When an employee handbook becomes a contract nobody meant to sign

The contradiction built into every handbook

A handbook is asked to do two incompatible things. It has to be specific enough to be useful — here is how leave is requested, here is what happens if you are late repeatedly, here is the complaints procedure — and vague enough that none of it can be held against the company later. Specificity is what makes it worth writing. Specificity is also what makes it look like a set of promises.

The resolution most employers reach is a disclaimer. That is the right instinct, and it works more often than not. But the disclaimer is doing more legal work than the people who paste it in usually realise, and it has a second effect that is almost always unintended.

How a handbook becomes an implied contract

The leading case is Woolley v. Hoffmann-La Roche, decided by the New Jersey Supreme Court in 1985 and followed in substance well beyond New Jersey. The court held that where an employment manual sets out comprehensive terms and conditions — particularly on termination and discipline — and is widely distributed, an implied promise that an employee will only be dismissed for cause can be enforceable, even though the employment is for an indefinite term and would otherwise be at will.

Two features do the damage. The first is comprehensiveness: a policy detailed enough that an employee could reasonably rely on it. The second is that it deals with job security. A handbook that explains the expenses process in exhaustive detail creates no exposure. A handbook that sets out a four-step disciplinary sequence has just described the circumstances in which someone will not be dismissed.

What makes a disclaimer work

Courts following Woolley have developed a two-part test, and both parts fail routinely. A disclaimer must be visually prominent — set off by capitalisation, boxing, bold or position so that it is likely to come to the attention of someone reading the handbook — and it must be written in language a reasonable employee can actually understand rather than confusing legalese. New Jersey courts have treated the effectiveness of a given disclaimer as a jury question rather than something settled on the page, which tells you how uncertain the drafting exercise is.

The last item is the one most often skipped and the cheapest to fix. A signed acknowledgment that restates the disclaimer is far better evidence than a disclaimer on page three of a PDF nobody opened.

The trap: your disclaimer can destroy your arbitration clause

This is the part that catches well-advised employers. If your handbook contains an arbitration provision, the disclaimer you added to defeat implied-contract claims may also defeat the arbitration agreement — because a document that expressly says it is not a contract is a poor place to put the one clause you need to be a contract.

A New Jersey federal court reached exactly that result in litigation involving Raymours Furniture. The handbook carried a prominent disclaimer stating that it was not a contract of employment, but the disclaimer did not exempt the arbitration clause from its scope, so the employee's agreement to arbitrate was not clearly and unambiguously established. The court added a second, independent ground: because the employer reserved the right to modify the handbook at any time without notice, the promise to arbitrate was illusory. Courts in other states have applied the same reasoning — an employer that can unilaterally rewrite or revoke an arbitration policy has not given consideration for the employee's promise to arbitrate.

Everything on the right belongs in the employment contract rather than the handbook. Leaving it inside a document that says it is not a contract is a combination that has repeatedly failed, on two independent grounds at once.

The layer most handbooks ignore entirely

Implied contract is a state-law question. Sitting on top of it is a federal one that applies whether or not there is a union anywhere near the business. Section 7 of the National Labor Relations Act protects the right of most non-supervisory employees to act together over pay and working conditions, and a workplace rule that interferes with that right is unlawful regardless of intent.

The Board's 2023 decision in Stericycle set the current test: a rule is presumptively unlawful if a reasonable employee could interpret it as having a tendency to chill protected concerted activity, and the employer then bears the burden of proving the rule is narrowly tailored to a legitimate and substantial business interest that cannot be served by something less restrictive. It applies retroactively to handbooks written before 2023. Enforcement priorities have shifted since, and the Board has spent long periods without a quorum, but Stericycle has not been overruled and administrative law judges have continued to apply it.

• Confidentiality rules broad enough to cover pay, or the terms of the handbook itself.
• Civility and professionalism policies phrased so that criticising management could breach them.
• Social media rules prohibiting disparagement of the company.
• Non-disclosure of investigations as a blanket rule rather than case by case.
• Bans on discussing terms of employment with colleagues, which are the clearest example of the problem.

The fix is narrowing rather than deleting. A rule that protects genuinely confidential business information is defensible; a rule that says employees may not discuss company matters is not. And a general savings clause at the front of the handbook saying nothing here is intended to restrict rights under the NLRA is not reliably sufficient on its own.

Can the employer just change it?

Generally yes, and almost every handbook reserves the right to. That reservation is what keeps the document flexible, and — as above — what can make anything inside it that needs to be binding fail. The practical position for an employee is that a handbook is not a set of vested entitlements: policies on remote work, bonuses, travel or leave above the statutory minimum can normally be revised going forward.

What cannot be changed retroactively is anything already earned. Leave that has accrued, commission that has been earned under the plan in force at the time, and wages for work performed are not policy questions. Where a handbook term has been incorporated into an individual contract by reference — which happens more often than employers intend — changing it may require the employee's agreement rather than a memo.

Reading it from the employee side

Two practical points. First, a handbook can help you. If it sets out a disciplinary procedure and the employer skipped it, that is a departure from its own stated process — evidence in an unfair treatment or discrimination argument even where the handbook is not contractual, and directly relevant to how a written warning or performance plan should be read. Keep a copy of the version in force when the events happened, because handbooks are silently updated.

Second, signing the acknowledgment is not a negotiation you can win, and refusing it changes very little — the policies apply either way, and the refusal is documented. If something in it genuinely concerns you, the useful move is to write, in the acknowledgment or by email, that you have received the handbook and are noting a specific question about a specific policy. That preserves the point without turning the first week into a dispute.

What a good handbook actually is

The best-drafted handbooks are shorter than the worst ones. They describe what the company does day to day, keep discretion visible in the language, put the disclaimer where it will be read, and route everything that has to bind — arbitration, confidentiality, IP, restrictive covenants — into signed agreements that stand on their own footing.

The failure mode is the opposite instinct: adding detail and formal-sounding procedure in the belief that it demonstrates fairness. It does demonstrate fairness. It also, in most US states, describes a promise — and a company that then departs from its own four-step process has handed the other side the argument.