(1) of the Information Technology Act, 2000 confirms that a legal retention requirement can be satisfied electronically:

"Where any law provides that documents, records or information shall be retained for any specific period, then, that requirement shall be deemed to have been satisfied if such documents, records or information are retained in the electronic form, if:" Source: Section 7, Information Technology Act, 2000, Indian Kanoon

This is the legal basis for treating your scanned repository as your working record day to day. But it does not make a scanned copy self-proving in court. In Anvar P.V. v. P.K. Basheer (2014) 10 SCC 473, the Supreme Court held that a certificate identifying how an electronic record was produced (then required under Section 65B of the Evidence Act, now the equivalent sits in Section 63 of the Bharatiya Sakshya Adhiniyam, 2023) is mandatory for admissibility, and oral testimony cannot substitute for it. Your filed scan is what you work from day to day, but someone needs to be able to certify how it was created and maintained if a dispute reaches court, which is why keeping the original signed paper alongside it is not optional.

Sources: Anvar P.V. v. P.K. Basheer, Indian Kanoon; Section 63, Bharatiya Sakshya Adhiniyam, 2023, Indian Kanoon.

Free vs paid options

You can build a usable repository for close to nothing, and many small teams should start there before buying software.

Approach	Cost	Good for	Where it breaks
Shared drive (Google Drive / OneDrive), strict folders and naming, spreadsheet as metadata index	Free	Under a few hundred contracts, one or two filers	No real search beyond filenames; one inconsistent filer breaks it for everyone
A structured database tool (Airtable, Notion) linking to files in the drive	Free to low-cost	Real metadata search and filtered views without CLM cost	Still manual entry; no automatic extraction or escalation
A dedicated contract repository or CLM	Paid	Role-based access, automatic alerts, audit trail, full-text search	Cost justified once volume or risk exceeds manual discipline

For a fuller comparison of the free tools standing in for each job a paid system bundles, see free CLM alternatives for small teams.

Red flags in a contract repository

Normal	Red flag	Why it matters
One canonical location, stamped executed copy clearly labelled master	Contracts scattered across personal drives, email, chat apps	No single place to check; a missing contract is noticed only in a crisis
A naming convention every contributor follows	Files named "final v2 FINAL(1).pdf", no party or date	Search becomes manual scrolling; duplicates go unnoticed
Metadata entered the same day a contract is filed	Metadata entered in a backlog, weeks later	Near-term deadlines can lapse before they are ever recorded
Renewal-notice date calculated from the clause, stored apart from expiry	Only the expiry date is stored	The real deadline, weeks before expiry, is missed
Access restricted by role for sensitive contracts	Every contract visible to everyone with a login	Pricing, salary, and settlement terms overexposed
A clearly labelled master copy, amendments filed as linked dated documents	Multiple similarly-named versions, no flagged master	Nobody can say with confidence which version governs

A clause that breaks your repository if it is written like this

Your metadata schema depends on the contract itself stating a clear effective date and term.

Bad: "This Agreement is effective as of the date of the last signature below and shall continue until terminated by either party."

What is wrong: there is no fixed calendar date for your effective-date field until you find the signature page, and no expiry date at all, since the term is open-ended. A repository built on this clause has an empty expiry field and nothing to calculate a renewal-notice date from.

Better: "This Agreement is effective from January 15, 2026 (the 'Effective Date') and shall continue for an initial term of twelve (12) months, expiring on January 14, 2027 (the 'Expiry Date'), unless earlier terminated in accordance with Clause 9 (Termination) or renewed in accordance with Clause 10 (Renewal)."

What changed: a stated calendar date and a defined, calculable expiry date turn two of your core metadata fields from a manual lookup into a direct copy from the contract text, and give you a fixed point to calculate the renewal-notice date against.

US and global contrast

US practice generally treats a signed PDF as a fully self-sufficient record once basic e-signature law (the federal ESIGN Act, or a state's UETA) is satisfied, with no separate stamping requirement, and registration reserved mainly for real property deeds. India layers two extra steps on top: stamp duty, enforced through Section 35's evidence bar, and compulsory registration for a narrower set of documents under Section 17. A repository copying a US-style "just store the signed PDF" template will quietly miss both.

FAQ

Do I need software to build a contract repository, or can I start with a shared drive? You can start with a shared drive and a spreadsheet, and many small teams should. Discipline matters more than the tool at low volume. Dedicated software earns its cost once volume or missed deadlines make manual discipline unreliable.

What is the difference between a contract repository and a CLM? A repository stores and organises executed contracts so you can find and track them. A CLM adds drafting, negotiation workflow, and approval routing on top. A repository is the foundation a CLM is built on, not a smaller version of the same thing.

Which version of a contract should I treat as the master copy? The fully executed, properly stamped version, not the last draft or redline exchanged before signature. Label it clearly and keep every other version visibly separate, ideally in its own subfolder.

Do I need to register every contract with the Sub-Registrar? No. Registration under Section 17 of the Registration Act, 1908 is compulsory mainly for documents affecting immovable property, including leases exceeding one year. Most commercial contracts, like an MSA or an NDA, only need stamping.

Is a scanned copy enough, or do I need to keep the original signed paper? Keep both where possible. Section 7 of the IT Act, 2000 lets electronic retention satisfy most legal requirements, but if authenticity is ever challenged in court, the Anvar P.V. v. P.K. Basheer standard means someone must certify how the electronic record was produced, easier when the original document still exists.

How many metadata fields should a small team start with? Start with the eight in this guide: parties, contract type, value, effective date, expiry date, renewal-notice date, owner, and governing law. Add fields only for a concrete recurring reason, not an exhaustive schema nobody keeps updated.

A repository built this way gets you a searchable record of what you have signed, and a defensible answer to "which version is real" and "when do we need to act." It does not tell you whether a specific document still needs registration, or whether a scan would survive a challenge in your own dispute; those are judgment calls this page cannot make for you. This is not legal advice. A qualified lawyer should confirm stamping and registration status for any high-value or property-related contract in your archive.