The phrase “jury consultant” suggests someone who predicts verdicts. That is the least of it, and firms that hire on that expectation are usually disappointed for reasons that have nothing to do with the consultant’s ability.
The work is mostly research design, question design and interpretation — and understanding which of those a piece of software can help with is the difference between spending money well and buying a number.
Designing the research. Deciding what question the firm needs answered, and constructing a mock trial, focus group or survey that could actually answer it. Most badly-run pretrial research fails here, before anyone is recruited, by testing whether people like the case rather than what changes their mind.
Writing the voir dire. Questions that surface attitudes rather than eliciting the answer the juror thinks is expected. This is a craft. “Can you be fair?” produces nothing. Questions about experience — what has happened to this person, what they have had to deal with — produce something usable.
Interpreting what came back. A focus group produces a great deal of talk. Deciding which reactions are signal, which are the loudest participant, and which are artefacts of the way the case was presented is judgement built on having watched many of them.
Telling the trial team something they do not want to hear. Arguably the most valuable function. A team six months into a case cannot see which parts of its theory are unpersuasive. An outsider can.
Scale in the cheap phase. Testing four framings of an opening across a large simulated sample, before spending money on human research, narrows what you take to real people. Use software to decide what to test, not to conclude.
Structuring what you already have. Organising questionnaire responses, coding open-text answers, and surfacing patterns across a panel is mechanical work done faster by machine.
Consistency in note-taking during voir dire. A structured record of what each panel member said, searchable afterwards, is more reliable than what anyone remembers by the afternoon.
Rehearsal. Testing sequence and comprehension repeatedly, at no marginal cost. Covered separately in trial preparation is a rehearsal problem.
Reading a room. Hesitation, the glance at another panel member, the answer that arrives half a second late — none of that is in the transcript, and it is a large part of what an experienced observer is doing.
Deciding what the research question is. Software answers questions; it does not tell you which one matters in this case.
Being wrong in a useful direction. A consultant who tells you the case theory is confusing has done something no model reliably does, because the model has been given the case by the people who already understand it.
Anything touching strike decisions. This is a professional-conduct question and not merely a tactical one. If a tool’s ranking could shape a peremptory challenge, the lawyer must understand its methodology well enough to know whether following it would produce an unlawful strike — see what ABA Formal Opinion 517 requires.
Most firms cannot afford a full consultant engagement on most cases. A sensible ladder:
The mistake is skipping to step five on a case that needed step one, or believing step three replaces step four. Simulation tells you what to ask real people. It does not tell you what real people will do.
Technology has made steps one to three enormously cheaper, which is a real democratisation — a solo practitioner can now do preparation that used to require a budget. It has not replaced the consultant’s actual contribution, which is judgement about research design and the willingness to deliver an unwelcome conclusion.
A firm that uses software to arrive at trial having tested its case ten times is better prepared than one that arrives with a prediction. The prediction is the part to distrust.
VerdictPilot is built for steps one to three: testing themes and sequence against modelled juror response, surfacing the objections the trial team has stopped seeing, and keeping the record of what was tested. It is preparation infrastructure, not a substitute for human research or for judgement about which strikes to make.