The five questions that reveal whether an eDiscovery quote is designed to be understood — or to be impossible to evaluate until the final invoice arrives.
Almost every eDiscovery quote is built from three pricing primitives: per-GB hosting, per-document review, and a menu of ancillary fees (processing, production, privilege log, project management). Each rate, presented in isolation, looks defensible. Per-document review at $1.50 sounds reasonable. Per-GB hosting at $25 sounds reasonable. Processing at $100/GB sounds plausible.
The problem is that no single rate reveals the total: only when you apply each rate to the actual matter parameters — 250,000 documents, 100 GB, six months, 2,000 privilege documents — does the full cost become visible. And that calculation is exactly what vendors avoid performing in the sales process. The line-item quote keeps each number in a separate mental bucket, so the total never gets computed at the table.
This is not an accident. The purpose of a line-item quote is to make the total cost of the matter impossible to evaluate before you have committed to the platform. Once the data is uploaded and the review is underway, the economics of switching are working against you. The quote format and the switching cost are two sides of the same lock-in strategy.
There is a practical remedy: demand that vendors apply their own rates to your matter parameters before you sign. The five questions below are the mechanism. They come directly from the vendor evaluation worksheet in the DecoverAI pricing white paper, and they work because they force the vendor to perform the calculation they have been carefully avoiding.
A vendor who refuses to provide a single all-in number for your matter is not being cautious — they are telling you something important about their pricing model. All-in pricing requires a vendor to accept the risk that their costs exceed their revenue on your matter; per-line-item pricing transfers all that risk to the client. The refusal to commit to a total is a statement that the vendor's costs are not bounded in advance — which means yours won't be either.
The correct interpretation of “it depends on the matter” is: “our pricing is structured so that we can charge more if we need to.” That is a rational business strategy from the vendor's perspective, but it is a procurement risk from yours. The question is not whether the vendor is being dishonest — it is whether the pricing model gives them the flexibility to charge significantly more than the quoted rates once the matter is underway.
Most do. Processing fees that were quoted at $100/GB apply to documents you didn't expect to collect. Project management hours that were quoted as “included up to 10 hours/month” run over when the matter gets complicated. Production fees that look like a standard per-page rate become significant when the document count is higher than projected. Each of these items is individually defensible; in aggregate, they are the mechanism by which a quote that looked reasonable becomes an invoice that doesn't.
Once you have an all-in vendor estimate, that number is your proportionality anchor under FRCP Rule 26(b)(1). A $460,000 all-in estimate on a $1.2 million commercial dispute is potentially disproportionate as a matter of law — and the post-2015 case law, beginning with Henry v. Morgan's Hotel Group, 2016 WL 303114 (S.D.N.Y. Jan. 25, 2016), holds that the requesting party bears the burden of establishing proportionality. If you can present the court with two all-in quotes for the same matter — one traditional, one AI-augmented — you have given the court something concrete to weigh in a Rule 26(b)(1) analysis.
The proportionality argument is only credible when the cost number is defensible. A vendor quote with line items that could flex upward is not a cost estimate; it is a cost floor. A published, fixed price that covers the full scope of the matter is a cost estimate. Courts weighing burden against benefit under Rule 26(b)(1) need the latter. For the full six-factor analysis and the case law that makes vendor pricing a first-class proportionality argument, see our guide to proportionality in eDiscovery.
Score each vendor on the ten criteria below (1 = poor, 5 = best-in-class). Vendors scoring below 3 on any single row should be re-quoted or replaced. The worksheet is deliberately binary in several rows: either the vendor charges per-seat fees or they don't. Either egress is free or it isn't. These are not matters of degree.
| Criterion | Vendor A | Vendor B | Vendor C |
|---|---|---|---|
| All-in pricing transparency (single number quoted) | ___ | ___ | ___ |
| No per-user / per-seat fees | ___ | ___ | ___ |
| Marginal cost scales linearly with data volume | ___ | ___ | ___ |
| Privilege log generation included in base price | ___ | ___ | ___ |
| Production (Bates, TIFF, load files) included | ___ | ___ | ___ |
| Free data egress / no lock-in | ___ | ___ | ___ |
| AI classification with documented validation | ___ | ___ | ___ |
| SOC 2 Type II + HIPAA compliance | ___ | ___ | ___ |
| Self-serve mode available (no PM hours required) | ___ | ___ | ___ |
| Contract terms: month-to-month, no minimums | ___ | ___ | ___ |
If a vendor cannot tell you the all-in cost for your matter before you sign, that is not uncertainty — that is a pricing model designed to be impossible to evaluate in advance. The evaluation worksheet exists precisely because the vendors who most need to be evaluated are the ones least willing to be evaluated clearly.