TL;DR
- The core decision: whether to keep writing policy in a form optimised for legal review or rewrite it for the manager who needs an answer in two minutes.
- When doing nothing is right: when managers already apply your policies consistently and your misapplication rate is bounded by a handful of known cases a year.
- What has to be true for this to work: somebody in HR owns the running order of every policy, not just the wording, and they have authority to push back on legal review when it buries the answer.
- How the options split: the choices are about layout and weighting, not length, and they can be combined without losing the thread.
- Decision rule: if a manager at the end of a shift can't reach the action in under ninety seconds of scanning, the policy is the wrong shape regardless of how accurate it's.
- Outcome to expect: fewer escalations of routine decisions, faster manager-led resolution, and a smaller gap between what the document says and what the floor does.
A manager, a missing answer, a Friday
A regional operations lead at a 600-person retailer is standing in the back of house at 4:10 p.m. on a Friday. Two of her team have just come back from lunch smelling of drink. One is a five-year employee, the other is in week two. She needs to know, in the next ninety seconds, whether she sends the new starter home unpaid and writes up the longer-tenured one, or whether she does something else, and she reaches for the company handbook on her phone. The substance of the policy on alcohol and drug use is sound. The opening of the policy is "Purpose and Scope", the second section is "Definitions", and the action she needs is on page three under "Disciplinary Procedure, Step One". She doesn't get there. She calls HR. HR is at a school pick-up. She sends both home and writes up nobody, and HR opens on Monday to a noisy thread about whether the call was right.
This is the reader's week. It isn't an edge case. The handbook is accurate, the manager is competent, and the policy has been reviewed. The manager didn't fail to read. The document failed to put the answer where the manager was.
But the real issue isn't whether managers can read a long policy. Plenty of them can. The real issue is that policies are written for the reader who has thirty quiet minutes and applied by the reader who has ninety seconds and a worker still in the room, and the document has been optimised for the wrong one of those two people for most of its working life.
Best tools for HR Operations
When the current shape is fine
Not every HR lead needs to rewrite their policies. Some genuinely don't. Four situations below describe readers for whom the existing approach is defensible, in increasing order of pressure.
Stage one: low misapplication, low volume, stable workforce. A 90-person professional services firm has had two policy escalations in the last eighteen months and both were handled in-house with no counsel involved. Manager tenure is high, turnover is low, and the head of HR wrote most of the policies herself. She can describe, without checking, what each policy requires a manager to do. If a manager misapplies a policy, it usually means the situation is unusual. This is the shape of a handbook that doesn't need a redesign. What it needs is protection against drift. A light annual review, an owner named on each policy, and a habit of rewriting any policy that has been escalated more than twice in a year.
Stage two: visible friction, contained cost. A 240-person health practice has five policies that managers routinely ask about: PTO, bereavement, expense, social media, and remote work. Each one has been clarified more than once. Nobody is in trouble, but every HR business partner spends roughly a quarter of their week re-answering the same questions with the same answer they have already written. The cost is real but bounded. The fix here's targeted: take the five most-asked-about policies and rewrite their running order so the answer leads. Leave the rest alone for now. Resist the temptation to do all forty.
Stage three: real risk exposure. A 1,100-person retailer has managers across seven states, all reading the same handbook. Misapplication has produced three written warnings in the last year that legal reviewed and one that legal didn't like at all. The handbook was drafted three years ago, has been amended eleven times, and cross-references itself in ways nobody can follow. This reader is past the friction stage. They have exposure. They need a deliberate project, not a touch-up, and they need the work to be led by somebody who can challenge the existing draft rather than amend it.
Stage four: edge case, don't generalise. A 3,500-person employer in a heavily regulated industry carries policies that have to be written in a specific form to satisfy an external auditor or a sector regulator. The reader isn't free to change the running order because the regulator reads the document too. For this reader, the answer is to keep the regulator-facing document as it's and build a parallel manager-facing companion: a one-page summary per policy that leads with the action, with the full document behind it. The two can coexist. They often have to.
Five questions this reader asks at 11 p.m.
These aren't the questions an HR lead announces in a steering meeting. They're the ones that surface when the laptop closes and the day catches up.
"Am I overthinking this?" Probably not. If managers are calling HR for decisions that are in the policy, the policy isn't working as a policy. The job of a policy is to be usable without HR in the room, and the test of that's whether a manager with a full inbox will reach the answer before they decide to call somebody. If the answer is no, the document has a usability problem, not a comprehensiveness problem.
"How did I miss this for so long?" Because most policies are drafted, reviewed by counsel, sent out, and then quietly not measured. There's no metric that says "percentage of managers who reached the correct action without HR intervention", because that metric is hard to define and harder to collect. The absence of the metric isn't the absence of the problem. The way to spot it without a metric is to log the inbound. Every HR question that can be answered by reading a policy is a vote against the current shape of that policy.
"What if I rewrite and make it legally weaker?" That's a real risk and it's the one that stops most rewrites. The mitigation is to keep the substantive wording intact and change the running order. The substance of "termination only for just cause after two written warnings" can sit on page two in a policy on misconduct, with the action on page one in plain language. The legal exposure of a particular phrase doesn't change when the action appears earlier. The exposure from vague aspirational language is a different matter, and that's covered below.
"Whose side am I on if I rewrite this?" Yours, and the manager's, and the employee's. A policy that managers can't apply isn't pro-employee, because employees don't get consistent treatment. A policy that legal can defend but managers can't apply isn't pro-employer, because the employer gets inconsistent outcomes. The job is to find the version that both sides can live with, and that means somebody in HR has to hold the pen against both legal review and operational reality.
"If I do nothing, what is the worst that happens?" The worst isn't a single incident. It's a slow accumulation of decisions made under uncertainty, each one defensible on its own and inconsistent in aggregate. Three years from now, an employment claim lands on the desk and the manager's notes tell a story that doesn't match the policy. The policy was accurate. The application was not. That's the failure mode and it's hard to recover from.
Three honest categories the approaches split into
Approaches to the running-order question fall into three families, and most readers end up mixing two of them.
The legal-led document, kept as is. What it is: a policy written and reviewed by counsel, with the substance intact and the structure unchanged. The opening section is scope or definitions, the action sits later, and the document is read end to end before a decision is made. When it's right: when the regulator reads the document, when the policy carries statutory wording that can't be paraphrased, when the reader's exposure is to a regulator more than to a manager's Friday afternoon. When it fails: when managers stop reading before they reach the action. The cost isn't visible because the missed decisions aren't logged as "policy failed"; they're logged as "manager judgment" and they accumulate. A real example: a community healthcare provider that rewrote its medication-error policy after a serious incident, and the legal-led version is now sitting on a manager's desktop unread while she makes a decision about a near-miss in real time. The policy is right. The manager isn't using it.
The manager-led rewrite. What it is: a deliberate project to restructure existing policies so the action leads, using plain words, with the substance preserved and the legal layer behind it. The opening section is the action. The next is the conditions under which it applies. The legal wording is later. When it's right: when misapplication is the dominant failure mode and the substance of the policy isn't contested. When it fails: when the rewrite softens the substance by accident, or when it removes words that turn out to matter. The discipline is to keep the substance intact and change only the running order. A real example: a logistics operator that pulled the action to the top of every policy that touched a manager decision, and saw HR call volume on routine policy questions drop inside a quarter. The win was not better policy. The win was that managers reached the answer without a phone call.
The companion summary. What it is: a one-page or two-page summary per policy, written in plain language and leading with the action, kept next to the full legal document. The full document does the work for counsel. The summary does the work for the manager. When it's right: when the full policy can't be rewritten without losing something the regulator or counsel cares about, or when the reader wants a fast win without a long project. When it fails: when the summary drifts from the full policy because nobody owns both. The two must be reviewed together and updated together, or the manager follows one document and HR enforces another. A real example: a financial services firm that built a companion summary for each top-asked policy and tied the summary's review to the full policy's review cycle, with the same owner on both. The summaries got used. The full documents kept their legal shape.
Five diagnostic questions you can answer tonight
These aren't "are you thinking about this?" questions. Each one comes with a way to answer it about your own organisation using things you already have.
Question one: how many manager questions does HR answer each week that the policy already covers? You can answer this by tagging your HR inbox for two weeks. Anything that says "the policy says X but in this case…" is a candidate. Anything that says "what does the policy say about…" is a stronger candidate. Count both. If the number is high, the document has a discoverability problem. If it's low, the document may be fine and the issue is training, not structure.
Question two: when a manager reads the policy cold, how long before they reach the action? Sit with a manager who has not read a given policy recently, hand them a scenario, and watch what they do. Don't coach. Time them. If they reach the action in under ninety seconds, the running order is working. If they read three sections and call you, the running order is the problem. Do this with five policies and five managers and the pattern is clear.
Question three: which policies are escalated most often? Pull the last year's case notes. Rank the policies by incident count. The top three are where misapplication is concentrated. They're also where a running-order rewrite will produce the most visible benefit, because the inbound is already measured. Don't start with the bottom of the list. Start where the cost is.
Question four: does any one person own the running order of every policy, or does each policy drift on its own? Walk through your policies. For each one, name the owner. If you can't, that policy is drifting. If the owners are spread across HR, legal, and compliance with no single point of accountability, the running order won't stay where you put it. Pick one owner. Give them authority over layout, not just wording.
Question five: if a manager applies the policy wrongly, can you tell from the policy text why they got it wrong? Take the last two misapplication cases. Read the policy they read. Ask yourself, honestly, whether a reasonable manager with no context could have applied it differently. If the answer is yes, and the manager did, the policy is the cause. If the answer is no, and the manager did, the manager is. The first is a document problem. The second is a people problem. They look identical from the case notes and they require different fixes.
Six Drafting Choices, Reviewed
Leading with the rule versus leading with the rationale
What it is: putting the action sentence first, with the rationale and context below, instead of opening with purpose, scope, and definitions. The reader who needs to act gets the action. The reader who needs to understand gets the rationale on the same page.
Why it earns a place: because the manager at 4 p.m. on a Friday isn't reading for context. They're reading to decide. Leading with the rationale forces them to do two passes when they have time for zero. Most policies were drafted by people who care about the why. Most policies are used by people who care about the what.
Where it genuinely falls short: it can read as brusque when the policy is contested or politically sensitive, and the rationale is what makes the policy tolerable to the workforce. Bereavement, leave for family care, and accommodations are the obvious examples. There the rationale is part of the acceptance of the policy, and leading with the rule alone can land as cold. The fix is to keep the rationale, but move it to a clearly marked second section rather than burying it. The reader can choose which one they need first.
Prescriptive wording versus principles
What it is: writing "send the employee home and pay them for the remainder of the shift" instead of "the manager should consider the circumstances and take appropriate action". Prescriptive language tells the manager what to do. Principle language tells the manager what to think about.
Why it earns a place: because a manager with ninety seconds and a worker in the room doesn't have time to reason from principles. They have time to follow a rule. Prescriptive wording also produces consistency across managers, which is the point of having a policy at all.
Where it genuinely falls short: it can't anticipate every case, and a rule that doesn't fit the situation will be ignored or bent, and bending rules quietly is how inconsistency enters. The discipline is to write prescriptive wording for the common case and pair it with a clearly marked escalation path for the uncommon one. "If the situation matches conditions A, B, and C, do X. If any of those aren't met, escalate to HR before acting." That sentence is two sentences and it does both jobs.
Naming the decision-maker
What it is: writing "the line manager decides, subject to HR review within 24 hours" rather than "the company will determine the appropriate action". Naming the human who decides, and the human who reviews.
Why it earns a place: because managers won't apply a policy that has no obvious owner of the decision. If the document says "the company", the manager reads it as "HR". If the document says "the line manager", the manager reads it as theirs. Naming the decision-maker also makes escalation paths legible, because the reader knows who they're escalating from and to.
Where it genuinely falls short: it can backfire when the named decision-maker is junior and the policy carries real exposure, and the document ends up transferring risk to a person who doesn't have the authority to bear it. The fix is to pair the named decision-maker with a named escalation path. Junior manager decides on the day. Senior manager or HR reviews within a stated timeframe. The reviewer owns the consequence, and the on-the-day decision-maker owns the speed.
Worked examples
What it is: a short illustrative example embedded in the policy, showing how the rule applies to a concrete case. "For example, if an employee calls in sick the morning after a late shift, treat as sickness absence unless the manager has independent evidence of misconduct."
Why it earns a place: because the manager reading the policy isn't thinking in abstractions. They're thinking in cases. The worked example gives them a case that maps to theirs, and the mapping is faster than the reading. Worked examples also expose ambiguity in the rule, because a rule that can't produce a sensible example is a rule that can't be applied.
Where it genuinely falls short: examples create loopholes. A manager will read the example and treat its facts as the rule. "If my case isn't exactly the example, the rule doesn't apply." The discipline is to label the example as illustrative, not exhaustive, and to write more than one. Three examples reduce the loophole problem because no one example covers the territory. One example amplifies it.
Defined terms
What it is: a defined-terms section at the front or end of the policy, giving precise meaning to words that the policy relies on. "Misconduct means… Gross misconduct means…"
Why it earns a place: because vague words produce vague decisions, and vague decisions produce inconsistent treatment, and inconsistent treatment is the seed of most employment claims. Defined terms force the drafter to be precise and force the reader to be precise back.
Where it genuinely falls short: defined terms that the manager can't remember, or that are buried in a section the manager won't read, do no work. The discipline is to keep the defined-terms list short, to put the definitions close to the words being defined where the layout allows it, and to test the policy by asking: if a manager reads only the action sentence and the definitions inline, can they apply it? If yes, the definitions are earning their place. If no, the policy still depends on a section the manager won't reach.
Cross-references to other policies
What it is: explicit links from one policy to another where they interact. "This policy should be read alongside the policy on absence management and the policy on data protection."
Why it earns a place: because policies do interact, and the manager making the decision often needs two policies in front of them. A cross-reference is the document telling the reader that the decision is bigger than this section.
Where it genuinely falls short: cross-references are a way to push work onto the reader. "See also policy X" is the document saying "we didn't want to deal with this in full here, so we will let you assemble it." The reader with ninety seconds won't assemble anything. Cross-references earn their place only when the second policy genuinely changes the action. If it doesn't, the cross-reference is decoration and should be cut.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| Stable workforce, low misapplication, professional services feel | Under 250 | Single office or small multi-site, low regulatory load | Drift over time, not present failure | Light annual review, named owner per policy, no rewrite |
| Concentrated misapplication on a small number of policies | 250 to 1,000 | Multi-site, mixed workforce, HR team of two or three | Repeated inbound on the same handful of policies | Targeted rewrite of the top three policies, action leads, substance preserved |
| Broad misapplication across many policies, real exposure | 1,000 to 3,000 | Multi-state or multi-site, HR team with specialist roles | Inconsistent application, legal exposure visible in case notes | Full rewrite project, manager-led structure, parallel legal review |
| Heavily regulated industry, document also read by regulator | Any size | Sector regulator, statutory wording required | Cannot rewrite without losing something | Keep regulator-facing document, build parallel manager-facing companion summary |
| Senior managers make most policy decisions themselves | Under 500, owner-led | Owner is hands-on, low HR headcount | Documents drift because nobody owns them | Pick one HR owner for the running order, give them authority, light review |
| Multi-state operation with one handbook | 500 to 5,000 | Cross-jurisdiction, varied state environments | Inconsistent application across sites | Manager-led running order plus per-state addenda for the variations |
| Rapid growth, last rewrite is over two years old | 200 to 800 | Hiring fast, policies not refreshed, new managers on boarded | New managers apply old policies inconsistently | Rewrite the top five policies first, queue the rest over two quarters |
| Hybrid workforce with distributed decision-makers | 300 to 2,000 | Remote and in-person mix, async communication | Manager decisions made without the document in front of them | Companion summary for every top-asked policy, kept current, easy to find |
The Shape of a Policy That Gets Followed
The running order isn't a stylistic choice. It's a structural choice, and it's what determines whether the document is used or ignored. The order is built from the manager's path through the document, not the lawyer's path through the regulation.
| Element | Position | Purpose |
|---|---|---|
| The action sentence | First, standalone paragraph | Gives the manager the answer in the first read |
| The conditions of application | Second, short list | Tells the manager when the action applies and when it does not |
| The escalation path | Third, named owner and timeframe | Tells the manager when to stop and who to call |
| The worked examples | Fourth, two or three short cases | Gives the manager a case to map to their own |
| The rationale | Fifth, optional but present | Lets the reader who needs the why find it without burying the who |
| The full substantive wording | Sixth | Carries the legal weight, reviewed by counsel |
| Defined terms | Inline where possible, short list at end if needed | Keeps precision without forcing a second pass |
| Cross-references | Last, only where they change the action | Earns its place or it is cut |
The test of whether this order works is whether the manager reads only the first three elements and applies the policy correctly. If yes, the rest of the document is documentation for the reader who has time. If no, the document is the wrong shape.
This is the order that puts the answer where the manager is. It isn't the only order, and it doesn't work for every policy. Policies that exist to satisfy a regulator need a different first element. Policies that carry statutory wording need the wording preserved. The discipline is to keep the order intentional and to know which policies are running which order, and why.
Testing It on a Manager
The legal review of a policy is necessary and it isn't sufficient. A policy that passes legal review can still fail at the point of use. The test that exposes this is a read-aloud test with a manager who has not read the policy recently.
Sit with a manager. Hand them a scenario that requires the policy. Don't coach. Ask them to read the policy aloud while they decide what to do, and to mark the moment they have enough information to act. Time the read. Note the section they reached. Note whether they had to backtrack.
The test exposes three things that legal review doesn't. It exposes the time to action, which the legal review doesn't measure. It exposes the number of sections the manager had to read before reaching the action, which the legal review doesn't measure. And it exposes the cases where the manager reached the wrong answer because the document was unclear, which the legal review may not catch because counsel reads for risk, not for use.
| Test signal | What it tells you |
|---|---|
| Manager reaches action in under ninety seconds | Running order is working |
| Manager reads three or more sections before acting | Running order needs the action moved up |
| Manager reaches action but gets it wrong | Wording is the problem, not position |
| Manager gives up and calls HR | Document has a discoverability problem, possibly both position and wording |
| Manager reaches action correctly but cannot justify it | Worked examples or defined terms are missing |
| Manager reaches action correctly and can justify it | Policy is in good shape for use |
Do this with five policies and five managers. The pattern is the diagnosis. If the pattern is consistent across policies, the running order is the fix. If the pattern is inconsistent, the problem is in the individual policies and each one needs its own review.
The test is cheap. It's also the test that catches what the legal review can't. A policy that passes counsel and fails the manager is a policy that will be misapplied, and a policy that will be misapplied is the policy that costs the most in the long run.
What to Put in Writing
A policy decision that's not written down is a policy decision that has to be re-made the next time the situation arises, and the next person will make it slightly differently. The artefacts below are the minimum to make this decision durable.
| Artefact | Who owns it | When it is written | What it prevents |
|---|---|---|---|
| The policy register, with one named owner per policy | Head of HR | Reviewed quarterly | Drift and orphaned policies |
| The running-order decision, recorded for each policy | Policy owner | At each policy review | Quiet reordering that erodes the manager-led structure |
| The read-aloud test notes, for each top-asked policy | Policy owner | After each test cycle | Repeating the same misapplication patterns |
| The inbound log, tagging policy questions by policy | HR operations lead | Continuous, reviewed monthly | A misapplication problem that nobody sees because the data is not collected |
| The acknowledgement record, with the disclaimer wording used | HR operations lead | At each policy issuance | Implied-contract exposure from policy wording that contradicts the disclaimer |
| The cross-reference map, showing which policies interact | Policy owner | At each policy change | A change in one policy that quietly breaks another |
| The review schedule, with dates and owners | Head of HR | Annually | Policies that age out without anyone noticing |
| The escalation log, capturing how often each policy is escalated and why | HR operations lead | Continuous, reviewed quarterly | Repeated escalations treated as one-off incidents |
The acknowledgement record is worth pausing on. In the United States, the implied-contract exception to at-will employment is recognised in roughly thirty-eight states, and thirteen states don't currently recognise it, being Delaware, Florida, Georgia, Indiana, Louisiana, Massachusetts, Missouri, Montana, North Carolina, Pennsylvania, Rhode Island, Texas and Virginia. In states that do recognise the exception, courts have generally held that disclaimer language alone doesn't establish at-will employment and an implied agreement may still exist, while courts in most states treat a clear disclaimer as controlling over arguably contrary policy language. Courts look for specific promissory language, for example a statement that employees will only be terminated for just cause, rather than general aspirational wording, and they weigh all the evidence together including the disclaimer, the length of employment, the policies themselves and the employment history. A clear disclaimer combined with a signed employee acknowledgement is the combination generally recommended to support at-will status. None of this tells the reader what to do specifically. It tells them the shape of the exposure, and they should take local advice.
The National Labor Relations Board position is also relevant to writing policy. On 2 August 2023 the Board decided Stericycle, which overruled the earlier Boeing standard and adopted a stricter test for whether a workplace rule is lawful. Under Boeing, rules were treated either as categorically lawful or weighed against business justification. Stericycle remains in place and has not been overruled as of April 2026, and administrative law judges continue to apply it, though whether it survives is uncertain. On 27 February 2026 the Board issued Memorandum GC-03, which signals a shift toward more employer-friendly and less aggressive enforcement by the Board and its regional offices. The shape of this exposure is real and shifting, and the reader should take local advice on the application of either document to their own situation.
Questions to Ask Before You Commit
These are the questions to put to an adviser, a reviewer, or your own team before the rewrite locks in.
Ownership: "Who owns the running order of every policy, and do they have authority over layout, not just wording?" A bad answer sounds like "we all do" or "legal reviews them". Ownership that's shared is ownership that's absent.
Substance: "Is the substance of every policy preserved exactly, with the running order being the only thing that changes?" A bad answer sounds like "we tightened the wording". Tightening wording during a structural rewrite is how substance drifts.
Measurement: "How will we know whether managers are reaching the action in under ninety seconds?" A bad answer sounds like "we'll see". The measurement has to be specified before the rewrite starts, not after.
Legal exposure: "What is our position on the disclaimer and acknowledgement combination, and how does it interact with each policy's wording?" A bad answer sounds like "we've a disclaimer". Disclaimers interact with policy wording, and the interaction is where implied-contract exposure lives.
NLRB posture: "Have we reviewed each policy against the Stericycle test, and have we tracked the GC-03 memorandum's effect on enforcement?" A bad answer sounds like "we reviewed the policies last year". The position shifts, and a review older than a year may be stale.
Companion documents: "Do any of our policies require a regulator-facing structure, and if so, which ones need a parallel manager-facing summary?" A bad answer sounds like "all of them, legal wants the full document". Some do. Most don't.
Test cycle: "When did we last run a read-aloud test with a manager on each of our top policies?" A bad answer sounds like "we don't really do that". The test is the only thing that tells you whether the running order is working.
Update discipline: "What is the cycle for reviewing the running order, and who triggers it?" A bad answer sounds like "when we get a chance". The cycle has a date and an owner or it doesn't happen.
Drift detection: "How do we know when a policy has drifted from the manager-led structure back to the legal-led structure?" A bad answer sounds like "we'd notice". You wouldn't. The drift detection has to be specific.
Outcome target: "What does success look like in six months, and how will we measure it?" A bad answer sounds like "fewer calls". Specify the metric. Pick a baseline. Set a target.
The cost of getting this wrong
The cost of leaving policies in a shape that managers can't apply isn't the cost of any single misapplication. It's the cost of the pattern. Three years of small decisions, each one made under uncertainty, each one defensible on its own, each one slightly inconsistent with the one before, and the pattern is what a claim lands on. A claim doesn't arrive with a single bad decision in its hand. It arrives with a year of marginal decisions and an argument that the inconsistency was the employer's choice.
The cost is also internal. A workforce that has been treated inconsistently by different managers applying the same policy is a workforce that has stopped trusting the policy. Trust in a policy is rebuilt slowly and only by the next policy being applied the same way twice. So the second-order cost of getting this wrong isn't a settlement, it's the next two years of every manager decision being re-litigated because the workforce has learned that the policy is what the manager decides it's.
So the question isn't whether you can afford to rewrite the running order of your top policies. The question is whether you can afford not to. Because the cost of doing nothing is paid in the small decisions, and the small decisions are the ones that don't appear on any invoice until the invoice arrives.
When you are ready to go further
If the diagnosis above matches your situation, the next step is to see how your peers are handling it. We publish independent comparison work across the HR operations stack, written for HR leads who are evaluating their own setup rather than shopping for a particular product. Our comparison pieces cover policy management, handbook distribution, acknowledgement tracking, and the operational layers around them, and they're written without vendor sponsorship or placement. We are a review publication. We sell nothing. We don't supply software, payroll services, or legal advice, and we never have.
If a comparison piece would help you decide what to look at next, the case studies below are a reasonable starting point. If you would rather talk through your situation with somebody who has done this kind of rewrite before, the expert network on the site is set up for that conversation.
Frequently Asked Questions
How long should a policy be?
A policy should be as long as it needs to be to carry the substance and no longer. For most policies, that's between one and three pages in the running-order form. The length isn't the variable that determines whether a manager reads it. The position of the action is.
Should a policy include the reasoning behind the rule?
Yes, but not at the top. The rationale earns its place when it sits in a clearly marked second section, after the action. The reader who needs the rationale finds it without having to pass through it to reach the action. The reader who doesn't need the rationale can skip it without losing the decision.
Should the policy name the decision-maker?
Yes. A policy that says "the company" will be read as "HR". A policy that says "the line manager, with HR review within 24 hours" makes the ownership legible and the escalation path concrete. The named decision-maker also makes the policy defensible, because the chain of decision is visible in the document.
How prescriptive should the wording be?
Prescriptive for the common case, with a clear escalation path for the uncommon one. "If conditions A, B, and C apply, do X. If any don't, escalate to HR before acting" covers most situations and gives the manager a path for the ones it doesn't. Pure principle wording forces the manager to reason from first principles under time pressure, and most of them won't.
Do worked examples help, or do they create loopholes?
Worked examples help when there are at least two of them, and when each is labelled as illustrative rather than exhaustive. One example amplifies the loophole problem because the manager treats its facts as the rule. Three examples reduce the problem because no one example covers the territory and the manager has to read the rule itself.
Who should write policy?
The person who owns the running order should write the first draft. Legal reviews for exposure. The owner reviews for usability. The first draft should not be written by legal, because legal is optimising for a different reader. The first draft should not be written by the manager, because the manager is optimising for their own situation. The owner sits in the middle and holds both.
How should defined terms be handled?
Defined terms should sit inline where the layout allows it, with a short list at the end of the policy if they can't be inlined. A long defined-terms section at the front forces a second pass before the manager can apply the rule, and most managers won't make the second pass. The discipline is to keep the list short and to test whether the policy works with only the inline definitions in front of the reader.
HROpsLab is a review publication. We test, we compare, we tell you what we found. No vendor pays for placement. No vendor sees our work before we publish. Start for free and read what is there before you decide anything.
