Our method, in the open
A grade is only useful if you can see how it was made. Here's exactly where our data comes from, how we turn it into grades, and what it can't tell you.
1. Where the data comes from
Every Report Card is built from North Carolina public court records:
- Case records from the statewide eCourts system — the register of actions (docket), parties and attorneys of record, hearings, judgments and dispositions.
- Filed documents in each case — complaints, motions, orders, consent orders and judgments — read to understand what was asked for and what the court decided.
- For Battle Research, recordings of prior public hearings, requested through official channels where they are lawfully available.
We never use sealed, juvenile, or confidential records, and we never contact the attorney, their firm, the parties, or the court about your order.
2. Finding the attorney's cases
We search for every case where the attorney appears as counsel of record in your lookback window (the last 12 or 24 months), normalize each case into the same structure, and note which side the attorney represented, the judge, the case type, and how and when it ended. If a case can only be seen at the docket level (no documents yet), the report says so.
3. Reviewing each case
Each case gets a structured review: a plain-English summary, how contested it was (none, low, moderate or high), how it was resolved (trial, hearing, consent order, mediated agreement, default, dismissal and so on), the outcome for each side, each issue that was decided and who it favored, money awarded, notable events (continuances, withdrawals, sanctions, appeals, emergency relief), and observations about each attorney's work tied to specific filings. Minor children are never named.
4. The four grades
| Grade | What it measures | Rule |
|---|---|---|
| Experience & volume | Cases on record and years of activity | A: 75+ cases over 5+ filing years · B: 40+ over 3+ · C: 15+ · D: 5+ · fewer: not graded |
| Contested results | Outcomes in cases that were actually fought | Smoothed score (wins + ½ ties + 2) ÷ (contested cases + 4): A ≥ 70% · B ≥ 60% · C ≥ 50% · D ≥ 40% · F below. Needs 3+ contested cases with a known result. |
| Speed to resolution | How long cases take vs. a reference time for that matter type | Median of (days to disposition ÷ reference days): A ≤ 75% · B ≤ 100% · C ≤ 125% · D ≤ 150% · F above. Needs 3+ closed cases. |
| Procedural reliability | Withdrawals, sanctions/contempt against the client's side, documented procedural problems | Concerns per reviewed case: A 0 · B ≤ 0.10 · C ≤ 0.20 · D ≤ 0.35 · F above. Needs 3+ reviewed cases. |
| Overall | Average of the graded categories | 4.0 scale: A ≥ 3.5 · A− ≥ 3.3 · B+ ≥ 3.0 · B ≥ 2.7 · B− ≥ 2.5 · C+ ≥ 2.2 · C ≥ 1.8 · D ≥ 1.0 · F below |
Why the smoothing? An attorney who went 2–0 in two contested cases hasn't proven a 100% record. Adding two "virtual" wins and two "virtual" losses pulls small samples toward the middle so a handful of cases can't produce an extreme grade. Uncontested matters — where nobody opposed anything — are never counted as wins.
Blank grades are a feature. If there aren't enough cases to grade a category fairly, we leave it blank and say why, rather than guess.
5. Patterns and the two lenses
After grading, we look across all the case reviews for patterns: strengths, concerns and recurring tactics (for example, "files for temporary custody with the complaint" or "requests attorney fees in most contested cases"). Every pattern lists the cases that show it, and a pattern needs more than one case — if something happened once, the report says "in one case."
- The hiring lens covers who the attorney fits well and poorly, what a client should expect, and questions to ask before hiring.
- The opposing lens covers what the other side should expect and practical preparation steps grounded in the record. It is general information, not legal advice.
6. Battle Research: from audio to patterns
For Battle Research we choose the attorney's prior public hearings that are most like yours (same hearing type and, where possible, the same judge), request the recordings through official channels, and convert them to text with speech-to-text software that labels each speaker. Transcripts are checked against the audio for the passages we quote. We then analyze them for patterns: openings, themes and repeated phrases, cross-examination style, objections, how they deal with self-represented parties, the relief they request, and how they react to pushback from the bench. Not every hearing is recorded, and availability differs by county and courtroom.
7. What court records can't tell you
- Records show outcomes, not how good a client's facts were. A lawyer who takes hard cases may show more losses.
- Most cases settle, and settlements can be good or bad for a client in ways the record doesn't show.
- Records can be incomplete or contain clerical errors, and some documents are not yet available online.
- Records say nothing about bedside manner, responsiveness, or fees.
That's why every report ends with a confidence note, and why we suggest using a Report Card alongside a conversation with the attorney — not instead of one.
8. Our limits
Firmbinder LLC is not a law firm and does not give legal advice, recommend attorneys, or predict the outcome of any case. Grades summarize what the public record shows under the method above; they are not a finding about any attorney's competence or ethics.