Jury Selection Software Trial Run: A 30-Minute Test Before You Buy

If you want to know whether jury selection software will hold up on a Tuesday morning with a judge waiting, don't read the feature list. Build a fake panel, run it through the product for 30 minutes, and try to break it. A product that survives a messy mock panel, an accidental edit, and a last-minute export request will probably survive your actual trial. A product that can't is telling you something a sales demo never will.

This isn't a review of any particular platform or a pricing breakdown. It's a test you run yourself, on your own laptop, before a dollar changes hands.

Why a demo isn't enough

Sales demos are built to work. The vendor's rep knows the data, knows the workflow, and has rehearsed the click path a dozen times. None of that tells you what happens when a paralegal uploads a CSV with three blank fields and a juror whose name has an apostrophe in it, or what happens when you accidentally delete a flag during a sidebar and need it back in ten seconds.

Jury selection software gets used under conditions that are nothing like a demo: bad courtroom wifi, a clerk handing you a corrected panel list at 9:05 a.m., opposing counsel watching you fumble with a laptop. The only way to find out how a product behaves under those conditions is to put it under similar pressure yourself, with low stakes, before trial.

Building the mock panel

Start by creating a synthetic panel of 25 to 40 fictional jurors in a spreadsheet. Give each one a juror number, an age range, occupation, residence, prior jury service history, litigation experience, relationships to parties, hardship status, and a notes field pulled from a fake questionnaire.

Then deliberately make it messy. Leave some fields blank. Write a few long, rambling notes the way a real juror might answer an open-ended question. Duplicate a name. Mix capitalization. Drop in at least one special character, an apostrophe, an ampersand, an accented letter. Real juror data is never clean, and a product that only works on a tidy sample spreadsheet isn't being honest with you about how it will perform on day one of voir dire.

One rule matters here: never upload real juror questionnaires to test a vendor's product. Federal district rules commonly treat juror questionnaires as confidential, limit their use to jury selection, and require counsel to protect them at least as carefully as confidential client information.[1][2] Some courts require that copies be returned or destroyed once selection is complete.[1][2] ABA Resolution 602 reinforces that questionnaires not made part of the record should be disposed of after trial.[7] A vendor's free trial account is not a place to test that obligation. Use fiction.

Minute 0 to 5: import

Upload the CSV and time yourself. Can you map the fields without calling the vendor's support line? Do the dates, notes, blanks, and special characters survive the import intact, or does something get truncated or silently dropped? Does the software flag the duplicate name, or does it overwrite one record with the other without telling you? Is there any audit trail showing what was imported and when?

Write down the total time from upload to a usable, reviewable panel. If it takes longer than five minutes with a clean mock file, it will take longer still with a real panel list handed to you minutes before jury selection starts.

Minute 5 to 10: build the seating chart

Arrange your fictional jurors into the platform's courtroom layout. Then shrink the browser window and try navigating with the keyboard alone, no mouse. Courtrooms are not always kind to a laptop screen angled toward counsel table, and you may not always have a free hand for a trackpad.

Check whether juror numbers stay visible at a glance, whether you can read a key note without opening a second or third window, and whether the chart prints or displays on a shared screen without text getting clipped at the margin. A fair pass-fail line: you should be able to locate any juror, read the note that matters, and move that juror within three actions. If it takes five clicks and two scrolls to find juror number 14's hardship note, that's a problem you'll feel acutely with a courtroom full of people waiting on you.

Testing jury selection software on flags, recovery, and exports

This is the part of the trial run that separates a usable product from an impressive-looking one. Feature lists tell you a tool can flag jurors. The next three steps tell you whether it does so in a way you can trust under pressure.

Minute 10 to 17: flag issues

Create at least five flags across your mock panel: hardship, cause concern, prior litigation experience, a potential relationship to a party, and a follow-up question that needs an answer before you can decide. Then test whether the flags are readable for colorblind reviewers, whether you can sort and filter by flag type, whether you can create a custom label, and whether the flags actually show up when you export.

If the product ranks or recommends jurors based on some internal scoring, stop and ask the vendor to explain the methodology and the underlying data. ABA Formal Opinion 517 makes clear that the lawyer, not the software, bears responsibility for understanding and supervising generative AI tools used in legal work.[10] A ranking you can't explain to a judge if asked is a ranking you shouldn't rely on. ABA guidance separately permits reviewing publicly available juror internet information during jury selection, subject to the court's own limits and a firm prohibition on any direct or indirect communication with jurors.[5] A flagging tool should help you organize what you've lawfully gathered, not infer protected characteristics or nudge you toward strikes you couldn't justify on the record.

Minute 17 to 22: recover from an accidental edit

Now break something on purpose. Change a juror's note, delete a flag, move a juror to a different seat, then close the matter or close the browser entirely. Try to undo each change. Look for version history, autosave, a backup copy, and a record of who made the change and when.

This is the step that matters most and gets skipped most often. A tool that can't clearly show who changed what and restore the prior state has failed a basic courtroom-reliability test, no matter how polished its interface looks. Jury selection moves fast, and someone on your team will eventually fat-finger a note or drag a juror to the wrong seat. The question isn't whether that happens. It's whether you can fix it in ten seconds or lose five minutes figuring out what went wrong while the judge is asking why you're not ready to proceed.

Minute 22 to 27: export a strike list

Export both a PDF and a spreadsheet containing juror number, seating position, flags, notes, and disposition. Check the pagination, the legibility at a normal print size, whether timestamps and metadata are attached, whether any redactions actually redact, and, critically, whether deleted information still lingers embedded somewhere in the file. Confirm the export reflects your current, edited panel and not an earlier snapshot from before you made your changes in minute 17.

An export that looks clean on screen but carries hidden metadata from a deleted note is a confidentiality problem waiting to surface, particularly if that file ever gets shared or produced.

Minute 27 to 30: score it

Rate each category on a 1 to 5 scale: import accuracy, seating-chart usability, flagging and search, undo and auditability, export quality, speed, and security and access controls. Then apply one hard rule: reject any product that scores below a 4 on recovery, export, or confidentiality, even if the aggregate score looks good. A high average can hide a single catastrophic weak point, and recovery, export, and confidentiality are the three places where a weak point costs you the most at the worst possible moment.

Close out the trial run with a written record: the defects you found, the vendor's answers to your questions about them, and the exact data retention and deletion terms the vendor commits to in writing. That document is worth more than any sales brochure when you're deciding between two products that otherwise look similar.

The plaintiff's perspective

Plaintiff's counsel often works with smaller teams and tighter budgets than the defense side of the same case. A tool that requires a paralegal's full attention just to keep the seating chart current isn't saving time, it's shifting work around. Run the mock panel test with the same staff you'd actually bring to trial, not your most tech-savvy associate, to see whether the product holds up with the team you'll really have in the room.

The criminal defense perspective

In criminal matters, the stakes around confidentiality and chain of custody for notes are especially sharp, since juror information can touch on sensitive personal history that has nothing to do with the case. The recovery test in minutes 17 to 22 matters most here: if a note about a juror's answer during voir dire gets altered or lost, and you can't show what it originally said, you've lost a record you may need later.

The corporate defense perspective

Corporate defense teams frequently run larger trial teams across multiple matters, which raises the stakes on permissions and access controls. During the trial run, create two user accounts with different permission levels and confirm that a lower-level user genuinely can't edit or delete what they shouldn't be able to touch. A single shared login that gives everyone full edit rights is a liability, not a convenience.

The prosecution's perspective

Prosecutors often select juries under institutional record-retention rules distinct from a private firm's own policies. Before adopting any tool, confirm in writing how the vendor handles data deletion once a case closes, and whether that process satisfies the office's retention and disposal obligations, not just the vendor's default settings.

FAQ

Is it safe to use real juror questionnaires to test a new software vendor? No. Federal district rules commonly treat juror questionnaires as confidential and limit their use to jury selection, with some courts requiring copies to be returned or destroyed afterward.[1][2] Use a synthetic panel instead, and confirm the vendor's security terms and the court's own rules before ever uploading real juror data to any third-party platform.

How long should a software acceptance test actually take? Thirty minutes is enough to cover the core workflow: a five-minute import, a five-minute seating chart build, seven minutes of flagging, five minutes testing recovery from an accidental edit, five minutes exporting a strike list, and three minutes to score the results. You can extend any step if the product surfaces a problem worth digging into further.

What should disqualify a product even if it scores well overall? A score below 4 out of 5 on recovery, export quality, or security and access controls should disqualify a product regardless of its aggregate score. These are the categories where a weakness surfaces at the worst possible time, mid-selection, with a judge and jury pool waiting.

Can jury selection software recommend which jurors to strike? Some products offer ranking or recommendation features. If a product does this, ask the vendor to explain the methodology and underlying data before relying on it. ABA Formal Opinion 517 places responsibility on the lawyer, not the software, for understanding and supervising AI-driven tools used in legal work.[10] Avoid any tool that appears to infer protected characteristics or that you couldn't explain if a judge asked about your strike rationale.

Is it permitted to look up jurors online during jury selection? ABA guidance permits reviewing publicly available juror internet information during jury selection, subject to the specific court's own limits, and with a firm prohibition on any direct or indirect communication with jurors.[5] Check your jurisdiction's local rules before relying on this during an actual trial.

A practical note

None of this replaces a lawyer's judgment about which jurors to strike or how to read a panel in the room. Software can organize what you already know and help your team move faster, but the call still belongs to counsel. At StrikeList AI we built our own tools with this kind of hands-on test in mind, precisely because a feature list never tells you how a product behaves when someone on your team makes a mistake at 9:15 on the morning of jury selection. Whatever platform you choose, run your own 30-minute test first.

Notes

1. United States District Court, Northern District of Illinois, Jury Questionnaire Confidentiality Policy, https://www.ilnd.uscourts.gov/
2. United States District Court, District of Massachusetts, Local Rules on Juror Questionnaires, https://www.mad.uscourts.gov/
3. American Bar Association, Formal Opinion on Lawyers' Use of Publicly Available Juror Information, https://www.americanbar.org/
4. American Bar Association, Resolution 602, Juror Questionnaire Disposal Guidance, https://www.americanbar.org/
5. American Bar Association, Formal Opinion 517, Generative Artificial Intelligence Tools, https://www.americanbar.org/