TL;DR: Most Indian legal research vendors do not publish prices at all, so any figure you see quoted second hand is unreliable. The subscription line is also the smallest part of the bill: the junior’s research hours, the duplicate subscription and the seat nobody logs into usually cost more. Work out your cost per matter, not your cost per seat, and the decision becomes arithmetic instead of argument.

On this page

The honest starting position on price

If you are trying to work out what legal research software costs in India, the first thing to accept is that the market does not want you to know. Of the major Indian legal research platforms, most publish no price at all. SCC Online’s Web Edition page offers a free trial and a contact route, with no figures. Manupatra routes buyers to a demo request. Westlaw Asia offers a free trial and nothing else. LexisNexis India asks you to connect with the sales team. All four positions were checked on 2 August 2026.

That is not an accident of web design. Quote-based pricing lets a vendor charge a Mumbai arbitration boutique and a Patna district court practitioner very different amounts for the same database, and it removes the reference point you would otherwise use to push back. It also means that any article confidently telling you “Manupatra costs X” is either quoting a stale page or making it up.

So this post does two things. It sets out the prices that are actually published, with the source and the date, and it gives you a model for working out what research costs your own practice, using your own numbers. The model matters more than the price list, because the subscription fee is rarely the biggest number in the calculation.

Two facts to anchor the rest. First, the Thomson Reuters 2025 Future of Professionals report, published on 26 June 2025 from an online survey of 2,275 professionals across legal, tax, trade, accounting and risk fields conducted in February and March 2025, found that respondents predicted AI would save professionals an average of 5 hours per week over the following year, up from 4 hours predicted in 2024. Second, the Thomson Reuters 2026 Report on the State of the US Legal Market records that technology spending at US law firms grew 9.7 percent and knowledge management costs 10.5 percent over 2024 levels. Both are worth reading with the caveat attached: the first is global and cross-professional, the second is US-only. Neither is an Indian benchmark, and neither should be treated as one.

What Indian vendors actually publish

Here is the complete set of published prices I could verify on 2 August 2026, with the page each came from. Where a vendor does not publish, the row says so rather than guessing.

ProductPublished priceSourceChecked
Indian Kanoon (search and read judgments)Free to useindiankanoon.org2 Aug 2026
Indian Kanoon Prism AI, Free planRs 0 per yearindiankanoon.org/prism/pricing2 Aug 2026
Indian Kanoon Prism AI, PremiumRs 500 per month, Rs 1,500 per quarter, Rs 2,700 half-yearly, Rs 5,000 per year, plus 18 percent GSTindiankanoon.org/prism/pricing2 Aug 2026
Indian Kanoon Prism AI, ProRs 1,500 per month, Rs 4,500 per quarter, Rs 8,000 half-yearly, Rs 15,000 per year, plus 18 percent GSTindiankanoon.org/prism/pricing2 Aug 2026
Indian Kanoon API, per requestSearch Rs 0.50, original document Rs 0.50, document Rs 0.20, document fragment Rs 0.05, document metainfo Rs 0.02, pre-paidapi.indiankanoon.org/pricing2 Aug 2026
SCC Online Web Edition Platinum Pack, annual subscriptionRs 33,500.00 (the listing states no GST position)EBC Webstore2 Aug 2026
SCC Online, other packsNot published, quote-basedscconline.com/web-edition2 Aug 2026
ManupatraNot published, quote-based (demo request)manupatra.ai2 Aug 2026
Westlaw Asia (India)Not published, quote-based (free trial route)westlawasia.com/subscriptions2 Aug 2026
LexisNexis IndiaNot published, quote-basedlexisnexis.com/en-in2 Aug 2026
Niyam, Advocate planRs 1,500 per month or Rs 15,000 per year, 1,800 credits per monthniyam.ai/pricing2 Aug 2026
Niyam, Professional planRs 3,000 per month or Rs 30,000 per year, 4,600 credits per monthniyam.ai/pricing2 Aug 2026
Niyam, Firm planRs 5,000 per month or Rs 50,000 per year, 10,000 credits per monthniyam.ai/pricing2 Aug 2026

Two observations about that table.

The first is how thin it is. Four of the largest names in Indian legal research publish nothing. If you want to compare them you have to run four separate sales conversations, and each vendor knows you cannot benchmark the quote you receive against anything public.

The second is that stale published prices are worse than none. A legacy Manupatra online payment page still carries a module price list, but the footnote reads “All prices are excluding Service Tax”. Service tax was subsumed into GST on 1 July 2017. A price list that has not been touched since before GST is not a price list, and quoting it as current would mislead you. That is why those numbers do not appear in the table above.

If you are still deciding which category of tool you need before you worry about price, the comparison in choosing an Indian case law search engine sets out what separates a citation database from a research system, and primary vs secondary legal sources covers what you should never be paying for at all.

Cost per seat is the wrong unit

Almost every vendor conversation is framed in cost per seat per year. That is convenient for the vendor because it scales with your headcount, which has nothing to do with how much research you do.

Consider two practices paying the same Rs 30,000 per seat per year. The first is a documentation-heavy conveyancing practice that closes 240 matters a year and researches almost none of them. Its research cost per matter is Rs 125. The second is a constitutional writ practice that files 30 petitions a year, every one of which needs a proper authority hunt. Its research cost per matter is Rs 1,000. Same invoice, eight times the unit cost, completely different decision about whether to renew.

Cost per matter is the unit that connects a software line item to the fee you actually charge. It also exposes the thing seat pricing hides, which is that a seat costs the same whether it is used every day or never opened.

The formula is not complicated:

Cost per matter = (annual software spend + annual research labour cost + annual error cost) divided by matters closed per year

The three numerator terms are the interesting part. The first is on your invoice. The second and third are not, and they are usually larger. The rest of this post works through all three for three practice shapes.

A note on how to read the scenarios: every rupee figure in them is an assumption I have invented for the purpose of the arithmetic, not a market rate. The tables mark each input as an assumption or a sourced fact so you can swap in your own numbers. There is no published Indian benchmark for what a district court advocate bills per hour, and inventing one would defeat the purpose of the exercise.

Scenario one: a solo practising in a district court

The profile: one advocate, criminal and civil matters in a district court, one clerk, no juniors. Around 90 active matters in a year, of which perhaps 25 involve a research question that goes beyond the bare provision.

The software line. A practitioner in this position has a genuinely free option that is not a compromise for a large share of the work. Indian Kanoon is free to search and read. The eCourts and NJDG portals are free for case status, and checking case status on eCourts and NJDG covers what those portals do and do not give you. Neutral citations from the Supreme Court’s e-SCR are free, and e-SCR neutral citations explains the format you will be expected to use.

So the honest software floor for a district court solo is Rs 0. Anything above that has to earn its place against free.

Above the floor, the published options are the Indian Kanoon Prism AI Premium plan at Rs 5,000 per year plus 18 percent GST, or the Pro plan at Rs 15,000 per year plus 18 percent GST, both from indiankanoon.org/prism/pricing as at 2 August 2026, and the Niyam Advocate plan at Rs 15,000 per year from Niyam’s pricing page as at the same date. SCC Online’s Platinum Pack at Rs 33,500 for an annual subscription, per the EBC Webstore listing checked on 2 August 2026, is a different order of spend for this profile.

The labour line. This is where the number moves. Suppose the 25 research matters take an average of 2.5 hours of your own time each. That is 62.5 hours a year. If you assume your own time is worth Rs 1,200 an hour, which is an assumption and not a benchmark, the labour cost of research is Rs 75,000 a year. That is five times the Advocate plan and more than twice the SCC Online Platinum Pack.

The error line. For a solo, the dominant error risk is arguing on an authority that has been overruled or distinguished into irrelevance, and discovering it in court. There is no clean way to price the reputational half of that. The measurable half is the rework: a wasted appearance, a fresh application, sometimes a costs order. If you assume one such event every two years costing Rs 20,000 in wasted work and travel, that is Rs 10,000 a year. Whether the number is right for you is a judgement call, but pretending it is zero is the one option that is definitely wrong. The mechanics of avoiding it are set out in good law checking.

The result. On these assumptions the solo’s total annual research cost is Rs 0 in software plus Rs 75,000 in labour plus Rs 10,000 in error, or Rs 85,000, across 90 matters. That is Rs 944 per matter, of which the software share is nil and the labour share is 88 percent.

Now add a paid tool at Rs 15,000 a year. Total goes to Rs 100,000, and cost per matter to Rs 1,111. The tool has to claw back more than 12.5 hours a year to break even at the assumed Rs 1,200 hourly value, which is 12 minutes saved on each of the 25 research matters. That is the whole test, and it is a test you can run in a fortnight rather than argue about in the abstract.

Scenario two: a five-lawyer litigation firm

The profile: two partners, one senior associate, two juniors. High Court and district court work, roughly 200 matters a year, of which 70 have a real research component.

The software line. This is where seat pricing starts to bite, because a five-lawyer firm is usually quoted per seat. Using the published figures: five Niyam Advocate seats at Rs 15,000 each would be Rs 75,000 a year, or the Firm plan at Rs 50,000 a year for 10,000 credits per month per Niyam’s pricing page as at 2 August 2026. One SCC Online Platinum Pack at Rs 33,500 covers one subscription; what five costs is not published, and firms are routinely quoted differently. The correct move is to ask for the multi-seat quote in writing and compare it against the single-seat list price you can verify.

The labour line. The 70 research matters do not consume partner time evenly. In most Indian litigation firms the first pass is done by a junior. Assume 5 hours of junior time per research matter and 1 hour of partner review. That is 350 junior hours and 70 partner hours a year.

Assume junior time is worth Rs 600 an hour and partner time Rs 4,000 an hour. Both are assumptions. The labour cost is Rs 210,000 for the juniors and Rs 280,000 for the partners, a total of Rs 490,000 a year.

Read that against the software line. Even at Rs 75,000 for five full-price seats, software is 13 percent of the firm’s research cost. The junior’s hours are 43 percent, and they are invisible because nobody invoices for them separately.

The error line. A five-lawyer firm has a different error profile from a solo. The risk is not usually the partner missing an overruling; it is the junior’s memo carrying a proposition the partner does not independently verify. Fabricated and misattributed citations are now a live category of this risk, which is why the duty to verify AI output and the checks in how to vet legal AI citation accuracy matter more at this size than at any other. If you assume two rework events a year at Rs 30,000 each in wasted associate time and client credits, that is Rs 60,000.

The result. Rs 75,000 software plus Rs 490,000 labour plus Rs 60,000 error is Rs 625,000 across 200 matters, or Rs 3,125 per matter. Software is 12 percent of it.

The lever that moves this number is not the subscription. It is the 350 junior hours. If a research system that produces linked source passages cuts the junior’s first pass from 5 hours to 3.5, the firm recovers 105 hours a year, worth Rs 63,000 on the assumed rate, which is most of the five-seat software bill. The practical differences between litigation and corporate research patterns are covered in AI legal research for litigators vs corporate lawyers.

Scenario three: a ten-person in-house team

The profile: a general counsel, three senior counsel, five counsel, one paralegal, inside a mid-size Indian company. Contract review, regulatory queries, a modest litigation docket managed through external counsel.

In-house economics are different in one decisive way: there is no billable hour, so research time has no revenue attached to it. That tempts finance teams to treat legal research software as pure overhead and cut it. The correct counter is that in-house research time has an opportunity cost and an outsourcing cost, both of which are real money.

The software line. Ten seats on a published Rs 15,000 per seat per year plan is Rs 150,000. The Niyam Firm plan at Rs 50,000 a year for 10,000 credits a month, per Niyam’s pricing page as at 2 August 2026, prices on consumption instead of headcount, which suits a team where two people do most of the research and eight do almost none. For traditional database vendors, ten seats will be quoted, not listed.

The labour line. Assume the team spends 600 hours a year across all ten people on legal research, and that the fully loaded cost of that time averages Rs 1,800 an hour. That is an assumption built from salary plus employer costs divided by working hours, and you can calculate the real figure for your own team in an afternoon from payroll. Labour cost: Rs 1,080,000 a year.

The outsourcing line. This is the in-house-specific term the other two scenarios do not have. Every research question the team cannot answer internally goes to external counsel at external counsel rates. If the team sends out 25 research questions a year at an assumed Rs 40,000 each, that is Rs 1,000,000. A tool that lets the team answer even six of those internally saves Rs 240,000, which is more than the entire ten-seat software line on the published Rs 15,000 per seat figure.

The result. Rs 150,000 software plus Rs 1,080,000 labour plus Rs 1,000,000 outsourcing is Rs 2,230,000 a year. If the team touches 400 matters a year, that is Rs 5,575 per matter, and software is 6.7 percent of it.

In-house teams also carry a constraint the other two do not: where the data sits and who processes it. A vendor whose answer to that question is vague is a procurement problem regardless of price, and legal AI data residency in India sets out what to ask.

The four hidden costs

Across all three scenarios, four costs sit outside the invoice and reliably exceed it.

1. The junior’s hours.

The largest single line in the five-lawyer scenario is 350 hours of junior research time. It is invisible because the junior is on a fixed salary, so the hours feel free. They are not. They are the most expensive part of the research process precisely because nobody measures them.

The test is simple and you can run it this month. Ask every junior to log research time against matter numbers for four weeks. Multiply the total by their fully loaded hourly cost. Most firms doing this for the first time find the number is between three and ten times their software spend. That ratio is my expectation based on the arithmetic above, not a surveyed figure, so treat it as a hypothesis to test rather than a fact to repeat.

2. The wrong-precedent risk.

The cost of relying on an authority that has been overruled, distinguished, or superseded by statute has three components: the wasted preparation, the adverse outcome or costs order, and the client relationship. Only the first is easy to price.

This risk has grown a category since generative AI entered legal drafting, because a fabricated citation looks exactly like a real one until you open it. AI hallucinated citations in India documents how this failure mode works and what courts have said about it, and the Supreme Court’s AI rules set out the disclosure position. The cost of one such incident, priced honestly, will usually exceed several years of any subscription in the table above.

3. The duplicate subscription.

This one is pure waste and it is common. A firm signs SCC Online at partner level. A senior associate separately expenses a second tool for citator work. A junior has a personal Indian Kanoon Prism subscription reimbursed as a professional expense. Nobody has looked at the three together, and the overlap is 70 percent.

The fix is an annual inventory: list every legal research and drafting subscription the firm pays for anywhere, including personal expense claims, with renewal date and named owner. Firms that have never done this usually find at least one line they had forgotten about.

4. The seat nobody uses.

Seat-based pricing charges you the same for a partner who runs 40 searches a week and a partner who has not logged in since onboarding. In a five-lawyer firm, one dormant seat on a Rs 15,000 plan is Rs 15,000 a year for nothing. In a ten-person in-house team with four dormant seats it is Rs 60,000.

Every vendor can tell you per-user login and query counts. Ask for that report before every renewal and make it a condition of the contract that you receive it quarterly. If a vendor will not give you usage data on the seats you pay for, that itself is information.

The cost model you can apply to your own numbers

Here is the whole model in one diagram.

flowchart TD
    A["Annual software spend in Rs"] --> D["Total annual research cost"]
    B["Research hours x fully loaded hourly cost"] --> D
    C["Error, rework and outsourcing cost"] --> D
    D --> E["Divide by matters closed per year"]
    E --> F["Cost per matter in Rs"]
    F --> G{"Compare against average fee per matter"}
    G -->|"Low share"| H["Research spend is not your problem"]
    G -->|"Middle share"| I["Normal: audit the hidden costs"]
    G -->|"High share"| J["Audit seats, duplicates and junior hours"]

To use it, fill in this table with your own figures. The right-hand column is the discipline: mark every input honestly as a sourced fact or an assumption, because a model built entirely of assumptions still tells you something useful about sensitivity, while a model that pretends assumptions are facts tells you nothing and feels authoritative doing it.

InputWhere it comes fromWorked example (five-lawyer firm)Assumption or sourced fact
Annual software spendYour invoices and expense claimsRs 75,000Sourced fact for you; the per-seat rate used here is the published Niyam Advocate price, Rs 15,000 per year, niyam.ai/pricing, 2 Aug 2026
Research matters per yearYour matter management system or diary70 of 200Assumption in this example, sourced fact for you
Junior hours per research matterFour-week time log5 hoursAssumption
Partner hours per research matterFour-week time log1 hourAssumption
Junior fully loaded hourly costPayroll divided by working hoursRs 600Assumption
Partner hourly valueYour own realised rateRs 4,000Assumption. No public Indian benchmark exists for litigation hourly rates
Rework events per yearYour own recollection, honestly counted2Assumption
Cost per rework eventWasted hours plus any costs orderRs 30,000Assumption
Matters closed per yearYour matter management system200Assumption in this example, sourced fact for you
Total annual research costSum of the three cost linesRs 625,000Derived
Cost per matterTotal divided by matters closedRs 3,125Derived
Software share of research costSoftware divided by total12 percentDerived

Three things to do with the finished number.

First, look at the software share. If software is under about a tenth of your total research cost, arguing about the subscription is a poor use of partner attention and the labour line is where the money is. That threshold is my own rule of thumb, not a published benchmark.

Second, run the break-even in minutes rather than rupees. A Rs 15,000 per year tool at an assumed Rs 600 junior hourly cost breaks even at 25 hours saved across the year. Divide by your research matters and you get a per-matter target. For the five-lawyer firm at 70 research matters, that is 21 minutes per matter. That is a claim you can test on real work in one billing cycle.

Third, test it on live matters rather than on a demo dataset. Take three questions from matters you have already closed, where you know the right answer, and run them through whatever you are evaluating. Count the minutes and check whether every citation opens to the case it claims to be. How to vet legal AI citation accuracy sets out a repeatable version of that test, and how to cite Indian judgments covers the citation format you should expect the output to produce.

For a consumption-priced tool, the arithmetic is even more direct. Niyam publishes its plans, so you can compute the break-even before you speak to anyone: Rs 15,000 a year for the Advocate plan with 1,800 credits a month, per its pricing page as at 2 August 2026. Whether that is good value depends entirely on the hourly figure you put in the model, which is why the model comes first and the price comparison second.

What no reliable Indian benchmark exists for

Being clear about the gaps is more useful than filling them with plausible numbers.

There is no free, reliable public benchmark for Indian legal billing rates. India Business Law Journal runs an annual billing rates survey, but it is not freely available and it covers corporate law firms rather than district court or High Court litigation practice. PayScale-style aggregators mix job titles and self-reported data in ways that make them unusable for this calculation. If you need an hourly figure for the model, derive it from your own realised fees divided by your own hours. That number is more accurate for your practice than any survey would be.

There is no published Indian benchmark for legal research time per matter. The Thomson Reuters figure of 5 hours per week saved through AI, from the 2025 Future of Professionals report, is a global cross-professional prediction from survey respondents, not a measurement of Indian legal research. Do not import it as an Indian number.

There is no published market-share or adoption figure for Indian legal research platforms that I can verify from a primary source. Vendor claims about user counts are marketing, not audited data.

There is one number that is public and worth knowing. The National Judicial Data Grid records over 5 crore cases pending across Indian district courts, comprising 1,13,36,273 civil and 3,90,06,045 criminal matters on the NJDG dashboard as accessed on 2 August 2026. That is a statement about the volume of work in the system, not about research cost, but it explains why speed of first-pass research has commercial value in Indian practice in a way it does not in a jurisdiction with a shorter docket.

For a picture of how Indian practitioners are actually adopting these tools, AI adoption among Indian lawyers in 2026 collects what is documented, and it is deliberately careful about what is not.

What to check before you sign

Whatever the price turns out to be, these are the contract terms that determine whether you get what you paid for. The columns compare a quote-based enterprise database subscription against a published-price consumption plan, on structural terms rather than on quality of content.

Term to checkQuote-based enterprise subscriptionPublished-price consumption plan
Price visible before you speak to sales✗✓
Price comparable against what another firm pays✗✓
Cost scales with usage rather than headcount✗✓
Per-user usage reporting available on request✓✓
Dormant seats still billed✓✗
Exit at short notice without penalty✗✓
Multi-year lock-in typical✓✗
Unused capacity carries forward✗✗
Price for a single user knowable in advance✗✓
Negotiation leverage from a public list price✗✓

The row that surprises people is the last one on carry-forward. Most consumption plans, including the published Niyam plans, do not roll unused credits into the next cycle, which means over-buying capacity is as wasteful as an unused seat. Size the plan to your actual monthly research volume, not to your busiest month.

Four questions to put in writing to any vendor before signing:

  1. What is the price for one seat, for five seats, and for ten seats, in writing, including GST treatment?
  2. Will you provide quarterly per-user login and query counts as a contractual term?
  3. What is the notice period to cancel, and what happens to data and saved research on exit?
  4. What is the update lag between a judgment being pronounced and it appearing in your database?

That fourth question is the one most likely to be answered vaguely, and it is the one that decides whether the subscription is fit for litigation work. Good law checking explains why a citator that is three months stale is worse than no citator, because it produces confidence rather than caution.

If you are comparing free tools against paid ones on capability rather than on price, free vs paid legal AI in India covers that comparison directly, and the best AI legal research tools in India sets out what each category of tool is actually for.

Frequently asked questions

There is no single answer because most Indian vendors do not publish prices. As at 2 August 2026, the verifiable published figures are Indian Kanoon Prism AI at Rs 5,000 or Rs 15,000 per year plus 18 percent GST, an SCC Online Web Edition Platinum Pack listed at Rs 33,500 for an annual subscription on the EBC Webstore, and Niyam plans at Rs 15,000, Rs 30,000 and Rs 50,000 per year. Manupatra, Westlaw Asia and LexisNexis India all require a quote.

Quote-based pricing lets a vendor charge different customers different amounts for the same product, and it prevents buyers from benchmarking. It is a commercial choice, not a technical constraint. The practical consequence for you is that you have no reference point when a quote arrives, which is why deriving your own cost per matter first gives you a defensible position in the conversation.

What does SCC Online cost per year?

The SCC Online Web Edition Platinum Pack is listed at Rs 33,500.00 for an annual subscription on the EBC Webstore, checked on 2 August 2026. The listing does not state the GST position. Other SCC Online packs and multi-seat arrangements are not published, and SCC Online’s own website directs buyers to a free trial or a contact form rather than a price list.

Is Indian Kanoon free?

Searching and reading judgments on Indian Kanoon is free. Its Prism AI research tools are sold separately, with a Free plan at Rs 0, a Premium plan at Rs 5,000 per year and a Pro plan at Rs 15,000 per year, both plus 18 percent GST, as published on the Prism pricing page on 2 August 2026. The Indian Kanoon API is separately priced per request, starting at Rs 0.02 for document metainfo.

What is cost per matter and why does it matter more than cost per seat?

Cost per matter is your total annual research cost, including software, labour and error, divided by matters closed in the year. It matters more than cost per seat because it connects the spend to the fee you charge. Two practices paying the same per seat can have unit costs that differ by a factor of eight, depending on how research-heavy their work is.

How do I work out what my own time is worth for this calculation?

Take your realised fee income for the last full year and divide it by the hours you actually worked, not the hours you billed. For salaried lawyers, take salary plus employer costs and divide by working hours in the year. Neither figure is a market rate and neither should be quoted as one, but both are accurate for your own practice, which is what the model needs.

Is there a published benchmark for Indian lawyers’ hourly rates?

Not one that is freely available and applicable to litigation practice. India Business Law Journal runs an annual billing rates survey covering corporate law firms, which is not free to access. Salary aggregators mix job titles and self-reported data. For the purpose of a cost model, derive the figure from your own realised fees rather than importing someone else’s number.

Not automatically. Indian Kanoon, eCourts, NJDG and the Supreme Court’s e-SCR are free and cover a large share of district court research. The case for paying arises when a paid tool saves enough time to beat its own cost. On an assumed hourly value of Rs 1,200, a Rs 15,000 per year plan breaks even at 12.5 hours saved across the year. Test that on real matters before committing.

List every legal research, citator and drafting subscription the firm pays for anywhere, including reimbursed personal expense claims, with renewal date and named owner. Compare coverage across them. Overlap between a general database, a citator and an AI research tool is common and often runs above half. Do this before each renewal cycle, not after.

What is a dormant seat and how much does it cost?

A dormant seat is a licensed user who does not log in. Seat-based pricing charges the same for a heavy user and a dormant one. One dormant seat on a Rs 15,000 per year plan costs Rs 15,000 a year for nothing. Ask your vendor for per-user login and query counts before every renewal, and make quarterly reporting a contract term.

On the published Niyam plans, no. Unused credits do not carry into the next billing cycle. Check this for any consumption-priced tool before buying, because over-buying capacity to cover a busy month wastes money in every other month. Size the plan to your typical monthly research volume and top up when a heavy month arrives.

In-house research time has no billable rate attached, so the argument has to run through opportunity cost and external counsel spend. Count the research questions your team sends to external counsel in a year and price them. If a tool lets the team answer even a fraction of those internally, the saving usually exceeds the entire software line. That is a number finance teams accept because it appears in the external legal spend budget.

The research hours of juniors and associates, because they sit inside a fixed salary and nobody invoices them separately. In the five-lawyer worked example, junior hours account for 43 percent of total research cost against 12 percent for software. The way to find your own figure is a four-week time log against matter numbers, multiplied by fully loaded hourly cost.

How much does a wrong precedent actually cost?

The measurable part is the rework: preparation thrown away, an extra appearance, sometimes a costs order. The unmeasurable part is the client relationship and, where a fabricated citation is involved, the professional exposure. Price only the measurable part in your model and treat the rest as a reason to be conservative. One serious incident will usually exceed several years of any published subscription price.

Does a cheaper tool mean a smaller corpus?

Not reliably, and price is a poor proxy for coverage. The questions that matter are which courts are covered, how far back the coverage runs, and what the lag is between pronouncement and availability. Ask for those three answers in writing. A tool that is current on High Court judgments is more useful for litigation than a larger corpus that lags by months.

How do I compare a quote against a published price?

Convert both to cost per seat per year including GST, then to cost per matter using your own matter volume. A quote that arrives as a bundle with modules and multi-year terms is deliberately hard to compare, so break it into a single annual per-seat figure before you evaluate it. Having one published price in the market, whatever it is, gives you a reference point in that conversation.

What should I test during a free trial?

Take three questions from matters you have already closed, where you know the correct answer, and run them. Time each one. Then open every citation the tool produces and confirm it is the case it claims to be and that it is still good law. A trial spent exploring features tells you nothing about cost per matter; a trial spent on closed matters gives you the minutes-saved figure the model needs.

Where does data residency fit into the cost calculation?

It is a constraint before it is a cost. For in-house teams and firms handling regulated client data, a vendor that cannot state where data is processed and whether it is used for model training may be unusable at any price. Settle that question before you compare figures, otherwise you may spend a procurement cycle pricing an option that cannot be approved.

Run the model on your own numbers before your next renewal date. Log four weeks of research time against matter numbers, count your matters, and calculate the cost per matter. If the software share comes out under a tenth, the renewal conversation is not where your money is.