Methodology · Published in full

The Dareen Scoping Method

A five-step framework for estimating the partner, associate and paralegal hours a legal matter will take — in which every hour traces back to an assumption printed next to it, so the partner can defend the number to a client line by line, and a reviewer can argue with the assumption instead of the total.

Developed by Dareen.ai and used across 27 transaction and dispute archetypes in four practice groups. This page is the complete method. It is ungated, and free to quote, cite, teach or implement with attribution.

Why most legal fee estimates fail

The common failure is not that estimates are wrong. It is that they are unauditable. A partner produces a number from experience; the client asks why; the answer is a narrative rather than a structure. When the matter runs over, nobody can say which part of the original thinking broke — so nothing is learned, and the next estimate is produced the same way.

The second failure is the point estimate. Real matters have a smooth path and a stressed path, and the distance between them is large — in Dareen's own worked banking example below, roughly 35%. Quoting the midpoint as a single figure hides the variable the client most needs to see and most influences: how the matter is run.

The method, in five steps

01Match the matter to an archetype

Start from a named baseline for this kind of matter — "mid-market bilateral loan", "buyer-side lower-middle-market acquisition", "post-discovery settlement" — scoped as a smooth case: the version where nothing goes wrong. The archetype is a real, phase-by-phase build, not a category label. Choosing the archetype is the single highest-leverage decision in the estimate; everything after it is adjustment.

02Derive hours from stated effort rates

Hours come from an explicit engine of effort assumptions, not from recall. For example, in Dareen's banking & finance methodology: a senior associate reads a complex legal document at 15–20 pages per hour and a partner at 25–30; drafting from precedent runs at 2–3 pages per hour, drafting from scratch at roughly 1; a paralegal registry filing takes about half an hour; closing-binder assembly runs about 0.1 hours per document. Every firm should calibrate these against its own realisation data — but they must be written down, because they are what a disagreement is actually about.

03Lay the matter out as phases × roles

Build the estimate as a matrix: each phase of the matter down one axis, each fee-earner tier across the other, with the governing assumption printed on every line. Two rules keep the matrix honest:

• Supervision is a line item, never a multiplier. Partner review, associate sign-off on paralegal work and partner oversight of paralegal-heavy phases appear as named rows in the phase they occur in. Applying a blanket supervision uplift on top of the total double-counts work that is already in the matrix.
• Self-review overhead is embedded, not re-applied. An associate's revision time (roughly +30–40% on drafting) belongs inside the drafting line, and is not added again later.

04Stress the baseline with scenario flags

Adjustments are made through a fixed vocabulary of qualitative factors, each with three states — smooth / standard / stressed — that map to a percentage uplift on one named phase. Data-room organisation, counterparty tone, number of negotiation rounds, signatory count, jurisdictions, security registrations: each is a question the partner can answer at intake without doing any arithmetic.

Three arithmetic rules govern how flags combine:

• Add, don't multiply. Two flags of +25% and +20% on the same phase give +45%, not +50%. Compounding overstates friction that is largely overlapping.
• A soft cap of +100% per phase. Past that point, the honest conclusion is that you have the wrong archetype — not a very stressed version of the right one. The cap is a signal to re-scope, not a ceiling to grind against.
• Where a flag and a matter-type modifier describe the same driver, take the larger of the two, not both. Cross-border and foreign filings, distressed and adversarial tone: these overlap, and stacking them is the most common way a scoping model inflates.

Uplifts scale all roles in the affected phase by default, not just associates. A messy data room adds paralegal time on registry re-pulls as well as associate re-review. The partner can override the role mix where a flag is genuinely one-sided.

05Deliver a range, and show the assumptions

The output is the smooth case and the stressed case together, with the assumption sheet attached — not a single number with the reasoning discarded. The spread is the conversation: it tells the client which behaviours (a clean data room, decisive instructions, a realistic signatory list) move the fee, and it gives the firm a defined trigger for a scope conversation when a flag flips mid-matter.

Worked example

A $20–50M secured, single-jurisdiction bilateral loan, taken through seven phases from term sheet to post-closing. The stressed case is the same deal with a document-dump data room, seven negotiation rounds, 18 signatories, two-jurisdiction execution and five security registrations.

Two things are worth reading off that table. First, the ~35% spread is the number to give a client, in place of a false-precision midpoint. Second, stress does not simply push work upward to partners — paralegal hours rise at almost the same rate, because friction generates administrative load (re-pulled searches, more signature wrangling, a larger closing book) as much as it generates judgment calls.

Where the method has been applied

The same five steps have been built out across 27 archetypes in four practice groups. Each archetype is a full phase-by-phase baseline with its own flag set.

How to implement it without software

Nothing here requires a product. A firm can run the method on a spreadsheet:

• Pick your three most common matter types and build a smooth-case phase × role baseline for each, calibrating effort rates against closed matters.
• Write down 4–6 scenario flags per phase, in three states, with a percentage on each state. Argue about the percentages once, then fix them.
• Estimate by selecting flags, adding within each phase, and reporting smooth-to-stressed as a range with the assumption sheet attached.
• After the matter closes, compare actuals to the baseline and correct the effort rates — not the individual estimate. The engine is what improves.

What software adds is the last step at scale: matching a new matter to the nearest archetype from the firm's own closed-matter history, and keeping the effort rates calibrated automatically rather than annually. That is what Dareen.ai does.