eDiscovery platforms vs a conversion tool
A review platform and a conversion tool are not competitors in any meaningful sense. One is a system for reviewing, culling and producing large document sets across a team. The other renders email into court-ready files.
The reason this page exists is that small firms genuinely do face the choice — not “which of these two”, but “do I need one of those at all for this matter”. That question has a real answer, and for a lot of matters the answer is yes.
What a platform does that this does not
This is the substantial list, and it is not a formality:
Deduplication. The same message exists in the sender’s mailbox and every recipient’s. Across four custodians you are looking at substantially the same mail four times. Deduplication is the single largest cost saving in a multi-custodian collection and there is no manual substitute at scale.
Email threading. Grouping a conversation so a reviewer reads the thread once rather than reading the first message twelve times as it is quoted down the chain.
Search and tagging across the corpus. Real search across millions of documents, with saved searches, tag propagation and review batches.
Multi-reviewer workflow. Assignment, progress tracking, quality control, and consistent privilege calls across a team — which is what stops two reviewers making opposite calls on the same document.
Analytics and technology-assisted review. Concept clustering, predictive coding, prioritised review. On a large matter this is the difference between proportionate and impossible.
Load-file production. Images, .dat, .opt, extracted text, parent-child
fields, all built to a specification and validated before delivery.
Defensible audit trail. A record of who reviewed what and when, which matters when the process itself is challenged.
A conversion tool does none of that. It renders .eml files into PDF or TIFF
with the header block intact and the attachments preserved.
Where a platform is genuine overkill
| Small-matter workflow | Review platform | |
|---|---|---|
| Typical scale | Hundreds of documents | Tens of thousands upward |
| Custodians | One or two | Several to many |
| Reviewers | One | A team |
| Deduplication / threading | Manual, or not needed | Automatic |
| Load-file production | Not usually required | Built in |
| Cost model | Per seat, or per conversion | Hosting plus per-GB, ongoing |
| Setup time | Minutes | Days, plus a vendor conversation |
| Ongoing cost after production | None | Hosting continues while data sits there |
That last row is the one small firms underestimate. Platform costs are typically ongoing for as long as the data is hosted, which for a matter that runs for two years is a different proposition from a one-off conversion.
The honest decision rule
Use a platform when there are multiple custodians, more than one reviewer, a load-file production requirement, a real prospect that collection method or review adequacy gets litigated, or a volume where deduplication and threading change the cost by an order of magnitude.
Do it in-house when it is one or two cooperative custodians, a few hundred responsive messages after culling, a single person reviewing, and a production that is going out as Bates-stamped PDFs to opposing counsel who will read them.
The failure mode to avoid is running a large production with small-firm tooling and discovering it at the numbering stage — because by then the conversion, the review and the formatting have all been done in a way that has to be redone. If you are genuinely unsure which side of the line a matter falls on, that uncertainty is itself a signal to get a vendor quote before starting rather than after.
Where this tool sits
At the conversion step of the small-matter workflow, and nowhere else. Export
the messages as .eml, convert the responsive set in one pass so pagination is
consistent, stamp in your numbering tool, produce.
How to produce emails in discovery
walks the whole sequence.
If a matter needs deduplication, threading, team review or a load file, it needs a platform, and no amount of conversion tooling substitutes for one. That is not a limitation being conceded reluctantly — it is a different category of software.
What is the actual threshold?
Less about document count than about three things: how many custodians, how many people are reviewing, and whether the collection method is likely to be challenged. One custodian, one reviewer, nobody contesting how the set was assembled — you do not need a platform. Reverse any of those and the calculation changes quickly.
Can I produce a load file without a platform?
With difficulty, and it is usually a false economy. A conventional
Concordance-style delivery needs image files, a .dat with correctly
ordered and delimited fields, an .opt mapping images to documents, and
per-document extracted text. Hand-assembling that is possible for a small
set and unforgiving — platforms reject malformed load files with
unhelpful errors. See load file.
Are platforms overkill for a two-party contract dispute?
Frequently, yes. A few hundred responsive emails from two cooperative custodians, produced as Bates-stamped PDFs, is a legitimate and common production. Paying platform hosting for that is a cost with no corresponding benefit.
Where does EML Exhibit fit if we are using a platform?
Usually nowhere — the platform handles ingestion and production itself. The exception is the last mile: rendering a specific set of messages into clean exhibits for a motion or a deposition, where a platform’s export is more cumbersome than converting the natives directly.