How Much Does eDiscovery Software Cost in 2026?

The bands a litigation data build falls into

Firms size this by matter volume. The budget follows two other things: how many source systems you collect from, and whether you are building around a review platform or attempting to replace one.

• The layer around review, $120,000 to $250,000, 16 to 22 weeks. Legal holds as ongoing objects with custodians drawn from your directory, acknowledgement tracking with escalation, evidenced reminders and a recorded release. Collection orchestration recording source, method, date range, custodian, operator and hashes. A pre hosting assessment layer running search term reports, threading, domain and date distributions and cross custodian deduplication. And a matter cost model that shows a promotion decision in currency at the moment it is made.
• Full platform including review, $400,000 to $900,000, 10 to 18 months. Adds coding panels, redaction, privilege logging, and production as a validated pipeline with pre release gates.

We will be blunt about the boundary. Reproducing what Relativity, Everlaw and DISCO do, including their analytics, their scale and their years of hardening, is a multi year programme well beyond the upper band and is almost never the right decision. The one situation that justifies it is a jurisdiction or client requirement that forbids data leaving your infrastructure, where hosting somewhere acceptable stops being optional.

What drives a litigation data build up

• Source system connectors, $12,000 to $35,000 each. Microsoft 365, Google Workspace, Slack, Teams and mobile forensic tooling are five separate integrations with five different export shapes and five different consent and permission models. Chat is materially harder than mail because a conversation has no natural document boundary.
• Data volume, $30,000 to $90,000. Terabyte scale text extraction, indexing and deduplication on your own infrastructure is genuine engineering rather than configuration, and it is the line most underestimated when a build is scoped around a feature list.
• Multi jurisdiction handling, $25,000 to $70,000. Where personal data cannot be exported for review, the architecture changes rather than the settings, and the residency requirement will drive cost more than any feature.
• Production pipeline, $45,000 to $110,000. Image format with load files, native production for spreadsheets, a Bates scheme that cannot collide across volumes, redactions burned correctly, confidentiality designations under a protective order, and automatic gates before release.
• Analytics or assisted review in house, $80,000 upward. A specialist workstream in its own right, and one where the purchased platforms have a substantial head start.

What keeps the number down

• Keep hosting review where it is. Build holds, collection and assessment, and continue to promote into your existing platform. This is what we recommend to most firms and it removes the largest cost driver in the category.
• Start with two source systems. Whatever accounts for most of your collections. The third and fourth connector cost roughly the same each, so add them once the chain of custody model is proven.
• Consume rather than rebuild processing. If a processing engine already handles your difficult formats, orchestrate it rather than replacing it. Processing is a solved and heavily invested problem.
• Cull before you host, not after. Date filtering, custodian scoping, domain analysis, cross custodian deduplication, threading to suppress inclusive duplicates and removal of known system files can take a large share out of a corpus before it reaches per gigabyte pricing. The assessment layer is cheap relative to what it prevents.
• Scope the hold module first. It is the smallest piece, it removes the preservation exposure that keeps general counsel awake, and it can go live long before the rest.

A worked example that adds up

A corporate legal department with roughly 60 live matters, collections spanning Microsoft 365 and Slack, review hosted with an existing platform under an outside counsel arrangement, holds currently tracked in a spreadsheet.

• Discovery, hold policy and preservation mapping across source systems: $14,000
• Legal hold objects with directory sync, acknowledgement escalation and recorded release: $36,000
• Custodian questionnaires and a maintained data map: $16,000
• Collection orchestration with chain of custody, hashing and an evidence log: $42,000
• Microsoft 365 and Slack connectors: $32,000
• Pre hosting assessment with search term reports, threading and distributions: $34,000
• Matter cost model showing promotion cost in currency: $22,000

Total $196,000, in the upper half of the recommended band. The line that repays fastest is the assessment layer, because a partner looking at a screen that says promoting this custodian set adds a specific monthly figure to the matter will scope differently from one who is told the total six weeks later on an invoice.

How the spend phases

• Discovery and preservation mapping, 6 to 10 percent. Which systems hold what, and which retention policies have to be suspended when.
• Legal hold and custodian management, 18 to 24 percent. Deliver this first and independently. It has value on its own.
• Collection orchestration and chain of custody, 20 to 26 percent.
• Source connectors, 14 to 20 percent. Split by how many systems you genuinely collect from.
• Assessment and culling, 16 to 20 percent.
• Cost model, migration and rollout, 10 to 14 percent. Run it alongside two live matters before it becomes the process of record.

The ongoing costs nobody quotes

• Support retainer, 15 to 20 percent of build cost a year. Source systems change their export interfaces on their own schedule, and a broken collector during an active matter is not a ticket that waits.
• Your own assessment infrastructure, $20,000 to $60,000 a year. Storage and compute for holding metadata and extracted text before promotion. It is cheaper than hosting the same data in a review platform, which is the entire point, but it is not free.
• Review platform hosting. Unchanged, because you are still using it. What changes is how much you promote into it.
• Connector maintenance, $8,000 to $18,000 per source per year. Chat platforms in particular evolve their export capabilities frequently.
• Security review and penetration testing, $12,000 to $30,000 a year. Client litigation data attracts diligence, and outside counsel guidelines increasingly require evidence of it.
• Deletion at matter close. Agree the process during the build. Retaining client litigation data after the obligation ends is a liability rather than a service.

Comparing a build against your current renewal

The comparison most firms attempt is licence against build, and it is the wrong one, because the dominant cost in this category is not a licence at all. It is per gigabyte hosted per month, and it is a variable your process controls rather than your procurement team.

So run the arithmetic on volume instead. Take your hosted volume across live matters, multiply by your rate, annualise it, and then ask what share of that volume would survive a disciplined assessment pass before promotion. Date filtering, custodian scoping, domain analysis, cross custodian deduplication and threading do not remove a marginal percentage. On a corpus assembled by promoting everything under time pressure, the reduction is usually substantial, and it recurs every month for the life of the matter.

Then add what the assessment layer prevents on the preservation side, which is harder to quantify and larger when it lands. A custodian who never acknowledged, a mailbox that followed standard retention because nobody told the systems team, a device wiped by ordinary offboarding. Under Rule 37(e) the consequences of losing electronically stored information that should have been preserved can be serious, and the first question asked is what steps you took.

Against all that, keep paying your review platform. It is a fair trade for what it does, and rebuilding it is the mistake this comparison should steer you away from rather than towards.

When buying beats building

Buy your review platform, without hesitation. Relativity is the most extensible option with the deepest ecosystem. Everlaw and DISCO offer simpler commercial models and strong usability. Nuix is the right answer when difficult processing is your actual bottleneck, and processing is a distinct problem from review. For a firm handling ordinary matter volumes, that stack plus disciplined process beats any build, and a build that tries to displace it will be slower, less capable and permanently behind.

Build the surrounding layer when two or more of these are true. Your legal hold tracking is a spreadsheet and you have had a preservation question raised against you. You cannot tell a client the cost consequence of a scoping decision at the moment the decision is being made. You handle matters where data cannot leave a jurisdiction or must stay on your own infrastructure. Your collections span several systems and your chain of custody evidence is assembled from emails after the fact. Or you are a corporate department managing several outside firms and need one cost and protocol model across all of them, which no single firm's platform will give you because it is not their data to model.

That last case is the strongest and the least served. A corporate department using the same cost model in reverse, to hold panel firms to an agreed protocol and compare cost per gigabyte across providers, tends to change panel decisions quickly. No purchased product will do that for you, because every product in this market sits on one side of that relationship.

If you would rather someone argued with your brief than agreed with it, Digital Heroes starts every engagement with a signed specification covering the data model, permissions and acceptance criteria, which is what keeps a fixed price fixed. You keep the specification either way.