TL;DR
- The real test: an item belongs in the handbook if an employee needs it to make a decision at work, and not otherwise.
- Default state: a thin handbook that everyone actually reads beats a complete one that nobody opens.
- What has to be true: someone has the authority to take policies out, not just add them, and they use it.
- How the options split: three honest categories exist, and forcing every policy into the same document is the choice to avoid.
- Decision rule: if the item changes more than once a year, or only a few people ever act on it, it doesn't live in the handbook.
- Outcome to expect: a shorter handbook, fewer acknowledgements to chase, and a faster answer when someone asks HR what the rule is.
The Handbook Is Not a Filing Cabinet
A people operations lead opens a Slack message at 4:47pm on a Thursday. The request is from a director who has just been to a conference: please add a policy on return-to-office scheduling, plus a clause about AI tool use, plus something on social media. The lead looks at the handbook. It's already 142 pages. Nobody has finished reading it in years. The lead's first instinct is to start drafting. The instinct is wrong. The real issue isn't which three policies to add. The real issue is that nobody has a rule for what stays out, and a document without that rule grows until the day an employee asks a question the handbook should answer and the handbook doesn't, because the answer is buried under content that was easier to write than to remove.
When the Reader Does Not Need to Act Yet
The handbook is genuinely fine. The team is under a hundred people. The last revision was within the past two years. An employee can find the answer to "how do I get paid" in under a minute. Manager procedures live in a separate manager guide. Statutory notices sit in an appendix that no one opens, and that's correct, because they exist to be posted, not read. This is a working system, and a rewrite would consume weeks for no gain. The right move is to leave it alone, run an annual review against the actual questions that came in this year, and revisit only when a real change forces the issue.
The handbook has friction but no risk. A new starter asks a question that lives in a procedure, not the handbook, and the team redirects them. That works until it doesn't. Friction shows up as the same question arriving in HR three or four times a quarter, or as a manager improvising an answer because they didn't know where the source was. The shape of the problem is a small one: a handful of policies sit in the wrong place, and a handful of procedures sit nowhere. Nobody has been harmed. A light audit, asking "where would an employee actually look" for the top ten questions from this year, finds the gaps. The fix is a handful of moves, not a rebuild.
Best tools for HR Operations
Real risk is now visible. A dispute has produced a discovery request that names the handbook. A manager has enforced a rule that was never written down, and an employee is asking how that's fair. Someone in legal has asked which version of the at-will disclaimer is in effect. None of these are hypotheticals, they're the kind of moment that shows up without warning. At this stage the reader is no longer choosing, they're responding. The job is to identify which policies are doing the work of protecting the organisation, which are decorative, and to cut the decorative ones fast so the protective ones can be reviewed with care. Cutting now is cheaper than cutting during a matter.
The edge case that looks like a problem but isn't. A board member wants a "complete" handbook. A peer company has a 200-page document. A vendor sells a template with 87 policies and the price looks like a discount. None of these are reasons to grow. A longer handbook doesn't mean a better one, and a template is a starting point, not a destination. The edge case to watch for is the moment a third party applies pressure that has nothing to do with what the reader's employees need. That's the moment to ask the only question that matters, which is the one in the next section.
The Five Questions This Reader Asks at 11pm
Do I really need to add this? Almost never. If the request came from a single event, the answer is a one-page guidance note, not a handbook clause. The reasoning is that policies are sticky: they survive the event, the people, and the director who asked for them. What goes wrong is that the policy outlives the problem it was written to solve and starts being cited in situations nobody imagined. The cost is years of misapplied text, and the cost is invisible until it isn't.
Can I keep it in an email and a manager briefing instead? Yes, for most procedure-style content. The reasoning is that what an employee needs to act on should be reachable, and what a manager needs to act on is better held in a place managers look. The risk of getting this wrong is that the policy drifts, and two managers end up enforcing different versions. The defence against drift is a single source of truth in a manager guide, with a one-line link in the handbook that points to it.
Will this be the policy that ends up in a hearing? Some policies do real work. A clear social media rule that respects the NLRB framework, a documented progressive discipline path, an at-will disclaimer with a signed acknowledgement: these are the policies that earn their place. The reasoning is that they're the ones an adjudicator will be shown, and they're the ones that have to be defensible. The cost of getting one of these wrong is the matter itself. The cost of getting any of the others wrong is a bad first impression in the matter, and that's enough reason to be selective.
What happens if I say no and the requester goes over my head? This is a real fear and worth naming. The reasoning is that "no" is a position, and a position without a written rule for saying it tends to lose to a director with a strong view. The defence is a written inclusion test, signed off once, so the next "no" isn't personal. Without the test, every refusal is a battle, and the handbook grows.
Who is allowed to take a policy out? This is the question that separates handbooks that work from handbooks that decay. The reasoning is that a document with no exit rule has only an entry rule, and entry rules are easy. The cost of not answering it now is that in three years the handbook is uncuttable, because every clause has a sponsor. The answer to this question is more important than the answer to any of the others, and it's the one the reader should write down first.
Three Honest Categories the Approaches Split Into
Single handbook, all content, periodic purge. Everything sits in one document. A named owner runs a review on a fixed cadence and removes anything that no longer earns its place. This works when the organisation is small enough that a single document can hold everything that matters, and when the named owner actually has the time and the authority to delete. It fails when the owner is too polite, or too junior, or when the review slides by a quarter and then a year. The concrete example is a 200-person professional services firm with a stable business and a People Director who has held the role for six years. The handbook is 70 pages, the review happens in January, and the owner has removed three clauses this year because they no longer reflected the way the firm actually works. This is the model that ages best when it's owned properly, and it ages worst when ownership is unclear.
Handbook plus separate manager guide. The employee-facing handbook holds only what an employee needs to make a decision. Manager procedures, including the harder conversations, sit in a separate guide that managers are trained on. This works when the manager population is large enough to warrant a different document, and when the line between "employee needs to know" and "manager needs to know" is respected. It fails when the line isn't respected, which is most of the time without a written rule. A reasonable example is a 600-person business with 80 people managers. The handbook is 45 pages, the manager guide is 60, and the two are linked by name. The weakness is that some procedures straddle the line, and the straddle is where disputes live. The fix is a default rule: if only a manager acts on it, it doesn't live in the employee handbook.
Handbook plus annex plus a living manager resource. The handbook covers the small core. Statutory notices and benefits plan documents live in an annex. Procedures, tool guides, and anything that changes often live in a wiki or a manager portal, with the handbook pointing to the source of truth. This works when the organisation has the discipline to maintain a wiki, and when the handbook is allowed to be short. It fails when the wiki rots because nobody owns it, and the handbook becomes a list of links to dead pages. A reasonable example is a 1,200-person company with a People Operations team of nine. The handbook is 30 pages, the annex is 20, and the manager portal holds 40 procedures. The weakness is real: the portal needs a named owner per procedure, and the moment a procedure loses its owner, it becomes folklore. The defence is a quarterly audit of broken links and orphan procedures, which is cheaper than the alternative.
Five Diagnostic Questions the Reader Can Self-Assess Against
When was the last time an employee used the handbook to settle a question on their own? Find the answer by looking at the tickets that came in this year, and asking which ones had a handbook link attached. If the answer is "rarely" or "we don't track that", the document isn't doing its job, and the fix is to make the top twenty questions findable, not to add new policies. Getting this wrong looks like growing the document to "be more useful", which makes the findability problem worse.
Which three policies would survive a discovery request without embarrassing the organisation? Walk through the policies with someone who has seen a real dispute. The exercise reveals which clauses are doing protective work and which are decorative. If fewer than half survive that walk, the document is over-extended. The right response is to compress the decorative ones into a manager briefing and let the protective ones be reviewed properly.
Who has the authority to remove a policy, and when did they last use it? Find the answer on paper. If there's no name, there's no removal function, and the document can only grow. If there's a name and the last removal was over a year ago, the function exists but isn't active, and the document will continue to grow. The right response is to set a quarterly review and to make one removal a target, because one removal a quarter changes the shape of the document inside a year.
How many of the policies in the handbook change more than once a year? Count them. The list will probably include anything tied to a benefit plan, a tooling stack, a regulatory window, or a tool vendor. Those items don't belong in a document employees sign once a year, because the document will be wrong the moment the underlying item changes. The right response is to move them to a source of truth that updates more often and link to them from the handbook.
What does an employee do when the handbook doesn't answer their question? Find out by asking three of them. The answer reveals the actual system, which may have nothing to do with the formal one. If the answer is "I ask my manager", then the manager guide is the real handbook, and the formal handbook is decoration. If the answer is "I ask HR", the team is the real handbook, and that's its own kind of risk during a busy quarter. The right response is to align the formal and the actual, which usually means writing less, not more.
Six Categories, and Where Each Should Live
Policies Employees Act on Daily
These are the items an employee opens the handbook to find. How time off is requested. How pay works. How to raise a concern. What the conduct standards are, in the language the NLRB framework expects. The reason this category earns its place is that the employee is the one making the decision, and they need a single source of truth to make it. The weakness is real and worth naming: the moment this section starts including every conceivable scenario, it stops being readable. A short, plain statement of the standard, with a link to a procedure for the edge cases, beats a long, comprehensive one. The trap is to write this section for the edge case and forget the median employee who needs the rule, not the nuance.
Policies Affecting a Small Population
These are the items that apply to a subset of employees: a sales compensation clawback, a parental leave extension for a specific group, a remote-work exception for a single role. The reason this category often doesn't earn its place in the main handbook is that most readers will skim past it, which means it isn't doing the work it was written to do. The right home is a role-specific addendum, linked from the handbook but not included in the body. The weakness is that addenda get lost, and a year later an employee in the affected role doesn't know it exists. The defence is a one-line mention in the relevant section of the handbook, with a direct link, and a list of who is in scope.
Manager-Only Procedures
These are the items only a manager acts on: how to run a difficult conversation, how to document a performance issue, when to escalate. The reason they don't earn a place in the employee handbook is that the audience is different. The right home is a manager guide, ideally a living one, with the handbook saying only that the guide exists. The weakness is that managers don't always read the guide, which means the procedure is held by the team and not the document. The defence is to make the guide short, searchable, and tied to the moments a manager actually faces, not the moments the writer imagined.
Statutory Notices
These are the items the law requires to be posted or provided, including federal and state-level postings that change as statutes change. The reason this category doesn't belong in the body of the handbook is that the body is a document an employee signs, and the notices are documents an employer posts. The right home is an appendix, refreshed on the local cadence the jurisdiction requires, and confirmed locally because the requirement moves. The weakness is that an appendix left alone goes stale, and a stale notice is worse than no notice. The defence is a named owner, a calendar entry for the refresh, and a check against the actual requirements in each location the organisation operates in.
Benefit Plan Documents
These are the summary plan descriptions and plan documents for health, retirement, and other benefits. The reason this category doesn't belong in the handbook is that the plan document is the legal artefact, and the handbook can't be the source of truth for two reasons: it can't update as fast as the plan, and a conflict between the two creates exposure. The right home is the plan document, with the handbook carrying a plain-language summary and a link. The weakness is that the plain-language summary drifts from the plan, and an employee reads the summary and the plan says something different. The defence is a rule that the summary is rewritten only when the plan changes, never in between.
Anything That Changes More Than Once a Year
This is the category the previous ones aren't. A tool stack, a security procedure, a tool approval workflow, a list of approved vendors. The reason this category doesn't belong in the handbook is that the document can't keep up. The right home is a wiki, a runbook, or a manager portal, with the handbook pointing to it. The weakness is the same one the third approach in the earlier section has: a wiki without an owner rots. The defence is a named owner per page, a quarterly audit, and a written rule that the handbook never tries to be the wiki.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| Stable business, no active dispute | Under 200 employees | Single handbook, named owner, annual review | Document is fine; resist the urge to add | One inclusion test, applied to the next three requests, with a yes or no each time |
| Growing team, manager base is widening | 200 to 800 employees | Handbook plus manager guide, line between them not yet written | Managers improvising; handbook says little about how | Write the line between the two, in writing, before adding anything new |
| Multi-state, several statutory windows | Any size | Handbook plus annex, annex refresh is not owned | Stale notices, conflicting summaries | A named owner for the annex, a calendar entry, and a written check against each location |
| Heavy tool stack, fast-moving procedures | Any size | Wiki plus handbook, but the wiki is unowned | Broken links, dead pages, employees asking instead of reading | A quarterly audit with a removal target, not an addition target |
| Recent dispute or active matter | Any size | Existing handbook, under review | Exposure is in the room | Identify the protective policies, review them with counsel, leave the decorative ones alone for now |
| Board pressure for "completeness" | Any size | Pressure is external, not driven by employees | Document grows because nobody has a rule for saying no | A written inclusion test, signed off at the top, applied to the next request |
| Templates or vendor content arrive | Any size | Strong template, weak ownership | Decorative content enters the document | Treat the template as a starting point, not a destination, and apply the inclusion test before anything lands |
| Existing handbook with no removal function | Any size | Add-only culture, every clause has a sponsor | Document is uncuttable, and that is the problem | Name a removal owner, set a quarterly review, and make one removal a target |
The Inclusion Test
A short test is more useful than a long one, and the one below is the one the publication's better older pieces carry. It's written to be applied, not admired, and the table that follows is the form to fill in when a request lands.
| Question | If yes | If no |
|---|---|---|
| Does an employee need this to make a decision at work? | Continue to the next question | It does not belong in the handbook. Consider a manager briefing, a wiki page, or a one-line email. |
| Is the policy stable enough to survive an annual review without going wrong? | Continue to the next question | It does not belong in the handbook. Move it to a source of truth that updates more often. |
| Would the policy survive a discovery request without embarrassment? | Continue to the next question | Rewrite it, remove it, or get advice before it lands. |
| Does the policy apply to most employees, or only a small group? | It earns a place in the body | It earns a place in a role-specific addendum, with a one-line link in the relevant section. |
| Has a written review against this test been recorded before the policy lands? | Add it, with a date and an owner | The policy does not land. The test exists so the next conversation is not personal. |
The reason this test is short is that a long test isn't used. The reason it's written in the negative is that "no" is the harder answer to give, and the test is built to support it.
Removing something already in there's harder than putting it in, and the difficulty is the whole point of this section. The first move is to name a single owner of removals, in writing, with a remit that includes the authority to take a policy out without the original sponsor's agreement. Without that remit, every removal is a negotiation, and the document will outlast the negotiation. The second move is a quarterly review on a fixed date, with a written record of what was considered, what was kept, and what was removed. The third move is a written answer to the question "who is affected if this policy goes away", and that answer has to be on paper, not in someone's head, because the head will change. The fourth move is a transition plan for anything being removed: a one-line communication to the people who acted on it, a manager briefing, and a date when the policy stops being enforced. The fifth move is to record the removal in the same place the addition is recorded, so the audit trail runs both ways. The hardest part isn't any of these steps. The hardest part is the first removal, and the value of doing it's that the second is easier.
| Step | Owner | Output | What it prevents |
|---|---|---|---|
| Name a single owner of removals, with written remit | HR lead | A named role and a one-page remit | Negotiated removals that take a year |
| Quarterly review on a fixed date | The named owner | A written record of what was considered, kept, removed | Silent drift of the document over the year |
| Written impact analysis for each removal candidate | The named owner | A short note on who is affected and how | A removal that surprises an affected group |
| Transition plan for any policy taken out | The named owner, with the relevant manager | A communication, a manager briefing, a stop date | A policy that is removed on paper and enforced in practice |
| Recording the removal in the same place as the addition | The named owner | A dated entry in the audit log | An unbalanced trail that records only growth |
What to Put in Writing
The decision is only as good as the paper it leaves behind. The artefacts below are the ones a future reader of the file, including a new HR lead, a counsel, or a judge, will look for. None of them are difficult to produce. All of them are skipped by teams that move on to the next request.
| Artefact | Who owns it | When it is written | What it prevents |
|---|---|---|---|
| The inclusion test, signed off at the top of the function | The HR lead | Once, and reviewed annually | Ad-hoc decisions that favour the loudest requester |
| A dated log of every policy added, with the requester and the reason | The named owner of the handbook | At the moment of approval | A document whose history is in someone's head |
| A dated log of every policy removed, with the reason and the impact note | The named owner of the handbook | At the moment of removal | An unbalanced trail that records only growth |
| A quarterly review record, with attendees and outcomes | The named owner of the handbook | On the day of the review | A review that slides and a document that drifts |
| A written line between the handbook and the manager guide | The HR lead and the senior manager group | Once, then reaffirmed annually | A document where the line is implied and the disputes live in the gap |
| A one-page index of what lives in the annex, with a refresh date per item | The named owner of the annex | Updated as items are added or changed | A stale annex and a missed local requirement |
| A written answer to "who is allowed to remove a policy" | The HR lead | Once, with the answer visible to the team | A document that can only grow |
| A one-page note for any policy that has been challenged or disputed | The named owner of the handbook | At the moment of the challenge | A repeat challenge that finds the team unprepared |
The list is short on purpose. A longer list isn't used. The reason to write each artefact isn't compliance, it's continuity. The HR lead who owns the function in two years isn't the one who owns it today, and the only way the decision survives the handover is on paper.
Questions to Ask Before You Commit
What is the rule for saying no. A bad answer is "we don't have one, we usually find a way to say yes." A good answer names a written test, with examples of recent decisions made under it.
Who can take a policy out, and when did they last do it. A bad answer is silence, or a name followed by a pause. A good answer is a name, a date, and a specific removal in the last twelve months.
What changed in the last year, and is the handbook still right. A bad answer is "nothing has changed" or "we updated the cover page." A good answer is a short list of substantive changes, with the policies that absorbed them.
What does the manager guide hold that the handbook doesn't. A bad answer is "they overlap, but it's fine." A good answer is a clear split, with the line written down.
What is in the annex, and when was each item last refreshed. A bad answer is "the usual things." A good answer is an index, with a date per item, and an owner per item.
How is the at-will position protected in writing. A bad answer is "we've a disclaimer somewhere." A good answer is a clear disclaimer in the handbook, supported by a signed acknowledgement, and a note that the disclaimer is the controlling statement where the document's own language might suggest otherwise.
How are conduct rules worded in light of the current NLRB framework. A bad answer is that the rules were drafted years ago and haven't been looked at since. A good answer is that the rules have been reviewed against the standard the Board applies, and the language is plain enough for an employee to understand what is and isn't permitted.
What is the version control story for the handbook. A bad answer is "we've a few versions in a folder." A good answer is a single source of truth, a version number per release, and a record of who approved each release.
What does the employee see when the handbook doesn't answer their question. A bad answer is "they call HR." A good answer is a short list of where the answer lives, with a contact named for each.
What is the plan for the next removal. A bad answer is that there's no plan. A good answer is a named candidate, a date, and an impact note in progress.
The Cost of Getting This Wrong
The invoice never arrives. A new policy is added in twenty minutes, the requester is pleased, the document grows by a page, and the cost is paid by the next person to look. The next person is usually an employee with a question the policy didn't address, and they give up and ask their manager. The manager improvises, because the manager guide doesn't cover this either, and the improvisation is the kind of detail that surfaces later. The second-order cost is the gap between the document and the actual practice of the organisation, and the gap widens every time a policy is added without a test. The second-order cost is also the time spent by HR answering questions the document should have answered, which compounds. So the right question isn't "can we afford to add this policy" but "can we afford to keep adding policies without a rule for what stays out", and the answer to that question is no, and the moment the answer becomes obvious is the moment after a matter, which is too late.
When You Are Ready to Go Further
If the decisions above have left you with a list of policies you suspect don't belong, and a manager guide that has never been written down, the next step is to look at how peer organisations have structured the same split. HROpsLab publishes independent comparison work on the platforms and approaches that hold these documents, and the reviews are written for HR leads who have already decided that the current setup is wrong and want to see what the alternatives actually do. The work is independent, the publication sells nothing, and the comparisons are updated as the platforms and the surrounding rules change. If that's the moment you're in, the next click is the one below.
Frequently Asked Questions
What absolutely has to be in a handbook?
The items an employee needs to make a decision at work: how pay works, how time off is requested, what the conduct standards are, how to raise a concern, and the at-will disclaimer where the jurisdiction recognises at-will employment. Beyond that core, the inclusion test is what decides the rest. A document that holds only this core is short, findable, and defensible. A document that holds more isn't automatically better, and the right test is whether the average employee can find the answer to a common question in under a minute.
Should benefits plan documents live in the handbook?
No. The plan document is the legal artefact, and the handbook can't be the source of truth for two reasons: it can't update as fast as the plan, and a conflict between the two creates exposure. The right pattern is a plain-language summary in the handbook, with a clear pointer to the plan document. The summary is rewritten only when the plan changes, so the two stay aligned. The risk of moving them in is that the summary drifts, the plan says something different, and an employee reads the wrong version first.
Where should manager-only procedures live?
In a separate manager guide, with the handbook saying only that the guide exists. The reason is that the audience is different, the moments are different, and the depth of detail is different. The line between the two should be written down, because without a written line the documents drift into each other, and the disputes live in the overlap. A reasonable rule is that anything only a manager acts on doesn't live in the employee handbook.
Do statutory notices go in the handbook?
They go in an appendix, not in the body. The reason is that the body is a document an employee signs, and the notices are documents an employer posts, and the refresh cadence is different. The appendix needs a named owner, a calendar entry for the refresh, and a check against the actual requirements in each location the organisation operates in. A stale notice is worse than no notice, and the defence is a written owner and a fixed cadence.
How long is too long for a handbook?
The answer isn't a page count, it's whether an employee can find the answer to a common question in under a minute. A long handbook that nobody reads isn't a long handbook, it's a non-handbook. The right test is functional, not numerical: ask three employees how they would find the answer to a real question, and time them. If any of them can't, the document is too long, and the fix is to move items out, not to add a table of contents.
Who decides what gets removed from the handbook?
A single named owner, in writing, with the authority to take a policy out without the original sponsor's agreement. Without that remit, every removal is a negotiation, and the document outlasts the negotiation. The owner runs a quarterly review, records what was considered, kept, and removed, and writes an impact note for each removal. The hardest part is the first removal, and the value of doing it's that the second is easier.
What do we do with policies that change often?
They don't live in the handbook. The handbook is a document an employee signs, and a policy that changes more than once a year can't survive that cycle without going wrong. The right home is a wiki, a runbook, or a manager portal, with the handbook pointing to the source of truth. The risk is that the source of truth rots without an owner, and the defence is a named owner per page and a quarterly audit. The handbook is the contract; the wiki is the practice.
Should we use a template we bought or were given?
Treat it as a starting point, not a destination. Templates arrive with 80-plus policies because the vendor doesn't know your organisation, and the ones that earn a place in your document are the ones that pass the inclusion test against your employees' actual decisions. The cost of accepting the template as-is is years of decorative content, and the cost of editing it down is a few days of work. The days are worth it.
HROpsLab is the independent review publication HR leads read before they change the system, and the work is free to use.
