“We’ll Win at Trial” — Reality-Testing in Caucus

The private room, hour two

The joint session is over. The mediator is in a private session with the plaintiff and her lawyer. The plaintiff is a former employee suing for wrongful termination. Her last demand was five times what the defendant has signaled it might pay.

"Honestly, I'm not worried. Our case is strong. We'll win at trial."

Her lawyer says nothing, which the mediator notices.

The mediator has spent the morning with both sides. She has a sense of the defendant's evidence, including some the plaintiff may not have fully considered. She can't share what the defendant told her in confidence. She can't tell the plaintiff she'll lose. And if she pushes too hard, the plaintiff will conclude that the mediator is working for the other side.

Testing reality without delivering a verdict

Reality testing in mediation is the process of helping a party evaluate the strength of their position, the costs and risks of continuing, and the realistic alternatives to settlement. It's one of the most important — and most delicate — tasks a mediator performs.

Four features make it difficult.

The party is invested in her story. Litigation often becomes part of a person's sense of what happened to them. Questioning the case can feel like questioning the experience.

The mediator has confidential information. Private sessions are typically confidential unless the party agrees otherwise. The mediator can't use what the other side said to argue with this one.

Mediators differ in style. Some mediators are more facilitative, helping parties reach their own assessments; others are more evaluative, offering views on the likely outcome where the parties want that. Which approach is appropriate depends on the mediator, the parties and what they've agreed.

And neutrality is at risk. A mediator who sounds like opposing counsel loses the trust that makes the private session useful.

Ask, don't tell

The core technique is questions that let the party assess her own risks.

Start with her view. "Walk me through how you see trial going." Listen fully before testing anything.

Ask about specific risks. "How do you think the jury will see the performance reviews from 2023?" "What happens if the judge excludes the email from your manager?" Specific questions invite specific thought.

Ask about the alternative, not just the case. "If this doesn't settle today, what does the next year look like?" Time, cost, stress, discovery, the possibility of an appeal. Many parties focus on the verdict and not the path to it.

Ask for a range, not a prediction. "What's the best outcome you can imagine at trial? The worst? What's realistic?" Parties who state a range often reassess their demand on their own.

Involve the lawyer. "Counsel, how do you see the risks?" The lawyer may have told the client things privately that she hasn't absorbed — or may welcome an opening to say them again.

Summarize neutrally. "So if I've got it right, you think liability is strong, but there's some uncertainty about how the reviews come in and about damages, and trial is likely a year or more away." A neutral summary often lands harder than any argument.

Timing matters

Listen first, test later. Many mediators use the first private session mainly to understand and acknowledge the party's position, and save the harder questions for a later session. A party who feels heard is much more willing to consider risks than one who feels challenged in the first ten minutes.

Let numbers follow reassessment. Asking for a new demand immediately after raising risks can feel like the point of the questions was to extract a concession. Give the party — and her lawyer — time to talk privately before the next move.

What not to do

Don't use the other side's confidences. Unless the other party has authorized sharing, what they said in private stays private.

Don't argue the other side's case. Questions are neutral; advocacy isn't.

Don't give a verdict unless evaluation has been agreed. If the parties want the mediator's view, give it carefully and as a view. If they haven't asked, don't impose one.

And don't rush. A party who feels hurried into reassessing her case will resist. Reality testing often takes several private sessions.

Four ways it goes wrong

The advocate for the other side, whose questions sound like cross-examination.

The confidence-breaker, who uses what the other party said privately to pressure this one.

The silent facilitator, who never tests the party's assumptions at all and lets unrealistic positions harden.

The rusher, who expects a party to abandon a position in one conversation.

Why this isn't trained

It's taught conceptually. Mediator training explains reality testing; practicing it with an invested, resistant party is harder to arrange.

Style differences create confusion. New mediators are often unsure how evaluative to be, and default either to silence or to verdicts.

Confidentiality is easy to breach by implication. A question can suggest what the other side said even without stating it. That subtlety needs practice.

And practice parties are too flexible. Role players often reconsider their positions quickly. Real parties who have spent years in litigation don't.

What reality-testing simulation can rehearse

A ten-minute simulation can put the mediator in a private session with an overconfident party and a quiet lawyer — so she practices asking specific risk questions, exploring the alternative and summarizing neutrally without breaching confidentiality or sounding like the other side. The AI agents in Foretell AI play the party and counsel consistently; the case, confidentiality terms and mediation style stay with the mediator or program.

Four versions to build:

  • The certain winner, who is confident about trial and hasn’t considered the path to it.
  • The invested party, for whom the case is about vindication more than money.
  • The quiet lawyer, who agrees with the mediator’s concerns but hasn’t said so to the client.
  • The party who asks what the mediator thinks, testing whether the mediator gives a view only as agreed and frames it carefully.

Design caution. Mediation confidentiality, the treatment of caucus communications, mediator standards on evaluation and court-program rules vary by jurisdiction and program. Modules must use the program's own rules and agreed mediation approach. Nothing here is legal advice.

Designing the module

Ten minutes, scored against a reality-testing rubric.

Pass one — understanding. Did the mediator elicit the party's view fully before testing it?

Pass two — questioning. Were questions specific, neutral and focused on risks and alternatives rather than on conclusions?

Pass three — boundaries. Was confidentiality protected? Was neutrality maintained? Was any evaluation given only as agreed?

Rubric on observable behavior: Was the party's view elicited first? Number of specific risk questions. Was the alternative to settlement explored? Was a range requested? Was the lawyer invited in? Was a neutral summary offered? Were any confidences used? Did any question sound like advocacy?

Specific risk questions is the measure. They are the difference between a party who reassesses and one who simply hears that the mediator disagrees.

For mediators, ADR providers and law firms

It moves negotiations. Parties who reassess risk on their own are more likely to settle — and more likely to be satisfied with the settlement.

It protects neutrality. Question-based testing preserves the trust that makes private sessions work.

It supports advocates. Lawyers who know how mediators reality-test can prepare clients for the conversation — and use it.

And it's hard to learn by observation. Private sessions aren't observable. Simulation lets mediators practice the most confidential part of their work.

For law schools, ADR courses can use it to show students the mediator's side of a negotiation they'll later participate in as advocates.

Frequently asked questions

What is reality testing in mediation? The process by which a mediator helps a party evaluate the strengths and weaknesses of their position, the risks and costs of continuing, and their realistic alternatives to settlement.

How do mediators reality-test without taking sides? By asking specific, neutral questions about risks and alternatives, inviting the party and their lawyer to assess them, and summarizing neutrally — rather than arguing the other side's case.

Can a mediator share what the other side said in caucus? Generally only with that party's permission. Caucus communications are typically confidential, subject to the jurisdiction's rules and the mediation agreement.

Should a mediator tell a party they will lose? Only if the parties have agreed the mediator may offer evaluations — and even then, as a view rather than a prediction. Many mediators rely primarily on questions.

The short version

"We'll win at trial." Her lawyer said nothing. The mediator knows things she can't repeat.

Ask her how trial goes. Ask about the specific pieces of evidence that could go the other way. Ask what the next year looks like. Ask for her best, worst and realistic outcomes. Bring her lawyer in. Then summarize what she said — neutrally — and let it land.

Reality testing works best when the party reaches the conclusion. The mediator's job is to ask the questions that get her there.

Foretell AI lets mediators and ADR programs build caucus simulations — including overconfident parties, confidentiality boundaries and question-based reality testing like the one above — with configurable AI parties and counsel, recordings and rubric-based evaluation. If your mediators learn caucus technique only in live sessions, we're happy to walk through how other programs have structured it.