TL;DR: Most Indian firms stay on a legal research platform they have outgrown because switching feels riskier than it is. It is not. Audit what you actually use, run a parallel trial on live matters, test ten authorities you rely on against the new tool, and keep a rollback plan until the parallel period ends. Nothing about the process requires losing a single citation.

On this page

Why firms stay on tools they have outgrown

The decision to leave a legal research platform is rarely blocked by the alternative being unproven. It is blocked by what the firm might lose in the move: years of saved searches, a junior associate’s muscle memory with a specific interface, a partner’s confidence that a citation typed into the old search bar will come back correct.

That anxiety is legitimate. A research platform sits underneath every matter a firm touches. If the switch goes badly mid-way through a filing deadline, the cost is not measured in subscription fees. It is measured in a missed authority or a citation format a registry rejects.

The anxiety is also, in most cases, disproportionate to the actual risk, because the risk is manageable with a checklist rather than a leap of faith. This piece is written for a firm or solo practice that has already decided the current platform no longer serves the practice well, whether on coverage, search quality, or cost, and wants to move without losing anything that matters. It assumes you have already done the comparison work. If you have not, our guide to choosing an Indian case-law search engine and our SCC Online vs Manupatra vs Indian Kanoon comparison cover that ground. This piece covers only the mechanics of the move itself.

Step 1: audit what you actually use before you move

Before you touch the new tool, write down what the old one actually does for your practice today. Not what it advertises. What you use.

Most subscriptions are sized for a use case the firm outgrew years ago. A partner who negotiated a firm-wide licence in 2019 may have bought a tier with tribunal coverage nobody in the current practice group touches. An associate may run three searches a week through a Boolean interface that a plain-English query, of the kind covered in our guide to AI legal research in India, would answer faster. The gap between the licence you pay for and the workflow you actually run is where switching decisions get made badly, because firms compare the new tool’s feature list against the old tool’s marketing page instead of against their own usage.

Concretely, before you shortlist anything:

Pull usage logs if your current vendor provides them. Some platforms show query volume, module usage, and per-seat activity to an account administrator. If yours does, get that report before you decide what you can afford to leave behind.

List the specific workflows that touch the platform daily. Citation verification before a filing. Precedent search for a new matter. Good-law checking on an authority a junior found. Statute cross-referencing against the current text on India Code, the government’s official repository of central legislation. Each of these is a separate capability, and a new platform may be strong on some and weak on others.

Identify who actually logs in. A firm-wide seat count often masks the fact that two partners and one associate account for 80 percent of the queries. Talk to them before anyone else about what they would miss.

Note anything niche that only the old tool provides. A specific tribunal’s decisions, an international law module, a document assembly feature bundled into the subscription. These are the items most likely to get forgotten in a move and most painful to discover missing three months later.

This audit is the input to every later step. Skipping it is why firms end up either overpaying for redundant coverage in the new tool or discovering a real gap after the old subscription has already lapsed.

Step 2: export saved searches and citation history

A research platform accumulates institutional memory: saved search folders, citation alerts, annotated case files, matter-tagged research trails. None of that memory transfers automatically to a new tool, and most of it cannot be recreated from scratch without redoing the underlying research.

Before you cancel anything, export what you can:

Saved searches and folders. If your current platform lets you export saved search queries or folder structures, do it now, even if you are not sure you will need them. A CSV or PDF export costs nothing to keep and a great deal to reconstruct later from memory.

Citation alerts and watch lists. If you have set up alerts for specific matters or authorities, note what they were tracking. You will need to recreate the equivalent monitoring in the new tool, and a list of what to recreate is worth writing down before the old dashboard disappears.

Annotated documents. Any highlighting, notes, or tags you have added to specific judgments inside the platform’s reader usually do not export cleanly. If a specific annotated case file matters for an active matter, screenshot or copy the annotations into your matter file directly, not into the platform.

Downloaded judgments already in your matter files. This is the one category that is usually safe. If your workflow has been to download and save PDFs of judgments into matter folders as you go, that archive is already outside the platform and does not depend on the subscription continuing. Case-status tracking through the National Judicial Data Grid and eCourts services is likewise independent of your research platform and does not need to move with the switch.

Whether your current vendor supports a clean export, and on what terms, is not something this piece can tell you in general. Export capability varies by platform and by plan tier, and it is not always documented publicly. Put the specific question to your account manager directly: what can be exported, in what format, and is there a time limit tied to your subscription’s end date. A reasonable answer names a format, a scope, and a deadline. A vague answer is a reason to start the export process earlier rather than later, before the conversation about cancellation puts any pressure on the vendor to be generous with your data.

Step 3: read the contract before you sign anything new

The contract questions in a platform switch fall into two categories: what you owe the platform you are leaving, and what you are committing to with the platform you are joining. Neither should be assumed from the sales conversation.

On the outgoing side, the specific things to check in your existing agreement, not to ask the vendor to summarise verbally: the notice period required to cancel without an auto-renewal charge, whether the contract is annual or multi-year, whether any portion of an unused term is refundable, and whether cancellation is required in writing through a specific channel. Auto-renewal clauses are the most common source of an unwanted extra year, because they typically require notice well before the renewal date, not on the renewal date itself. General principles of offer, acceptance, and termination under the Indian Contract Act, 1872 govern how a notice clause operates once you have signed, which is one more reason to read the actual clause rather than rely on what a salesperson describes.

On the incoming side, before you sign: what the minimum commitment period is, whether pricing is locked for that period or subject to change, what the seat count actually covers, and what happens to your data and your export rights if you decide to leave this platform too, down the line. A vendor with confidence in its product does not need a long lock-in to keep you; a short initial term with the option to extend is a reasonable ask, and it is fair to ask for it directly.

Where you cannot find a term published anywhere, whether the current vendor’s own site, your signed agreement, or the new vendor’s terms of service, do not assume an answer and do not accept a verbal summary as a substitute for the document. Ask for the specific clause in writing. If it genuinely is not published, that absence is itself useful information about how the vendor operates, and it belongs in your decision, not in a guess about what the term probably is.

Step 4: run a parallel trial on live matters

The single most useful step in a platform switch is running both tools side by side on real, current matters before you cut over fully. Not a demo. Not a sandbox with sample queries. The actual research questions your team is working on this week.

A parallel trial does two things a sales demo cannot. It tests the new tool against your firm’s actual query patterns, which are often narrower and more idiosyncratic than a generic demo covers. And it gives your team a real basis for comparison instead of a marketing claim, which matters enormously for adoption later: an associate who found the same authority faster in the new tool during a live matter will trust it far more than one who was simply told to switch.

Structure the parallel period deliberately rather than letting it happen informally:

Pick two or three active matters that represent your typical research load, ideally spanning at least one contentious matter and one advisory or transactional one if your practice covers both. For each research question that comes up on those matters during the trial window, run it through both platforms and note where the results diverged, not just whether the new tool “worked.”

Set a fixed length for the trial, long enough to hit a real range of query types but short enough to have a firm end date. Two to four weeks is typically enough to surface most gaps, though the right length depends on how frequently your practice generates new research questions.

Do not run the parallel trial only through the partner who championed the switch. Include at least one sceptic and at least one junior who will do the bulk of the day-to-day querying once the switch is final. Their friction points are the ones that will actually determine whether the new tool sticks.

Step 5: the coverage test, ten authorities

Corpus size claims from any vendor are, by their nature, aggregate marketing numbers that do not tell you whether the specific authorities your practice relies on are actually indexed, whether primary judgments or the secondary commentary discussed in our guide to primary versus secondary legal sources. The only way to know is to check directly, and the check is simple enough to run in an afternoon.

Pick ten authorities you cite regularly. Not ten random famous cases. Ten that come up in your actual practice: the leading Supreme Court decision on the point you argue most often, the High Court judgment your jurisdiction treats as settled, a recent tribunal order your practice group has been tracking, a statute section with its current amended text, and a mix of older foundational authority and something decided in the last twelve months.

For each one, run the same test in the new platform: can you find it by citation, can you find it by party name, and can you find it by describing the legal question in your own words without knowing the citation. Note whether the result links to the full judgment text or only a summary, and whether the platform tells you anything about the authority’s subsequent treatment, which is the specific check covered in our guide to good-law checking.

A platform that finds eight or nine of your ten authorities cleanly is workable, provided the two misses are not central to your practice. A platform that misses more than two or three, especially among the authorities you would consider foundational, has a coverage gap that no amount of search-quality improvement elsewhere will offset. This test matters more than any comparison table a vendor publishes, because it is built entirely out of what your practice actually needs, not out of what the vendor chose to measure.

Step 6: citation format compatibility with your court

A citation that reads correctly on screen is not automatically a citation your registry will accept in a filing. This step is easy to skip because it looks like a formatting detail, and it is the one most likely to cause a problem at the worst possible moment, mid-filing.

Since July 2023, the Supreme Court of India has assigned every judgment a neutral citation in the format YYYY INSC NNN at the point it is signed, independent of any private publisher, according to the Supreme Court’s own SCR portal, which now hosts the official neutral citation record. As of 2026, filings that carry the neutral citation alongside the traditional reporter citation are generally preferred by the registry. Traditional formats such as SCC and AIR remain widely used and cited in parallel.

Before you commit to a new research platform, check three things directly against your own filing practice, not against a general claim about the industry: does the platform display the neutral citation for judgments where one exists, does it display it in the format your registry expects it written, and does the platform’s citation search let you find a judgment when you only have the neutral citation, which matters when you are verifying a citation from an opposing party’s submission that used only the newer format. This is the exact problem Niyam’s citation handling is built to solve, resolving a case across neutral, reporter, and tribunal-specific formats rather than treating each as a different judgment. For the mechanics of the current citation formats, see our guide to citing Indian judgments correctly and our dedicated piece on neutral citations and e-SCR.

If your High Court has its own local citation convention or a specific preferred format for the bench you appear before, check the platform against that directly. General claims about “court-accepted formats” from a vendor should be read as claims, not guarantees, until you have confirmed the specific format against your own registry’s current practice.

Step 7: train the team before the old tool goes dark

The most predictable failure mode in a platform switch is not a missing feature. It is a team that reverts to old habits under deadline pressure because nobody built the new muscle memory before the old tool disappeared.

Training does not need to be elaborate, but it needs to happen before the cutover date, not after. A short session covering how the team’s five or six most common query types map onto the new interface is worth more than a full feature walkthrough nobody remembers a week later. If the new platform supports plain-English queries rather than the Boolean strings your team is used to constructing, that alone is worth a dedicated session, because it is a genuinely different way of asking a question and the instinct to fall back on old syntax is strong. Our roundup of AI legal research tools for Indian lawyers is a useful reference for what a plain-English workflow actually looks like in practice.

Identify one or two people per practice group who become the first point of contact for “how do I do this in the new tool” questions. This does not need to be a formal role. It needs to be someone who has actually run the coverage test and the parallel trial themselves, so their answers come from direct experience rather than the vendor’s documentation.

Build in a short overlap window where both platforms remain accessible even after the formal cutover date. A team member mid-way through an unfamiliar workflow should be able to fall back to the tool they know for that one search, rather than being forced into a mistake under deadline pressure because the safety net was removed on day one.

Step 8: assign a cutover owner

A platform switch that is “everyone’s responsibility” tends to become nobody’s responsibility, and the gap shows up as a missed export deadline or a junior associate who never got trained.

One person, not a committee, should own the cutover: they track the notice period on the outgoing contract, confirm the export is complete, run or supervise the coverage test, schedule the training sessions, and set the actual cutover date. This does not need to be the most senior partner. It is more often the person who will actually use the new tool daily and has the standing to chase other people for the things they need to do, such as attending a training session or confirming their saved searches are exported.

Give the cutover owner a single deadline to work backward from: the date the old subscription’s notice period requires cancellation to avoid an unwanted renewal. Every other step in this checklist should be scheduled to complete before that date, with enough margin to catch a problem in the export or the coverage test without being forced into a rushed decision.

Step 9: keep a rollback plan

Every migration plan should include the version where it does not go as planned, and a platform switch is no exception. The rollback plan is not a sign the switch is likely to fail. It is what makes it safe to commit to the switch at all.

The rollback plan has three parts. First, do not cancel the old subscription until the new platform has cleared the coverage test and the parallel trial, and until the team has had the training session. Cancelling early to save a month of overlapping cost is a false economy if it removes your fallback before you have confirmed the new tool actually works for your practice.

Second, define in advance what would trigger a rollback: a coverage gap discovered after cutover on an authority central to an active matter, a citation format issue the registry flags, or a team-wide inability to complete a common workflow within the first month. Naming the trigger conditions before you are under deadline pressure means the decision to fall back, if it comes to that, is made calmly rather than in the middle of a crisis.

Third, keep the exported saved searches, citation history, and coverage-test results from step 5 somewhere durable, outside either platform. If a rollback becomes necessary, that archive is what lets you resume the old workflow without having lost the institutional memory you were trying to protect in the first place.

A parallel-trial period that ends with a clean coverage test, a trained team, and a documented decision to proceed is not a leap of faith. It is a completed audit.

What to test before you sign vs what can wait

To test before you signCan wait until after
✓ Coverage test on your ten authorities✗ Full feature list walkthrough
✓ Citation format matches your court’s current practice✗ Every module in the platform, not just the ones you use
✓ Export format and notice period on the outgoing contract✗ Advanced settings and admin customisation
✓ Parallel trial on at least one active, real matter✗ Integration with tools your firm does not currently use
✓ Data privacy and confidentiality terms for research queries✗ Mobile app if your workflow is desktop-first
✓ Pricing terms and minimum commitment period✗ Every keyboard shortcut and interface preference
✓ Whether the sceptic on your team can find what they need✗ Cosmetic interface differences from the old tool

The full cutover checklist

Work through this in order. Each step assumes the ones before it are complete.

  1. Audit current usage: which workflows, which modules, which people actually use the existing platform.
  2. Export saved searches, citation alerts, and any annotated documents from the outgoing platform.
  3. Confirm the notice period, auto-renewal terms, and refund policy in the current contract.
  4. Confirm the minimum commitment, pricing lock, and export rights in the new contract before signing.
  5. Run the coverage test: ten authorities central to your practice, checked by citation, party name, and plain-English query.
  6. Confirm citation format compatibility, including neutral citation support, against your registry’s current practice.
  7. Start a parallel trial of at least two weeks on real, active matters, including at least one sceptical team member.
  8. Log every divergence between the two platforms during the trial: missed authority, wrong citation format, slower search.
  9. Assign a single cutover owner responsible for the deadline and every step above.
  10. Schedule team training before the cutover date, not after.
  11. Set the cutover date working backward from the outgoing contract’s notice deadline, with margin.
  12. Define rollback trigger conditions in writing before cutover.
  13. Keep the outgoing subscription active through the full parallel trial; do not cancel early to save cost.
  14. On the cutover date, confirm every team member can complete their most common workflow in the new tool.
  15. Cancel the outgoing subscription only after the rollback window has passed without a trigger condition firing.
  16. Archive the exported saved searches and coverage-test results outside both platforms, for long-term reference.

The parallel-trial timeline

flowchart LR
    A[Week 0: Audit + export] --> B[Week 1: Sign new contract, confirm terms]
    B --> C[Week 1-2: Coverage test on 10 authorities]
    C --> D[Week 2-4: Parallel trial on live matters]
    D --> E{Coverage and citation<br/>tests pass?}
    E -->|Yes| F[Week 4: Team training]
    E -->|No, gaps found| G[Extend trial or<br/>trigger rollback]
    F --> H[Week 5: Cutover date]
    H --> I[Week 5-6: Rollback window,<br/>old platform still active]
    I --> J{Any trigger<br/>condition fires?}
    J -->|No| K[Week 6+: Cancel old subscription]
    J -->|Yes| G

Common mistakes firms make when switching

Cancelling the old subscription before the parallel trial is complete. This removes the fallback exactly when it is most useful and turns every discovered gap into a crisis instead of a data point.

Running the coverage test on famous cases instead of the firm’s actual authorities. A platform that indexes every landmark constitutional case can still miss the specific tribunal order or district-level judgment your practice relies on weekly.

Letting one enthusiastic partner run the whole evaluation. A parallel trial that never includes a sceptic or a junior associate finds only the problems the champion happened to think of, not the ones the rest of the team will actually hit.

Treating the sales conversation as the contract. A feature described verbally by a salesperson is not the same as a term written into the agreement. Every claim about export rights, pricing lock, and notice periods belongs in the document you sign, not in your notes from a call. The same discipline applies to any AI feature the vendor demonstrates; see our guide on how to vet legal AI citation accuracy before you take a demo result at face value.

Skipping citation format verification until a filing is due. This is the single most avoidable failure mode, because the check takes minutes and the consequence of skipping it can cost you a filing deadline.

A citator that flags when an authority might have been distinguished or overruled since your last check is genuinely useful during a coverage test, because it lets you confirm not just that the new platform has a case but that it has kept up with what happened to it. Niyam’s citator surfaces that signal within its indexed corpus, which is one of the specific things worth checking during your own ten-authority test rather than taking on faith.

Frequently asked questions

Two to four weeks is typically enough to cover a representative range of query types for most practices, though the right length depends on how frequently your practice generates new research questions. A litigation practice with several active matters may generate enough queries in two weeks; a practice with fewer, more sporadic research needs may need closer to a month to get a fair sample.

What is the coverage test and how many authorities should I check?

The coverage test means picking a fixed number of authorities central to your actual practice, not famous cases in general, and checking whether the new platform finds each one by citation, by party name, and by a plain-English description of the legal question. Ten is a practical number: enough to be representative, few enough to complete in an afternoon.

Export capability varies by vendor and plan tier, and it is not always documented publicly. Ask your current account manager directly what can be exported, in what format, and by what deadline, before you begin the cancellation process. Do not assume a standard export exists just because other software categories commonly offer one.

The biggest risk is discovering a coverage gap or a citation format mismatch after you have already relied on the new platform for a filing. The coverage test and citation format check in this piece are designed specifically to surface that risk before cutover, on a timeline that does not coincide with an actual deadline. This is the same verification discipline covered in our guide on a lawyer’s duty to verify AI output.

Should I cancel my old subscription before or after the parallel trial?

After. Keep the outgoing subscription active through the full parallel trial and the rollback window that follows cutover. Cancelling early to save a month of overlapping cost removes your fallback exactly when a genuine gap is most likely to surface.

Check directly against your own filing practice rather than a general vendor claim. Confirm whether the platform displays the neutral citation (format YYYY INSC NNN for Supreme Court judgments, in effect since July 2023) alongside the traditional reporter citation, and whether you can search by the neutral citation alone. If your High Court has its own local convention, verify that specifically.

One named person, not a committee. The cutover owner tracks the outgoing contract’s notice deadline, confirms exports are complete, runs or supervises the coverage test, schedules training, and sets the final cutover date. This person does not need to be the most senior partner, but should be someone who will use the new tool daily.

What should trigger a rollback to the old research platform?

Define this in writing before cutover, not during a crisis. Reasonable trigger conditions include a coverage gap discovered on an authority central to an active matter, a citation format the registry flags as unacceptable, or a team-wide inability to complete a common workflow within the first month after cutover.

Not if those judgments are already saved into your matter files outside the platform, which is the safest place for anything tied to an active or closed matter regardless of which research tool you use next. Judgments accessed only through the platform’s own reader, without a separate download, are the ones at risk if your access ends before you have reviewed what you need.

Run short, workflow-specific sessions covering your team’s five or six most common query types rather than a full feature walkthrough. Identify one or two people per practice group as the first point of contact for questions, ideally people who ran the coverage test and parallel trial themselves. Keep both platforms accessible for a short overlap window after the formal cutover date.

Confirm the minimum commitment period, whether pricing is locked for that period, what the seat count actually covers, and what your export rights are if you later decide to leave this platform too. Where a term is not published anywhere, ask for it in writing rather than accepting a verbal summary, and treat an unwillingness to put it in writing as information in itself.

Notice period terms vary by vendor and are set out in your specific signed agreement, not in a general industry standard. Read your own contract for the exact notice period and auto-renewal terms rather than assuming a figure. If the term is unclear, ask your account manager for the specific clause in writing before you plan your cutover date around it.

What if the new platform fails the coverage test on one or two authorities but passes on the rest?

Weigh the specific misses against how central those authorities are to your practice, not just the pass rate. A platform that misses two authorities you cite occasionally is workable. A platform that misses even one authority your practice treats as foundational is a real gap worth raising directly with the vendor before you commit, since coverage sometimes expands after a documented gap is flagged. For a broader framework on weighing coverage claims, see our guide to choosing an Indian case-law search engine.

Can I run a parallel trial on a real, active matter without creating a conflict or confidentiality issue?

Running the same research query through two platforms does not itself create a confidentiality problem, but it does mean two vendors are receiving your query content. Check each platform’s data retention and confidentiality terms before you run live-matter queries through it during a trial, the same way you would before committing to it as your primary tool, consistent with the professional obligations set out in the Bar Council of India Rules. See our guide to choosing an Indian case-law search engine for the specific questions to ask about query retention and model training.

How much does switching actually cost, beyond the new subscription price?

The direct cost is usually a short period of running two subscriptions in parallel during the trial and rollback window, plus the time cost of training and the coverage test. For a broader view of what legal research software costs across firm sizes, which is useful context when budgeting for an overlap period, see our guide to legal research software cost in India.

What should go in the rollback archive if I do end up reverting to the old platform?

Keep the exported saved searches and citation history from step 2, the results of your coverage test from step 5, and a written note of exactly which authorities or workflows triggered the rollback. This archive is what makes a rollback fast rather than a second full re-evaluation, and it is also the starting point if you decide to try a different platform again later.

Switching platforms is a project with a defined start and end, not an open-ended risk. Audit what you use, export what you can, test the ten authorities that actually matter to your practice, run the parallel trial on real work, and keep the old subscription alive until the rollback window closes. None of that requires losing a citation, and all of it fits inside a few weeks of overlap most firms can absorb.

For the comparison work that should come before this checklist, see our guide to choosing an Indian case-law search engine and our SCC Online vs Manupatra vs Indian Kanoon comparison. For the cost side of the decision, see our guide to legal research software cost in India.