Ask a finance team about eDiscovery invoices and you’ll usually hear the same complaint: the quote and the bill don’t match.
The pricing model is usually why.
How per-GB pricing works
A typical per-GB invoice has several meters running at once:
- Processing, charged per GB when data is ingested and made searchable
- Hosting, charged per GB per month for as long as the data sits in the platform
- User fees for each reviewer with access
- Production and export fees when documents go out the door
Each meter looks reasonable on its own. Together they make the total hard to predict.
Why the bill keeps growing
The volume is a guess at the start. The collection you scope isn’t the collection you process. Attachments, email threads, embedded files, and duplicates expand the data set, and you won’t know its real size until processing is done.
Hosting doesn’t stop when review does. Matters stay open for months, sometimes years. A monthly per-GB hosting fee on a large collection keeps adding up long after the review work is finished.
Scope moves. New custodians, wider date ranges, and follow-up requests from opposing counsel all add data, and every new gigabyte starts its own meter.
For a CFO, that’s the hardest kind of cost to defend: necessary, recurring, and impossible to forecast.
How per-matter pricing works
Per-matter pricing sets the price based on how complex a matter is, not how many gigabytes move through the system.
Discovery AI prices matters in a small number of complexity tiers, based on things like collection size, number of custodians, privilege complexity, and timeline. You agree on the tier before work starts. If a matter grows past it, we talk about it before anything changes.
The AI itself runs in your own Azure environment. Microsoft bills that usage at its list price, and we don’t mark it up.
Side by side
| Per-GB | Per-matter | |
|---|---|---|
| Priced on | Data volume, known only after processing | Matter complexity, agreed before work starts |
| Hosting over time | Monthly per-GB fees for the life of the matter | No per-GB hosting meter |
| Forecasting | Hard | Straightforward |
| Fits a budget line | Poorly | Cleanly |
| AI costs | Bundled into the vendor’s price | Billed by Microsoft through your Azure at list price |
Questions to ask any vendor before you sign
- What expansion do you expect from collected to processed data, and who pays for it?
- What does hosting cost per month, and does it stop when review ends?
- Are there separate fees for users, productions, or exports?
- If AI features are included, how are they billed, and is there a markup?
- What happens to the price if the matter grows?
Get the answers in writing. They’re the difference between a quote and a forecast.
Why this matters for an in-house team
A legal department is accountable for a budget. Per-matter pricing gives finance a number before the work starts, and that’s what makes bringing review in-house easy to approve.
If you’d like to compare a recent per-GB invoice against a per-matter model, bring it to a call and we’ll walk through it together.