TL;DR
- The decision is forced either way: doing nothing chooses one of the two options silently, so a quiet review is a decision in itself.
- When to leave well enough alone: friction only, no protected-class risk, no contractual exposure, and a credible path back to a single answer in a defined window.
- What has to be true for a reset to work: the gap is bounded, the change can be defended in writing, and the people affected can be reached in a way that doesn't look like you're rewriting history.
- How the options split: the size and shape of the gap push you toward enforcement, rewriting, or a hybrid that protects the rule for some and lets the past stay where it's.
- A working decision rule: if you can name, in one sentence, the practice you would want to see in place six months from now, you're ready to choose. If you can't, the reset is premature.
- Outcome to expect: the cleanest version ends with a written rule that matches what the organisation actually does, a record of how you got there, and no one surprised by what changed.
The Tuesday Morning Memo
A senior HR lead at a mid-sized professional services firm opens her inbox and finds a one-line question from the head of operations: "Are we still asking for three written references, or has that gone?" She knows the handbook says three. She also knows, because she has run the last four onboarding cycles, that managers have been hiring on one verbal and two written for over a year. Nobody complained. The new joiners are fine. The reference-checking vendor, a different one, has been paid without protest. The handbook is on the intranet, unchanged, signed off by the CEO two budgets ago. She now has to decide whether to send a one-line reply that quietly confirms the new practice, write a longer note to managers reminding them the rule is three, or treat this as the moment to fix the document so it matches what people are doing. None of the three feels right. The first ratifies a drift she never approved. The second looks arbitrary to managers who have been doing it the other way for twelve months. The third asks her to admit, in writing, that a policy she is supposed to own has been ignored for a year on her watch.
The real issue isn't whether the rule is three references or two. The real issue is that a policy which the organisation isn't applying consistently is now a document she has to either defend or rewrite, and the cost of either move depends on factors she can't see from her inbox.
When You Do Not Need to Act Yet
Most divergence between handbook and practice isn't a crisis. Some of it's a normal feature of a working organisation. The first job is to tell the two apart, because treating friction as failure is how you create the kind of policy churn that exhausts managers and trains them to ignore the next document you publish.
Best tools for HR Operations
Your current setup is genuinely fine. A small team with one written rule that everyone interprets slightly differently, where the interpretation has been stable for years and where no one has raised a concern, isn't a gap. It's a custom. The handbook describes the ceiling. The practice describes the floor. As long as no one is crossing either, the document and the behaviour are doing different jobs and that's fine. A three-person studio with an expense policy that no one reads because everyone files the same way and the finance team signs it off isn't broken. It's local. Don't write a new expense policy to fix a feeling.
Friction only. Practice has drifted in a way that produces a small amount of friction, and the friction is contained to the people who already know about it. A manager is mildly embarrassed when a new joiner asks for the official process and gets a different answer from the team. Someone in HR spends twenty minutes a month reconciling what the handbook says with what gets done. That's a real cost, and it's also a cost you can absorb. The signal that you've left this category is when the friction starts costing other people time, not just the people who already understand the gap.
Real risk, contained. Practice has drifted in a way that could create exposure, but the exposure is contained to a known group and the gap is small. A manager has been approving flexible-working requests that fall outside the written rule, but only for two named employees, both in a single team, and you know who they're. You can pull the manager aside, narrow the practice, and reset the rule without making the change look like a company-wide enforcement. You're still in scope to fix this quietly, and the document you write at the end can be a clarification rather than a reversal.
The edge case you can't leave alone. Practice has drifted in a way that touches a protected characteristic, a contractual entitlement, or a regulatory threshold, and the people who have been getting the looser interpretation are a recognisable group. Even if the group is small, even if no one has complained, you're now one resignation or one complaint away from a problem that the written policy makes worse, not better, because the document says one thing and the evidence of what actually happened says another. This is the version where the absence of a decision is itself the decision, and the decision is to ratify the gap.
The Five Questions You Ask Yourself at 11pm
If I rewrite the rule, am I lying about what happened? No, but only if you describe the change as a clarification, not a correction. If the new rule says "we check two written references" and you can't defend, in a sentence, why the old rule said three, the rewrite reads as a cover-up. The honest framing is that the old rule was written for a hiring market that no longer exists, and the new rule describes what the organisation has actually been doing. The risk is when the rewrite can't survive the question "what changed?"
If I enforce the old rule, who pays the cost? The managers who adapted the practice, and the new joiners who were hired under it. They're the ones who will experience enforcement as a reversal. The cost is reputational with the manager group and operational in the next hiring cycle, when managers who learned the rule informally will be slow to trust any future rule. The cost is rarely the work itself, because if managers have been doing it the other way for a year, the alternative works.
What does the document look like in court? This is the question you don't want to ask, and the one that decides a lot of cases. If an employee who was terminated under the old rule can point to a year of practice under a different rule and a signed handbook that was never enforced, you've handed them an argument you would rather not have. In states that recognise the implied-contract exception to at-will employment, courts weigh the policies themselves and the employment history alongside any disclaimer language. An unenforced policy is part of that evidence. It doesn't bind you, but it's something a decision-maker will look at.
Who else knows, and what do they think they know? A practice that two managers and a recruiter know about is fixable. A practice that the leadership team has signed off on, even informally, is a different conversation. If the head of operations has told the CEO that the team has been hiring on two references and the CEO has nodded, the gap isn't between handbook and practice, it's between handbook and the operating model. The reset is bigger and the political cost is higher.
If I do nothing, what does the gap look like in a year? A small, stable drift becomes a normal feature of the organisation. New joiners learn the practice from their managers, not the document. The handbook becomes a reference for the rare case, not the rule for the common one. In five years, you've a document that's fiction and an organisation that runs on folklore. That's not a disaster in itself. It's, however, an exposure you've chosen without choosing it, and the choice will surface when someone who never read the handbook is surprised by what it says.
Three Honest Categories the Approaches Split Into
The quick correction. A single memo, a single line of communication, a return to the written rule, a brief period of grace, then enforcement. This is the right move when the practice has drifted in a narrow, well-defined way, when the people affected are a known group, when the underlying rule still makes sense, and when the cost of the rule isn't the reason people stopped following it. A team that stopped following the three-reference rule because the hiring market shifted and the rule was not designed for the current shape of candidates. The short version: send a note, give managers thirty days, then hold the line. The failure mode is that the reason people stopped following the rule is still there, and the correction is a temporary fix. The rule will drift again. You will be sending another memo in six months, and the second one will land worse than the first.
The rewrite to match practice. The document changes to describe what the organisation actually does, with a clear written record of when and why the change was made. This is the right move when the practice is sustainable, when the underlying logic of the original rule no longer applies, and when the cost of going back to the old rule would be higher than the cost of formalising the new one. A team that has been letting employees expense home-office equipment up to a higher limit than the handbook allows, because the limit was set before remote work was normal. The short version: write the new rule, communicate it as a deliberate update, archive the old one. The failure mode is that you're ratifying whatever happened to develop, and whatever happened to develop may have been a workaround for a manager who should not have been making policy on their own. The rewrite legitimises a precedent you didn't mean to set.
The hybrid: keep the rule, run an exception. The document stays as it's, but a formal exception process is added that lets managers approve deviations in defined circumstances. This is the right move when the rule is broadly right but doesn't fit every case, when the people who need the exception are a known and limited group, and when you want to preserve the principle without forcing every team to live inside a rule that doesn't fit them. A flexible-working policy that says core hours are 10 to 4, with a formal exception process for teams that operate across time zones. The short version: keep the rule, name the exception, write down who can approve it. The failure mode is that the exception becomes the rule in practice, and you've not solved the gap, you've just moved it one layer down. If the exception is used for more than a small minority of cases, the rule is wrong and should be rewritten.
Five Diagnostic Questions You Can Answer Today
What does the handbook say, exactly, and where did it come from? Pull the current text and the version history. The honest question isn't "what does it say" but "what was it designed to do?" A rule that was written to solve a specific problem five years ago and that problem no longer exists is a candidate for rewriting. A rule that was written to manage a recurring risk and the risk is still there's a candidate for enforcement. The action is to read the policy with the original author if you can, and to ask what the rule was trying to prevent. If the answer is "I don't know", the rule was probably a copy of a template and is doing whatever the current practice is doing, which means the document is decorative.
How long has the practice been in place, and how did it start? A practice that has been stable for two years is a custom. A practice that started six months ago after a senior manager made an exception is a drift. The action is to look at the timeline. If you can't find the moment the practice began, it has been there long enough that the reset will be felt as a reversal. If you can find the moment, you can find the people involved and the conversation that needs to happen.
Who is following the practice, and who is following the handbook? A gap that affects a single team is a management problem. A gap that affects every team is a policy problem. The action is to map the practice across the organisation, team by team, and to count. A two-team organisation with a one-team gap has a different decision in front of it than a twenty-team organisation with a five-team gap, even if the percentage looks similar. The first is a conversation with one manager. The second is a rewrite.
What is the worst plausible outcome if the gap is left alone for another year? Not the worst possible outcome, the worst plausible one. A redundancy round in eighteen months that has to be defended in writing. A senior hire who leaves and claims the written rule was the one they relied on. A regulator asking for a copy of the policy and being shown a document that no one follows. The action is to write down the two or three scenarios that would make you wish you had acted, and to ask whether any of them are realistic. If none of them are, the gap is tolerable. If one of them is, the gap is the next conversation you should be having with the leadership team.
Can the people affected be reached in a way that doesn't look like you're rewriting history? A change that's announced as a clarification can be defended. A change that's announced as a correction implies that everyone who followed the practice was in the wrong. The action is to draft the announcement and to read it back as if you were a manager who had been doing it the other way for a year. If the announcement reads as an admission that the organisation got it wrong, it will land as one. If the announcement reads as a deliberate update in response to a changed environment, it will land as a decision.
Five Responses to a Gap, Reviewed
Enforce the written policy immediately
This is the move where you tell managers, in writing, that the rule is the rule, that practice has drifted, and that from a named date forward the policy will be applied as written. Past deviations aren't addressed. The message is: the document is correct, the practice was wrong, and the future is the handbook.
It earns a place because some rules can't be allowed to drift, and the cost of formalising a bad practice is higher than the cost of reversing it. Rules that touch protected characteristics, regulatory thresholds, or contractual entitlements fall into this category. A safety rule, a data-handling rule, an anti-discrimination rule. The drift may be small, but the cost of a written policy that's not enforced isn't zero, because the document is now part of the evidence of what the organisation actually does, and a decision-maker weighing that evidence won't be able to tell the difference between a deliberate practice and an accident.
The weakness is that immediate enforcement looks arbitrary to the people who have been doing it the other way, and it's. You're telling a manager that a practice they adopted, often for good reason, is now wrong, and you're not explaining why the practice was allowed to develop in the first place. The manager's question, which they won't always ask out loud, is "why was this okay for a year and not okay now?" If you don't have an answer to that question that you can put in the memo, you don't have a defensible enforcement.
Enforce it prospectively with notice
This is the move where you announce that the policy will be enforced from a named date, that the gap is acknowledged, and that managers have a defined period to come into line. Past deviations aren't retroactively punished. The message is: the rule is correct, the practice was a drift, the drift ends on a date you can name.
It earns a place because it does the thing immediate enforcement doesn't do, which is to acknowledge the gap without ratifying it. You get the rule back. You give the people who have been following the practice time to adjust. You create a written record that the change was deliberate, which is the part that matters when the change is later read as evidence of what the organisation agreed to.
The weakness is the period of notice itself. A thirty-day window is enough for managers who want to comply and not enough for managers who are going to push back. A six-month window lets the practice harden. Either way, you're running two systems at once for a defined period, and the people who run the looser system will quietly continue to run it because the deadline isn't yet binding. You will need to monitor compliance during the notice period, and the monitoring itself is a cost.
Rewrite the policy to match practice
This is the move where the document changes, in writing, to describe what the organisation has been doing, and the change is announced as a deliberate update. The message is: the rule has been updated to reflect how we actually work, here's what changed, here's when it changed, here's why.
It earns a place because the cleanest version of a handbook is one that describes what the organisation actually does, and the cleanest version of this conversation is one that ends with the document and the practice pointing at the same thing. If the practice is sustainable, if the underlying logic of the old rule no longer applies, and if you can defend the change in a sentence, the rewrite is the right move. It also closes the door on a category of exposure: an employee who can point to a written policy that has not been enforced for a year has a stronger argument than an employee who can point to a written policy that has been formally updated.
The weakness is that the rewrite ratifies whatever happened to develop, and whatever happened to develop may not have been a good idea. The practice that became the rule may have been a workaround for a single manager, an exception that metastasised, a response to a one-off problem that no longer exists. You're about to write, in a document that will outlast everyone in the current conversation, that the organisation has decided to do it this way. Read the practice back as if you were a new joiner who had no context. If the practice still looks like a workaround, you're locking in a workaround.
Run a formal exception process
This is the move where the policy stays as it's, and a formal exception process is added that lets managers approve defined deviations in defined circumstances. The message is: the rule is the rule, and here's how you ask to do something different.
It earns a place because the rule may be broadly right without being universally right, and the organisation may have a legitimate need for the exception that the rule was not designed to cover. A flexible-working rule that doesn't fit teams across time zones. An expense rule that doesn't fit a team that travels to a particular region. The exception process lets you keep the principle and lose the friction, and the written process makes the exception a decision rather than a drift.
The weakness is that exceptions become the rule. If more than a small minority of cases are being run through the exception, the rule is wrong, and the exception is the actual policy, and you're now in a worse position than you were before, because you've two documents and a process that everyone has learned to use. The exception process also requires someone to own it, and the owner will, over time, become the de facto author of policy. If that person is a good policy author, you're lucky. If they're not, the exception process will produce a body of decisions that no one has reviewed in aggregate, and the aggregate will, eventually, become the unwritten rule.
Leave it and accept the inconsistency
This is the move where nothing changes. The handbook stays as it's, the practice continues, and the gap is allowed to be a feature. The message is: the document is one thing, the practice is another, and that's how it's for now.
It earns a place because some gaps aren't worth the cost of closing, and because some decisions are best made by waiting for the next forcing event rather than by acting on a quiet Tuesday morning. A gap that affects a small group, that doesn't touch a protected characteristic, that has not been raised by anyone affected, and that can be closed in a single conversation when it becomes important, is a gap that you can leave alone. The leadership team has more important things to do.
The weakness is that doing nothing is a decision, and the decision is to ratify the gap. If the gap later becomes a problem, the defence "we knew about it and chose not to act" is a different defence from "we knew about it and chose to act." The first is harder to make, because it asks the decision-maker to accept that the organisation saw the risk and walked past it. You're also training the people around you that the handbook is a reference document, not a binding one, and the next time you need the handbook to be binding, you'll find that the habit of ignoring it has already been learned.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| Protected-class or regulatory exposure | Any | A rule that touches a protected characteristic or a regulatory threshold has drifted, regardless of how small the drift | The document is now part of the evidence of what the organisation actually does | Enforce prospectively with notice, document the gap, plan the rewrite for the next policy cycle |
| Single-team drift, known manager | One team | A single manager has been running a different practice, the gap is contained, no one else is affected | The manager is a respected operator and the practice is not obviously wrong | One conversation with the manager, narrow the practice, reset the rule informally, then formalise the outcome in writing |
| Cross-organisation drift, stable practice | Several teams | A practice has been in place across multiple teams for more than a year, the original rule no longer fits the environment | Managers have learned to ignore the handbook on this point, the document is fiction | Rewrite the policy to match practice, with a clear written record of why the rule changed and when |
| Rule is broadly right, exceptions are common | Several teams | The rule fits most cases but a defined group needs to do something different, the exception has become the default in that group | The exception is not in writing and is being approved by the wrong people | Run a formal exception process, name the circumstances, name the approver, audit the exceptions at the end of the first year |
| Gap is small, stable, no one has raised it | One or two teams | A practice has been in place for years, the cost of the gap is contained, the people affected are not asking for a change | The friction is a small amount of HR time and occasional manager embarrassment | Leave it, with a written note in the policy file recording that the gap was reviewed and accepted, and a date for the next review |
| Gap surfaced in a grievance or a leaver's claim | Any | The gap has become evidence in an individual case, the document and the practice are now both in the file | The document makes the practice look like a drift you chose to ignore | Take advice before acting on the individual case, then run a reset that does not depend on the outcome of the case |
| New joiners have been hired under the practice | Several hires | A practice has been used to onboard a defined group of recent joiners, the handbook says something different | Reversing the practice means telling recent joiners that the rule they were told about is not the rule | Rewrite the policy, with a communication that names the change and explains why, and that does not imply the joiners were misled |
Sizing the Gap Before You Act
Before you choose a response, you need a number, even a rough one, for how many people are affected and how long the gap has been in place. A gap that has been running for three months and affects a single team is a different decision from a gap that has been running for three years and affects the whole organisation, even if the rule in question is the same rule. The size of the gap changes the cost of the reset, the cost of leaving it, and the credibility of the explanation you'll need to give.
A useful exercise is to pull the relevant records and to write down, for each affected policy, the date the practice started, the teams that have been following it, the number of people on the receiving end of the looser or tighter interpretation, and the most recent moment the practice was visibly applied. The exercise doesn't have to be a formal audit. It has to be a record that you can put in front of the leadership team without being challenged on the numbers.
The table below sets out the findings that matter, what each one implies, and the response that the finding supports. The findings are listed in the order you're likely to encounter them, starting with the most common and ending with the ones that force a reset.
| Finding | What it implies | The response it supports |
|---|---|---|
| Practice has been stable for under six months, single team | A recent drift, probably traceable to a single decision | One conversation with the manager, narrow the practice, reset the rule informally, formalise the outcome in the next policy review |
| Practice has been stable for more than a year, multiple teams | A custom that has replaced the rule | Rewrite the policy to match practice, with a written record of why the rule changed and when |
| Practice affects a protected characteristic or a regulatory threshold | Exposure that does not shrink with time | Enforce prospectively with notice, document the gap, plan the rewrite for the next policy cycle |
| New joiners have been onboarded under the practice | The practice has shaped what people were told to expect | Communicate the change as a deliberate update, not a correction, and do not imply that the joiners were misled |
| One manager is responsible for the drift | A leadership problem, not a policy problem | One conversation with the manager, in writing, followed by a check at thirty days, then at ninety |
| The leadership team has informally approved the practice | The gap is between the handbook and the operating model | Bring the gap to the leadership team formally, with options, before acting |
| The gap has surfaced in a grievance or a leaver's claim | The gap is now evidence in an individual case | Take advice before acting on the case, run a reset that does not depend on the case outcome |
Announcing a Reset
A reset announcement is a small document that does a lot of work. It has to acknowledge that the practice existed, without implying that the practice was wrong. It has to name the new rule, without suggesting that the old rule was nonsense. It has to give managers and employees a way to ask questions, without opening a conversation that becomes a referendum on the change. Most reset announcements fail because they're written for the leadership team, not for the manager who has been doing it the other way for a year.
The first paragraph of a reset announcement should answer the question "what is changing?" in one sentence. The second paragraph should answer the question "why is it changing?" in two or three sentences, with the reason framed as a deliberate update in response to a changed environment, not as a correction of a past mistake. The third paragraph should answer the question "what do I do now?" with a specific instruction for managers and a specific instruction for employees, and a date by which the change is in effect. A fourth paragraph should name who to contact with questions, and should not be the HR lead's direct line, because the questions won't be about the change.
The language matters more than the structure. A reset that says "we've become aware that the policy has not been consistently applied" reads as an admission that the organisation was not paying attention. A reset that says "we've reviewed the policy in light of how our work has changed and are updating it to reflect current practice" reads as a decision. Both sentences describe the same gap. The second one is the one you can defend.
A reset that implies that the people who followed the practice were in the wrong will be received as a criticism by the people who followed it, and as a confirmation by the people who didn't. Neither reaction helps. The reset should be written so that a manager who has been doing it the other way for a year can read it and think "this is reasonable" rather than "I have been told off." If the manager can't think that, the reset isn't finished.
The table below sets out the moving parts of a reset, what each one does, and the failure mode if you get it wrong. The reset is a small document, but it's a document, and the parts of it aren't optional.
| Element | What it does | Failure mode if you get it wrong |
|---|---|---|
| One-sentence summary of the change | Tells the reader what is different | Reader has to read the whole document to find out whether it applies to them |
| Reason for the change, in plain language | Defends the change against the question "what changed?" | Reader assumes the change is a cover-up for a mistake |
| Effective date, named | Gives the reader a deadline they can plan against | The change drifts because the deadline is vague |
| Instruction for managers | Tells managers what to do in the next thirty days | Managers continue the old practice because the new rule is not yet operational |
| Instruction for employees | Tells employees what to expect, and what to do if they have been treated under the old practice | Employees who were treated under the old practice are surprised by the new rule |
| Contact for questions | Gives the reader a way to ask without emailing the HR lead | The HR lead's inbox becomes the policy helpdesk |
What to Put in Writing
A good decision is one that you can defend in writing a year later, and a defensible decision is one that has left a record of what was found, what was decided, and why. The record isn't the new policy. The record is the trail of working that produced the new policy, and the trail is what turns a good decision into one that survives a challenge. The table below sets out the artefacts that this decision needs to leave behind, who owns each one, when it's written, and what each one prevents. The list isn't exhaustive. It's the minimum.
| Artefact | Who owns it | When it is written | What it prevents |
|---|---|---|---|
| The current policy text and version history | HR operations | Already in place, pulled at the start of the review | A rewrite that contradicts the original rule without acknowledging the change |
| A short note on the gap, dated, with the date the practice started and the teams affected | HR lead | Within a week of the gap being identified | A reset that cannot answer the question "how long has this been going on?" |
| A decision memo naming the chosen response, the reason for the choice, and the alternatives considered | HR lead | Before the reset is announced | A reset that reads as a personal decision rather than an organisational one |
| A leadership sign-off, in writing, on the chosen response | Head of HR or equivalent | Before the reset is announced | A reset that is challenged by a senior leader who was not consulted |
| The reset announcement itself, with the date it was sent and the audience it reached | Internal communications or HR operations | On the day of the reset | A reset that is described differently by different parts of the organisation |
| A record of the questions raised during the reset period, and the answers given | HR operations | Updated during the reset period, summarised at the end | A reset that is relitigated in the next policy cycle because the questions were not answered |
| The updated policy text, with version history showing the change and the date | HR operations | At the end of the reset period | A policy file that does not show the change, leaving the next reviewer to rediscover the gap |
| A note on the exceptions, if any, granted during the reset period, with the approver and the reason | HR operations | At the end of the reset period | An exception process that runs without audit, and that becomes the rule by stealth |
| A scheduled review date for the policy, in the calendar | HR operations | At the end of the reset period | A reset that hardens into a new drift because no one is checking |
Questions to Ask Before You Commit
Who made the call to let the practice develop? If the answer is "no one", the gap is a drift and the reset is a clarification. If the answer is "the head of X", the gap is a leadership decision and the reset is a reversal. The bad answer to this question is a name followed by "but they have left," because that answer means the gap is now yours to explain and you don't have the context that produced it.
What did the rule cost the people who stopped following it? A rule that was expensive in time, money, or friction, and that people worked around, is a rule that was probably wrong. A rule that was cheap to follow and that people ignored anyway is a rule that was probably misunderstood. The bad answer is "I don't know," because it means the reset is being designed without understanding the behaviour it's trying to change.
What is the smallest change that closes the gap? The smallest change is the one that does the least damage to the people who have been following the practice. A one-sentence clarification is better than a full rewrite. A targeted exception process is better than a new policy. The bad answer is the one that proposes a larger change than the gap requires, because the larger change will be read as a power move and will produce resistance that the gap didn't have.
Who will be embarrassed by the reset? Every reset embarrasses someone. The head of operations who approved the practice informally. The manager who built a workflow around the looser interpretation. The HR lead who owns the handbook and didn't notice the gap. The bad answer is "no one," because it means the reset has not been pressure-tested against the people who will have to live with it.
What is the deadline for the reset, and what happens on that date? A reset without a deadline is a memo. A reset with a deadline is a change. The bad answer is a deadline followed by "and then we will see," because that's not a deadline, it's a date.
How will you know the reset has worked? A reset that doesn't have a way to be measured is a reset that can't be defended at the next review. The bad answer is "we will hear about it if it doesn't," because that's how the gap developed in the first place.
What is the cost of getting this wrong? The cost of getting it wrong isn't the work of doing it again. The cost is that the next reset will land on a workforce that has already learned to ignore the handbook. The bad answer is one that only counts the immediate work, because the immediate work is the cheap part.
Who is the audience for the reset announcement, and what do they need to hear? The audience isn't the leadership team. The audience is the manager who has been doing it the other way for a year, and the new joiner who was hired under the practice, and the employee who has been on the receiving end of the looser or tighter interpretation. The bad answer is one that's written for the leadership team and then sent to everyone else.
What will you say at the next policy review, when someone asks why the rule changed? The reset has to survive that question, and the question will be asked by someone who has not read the memo. The bad answer is a sentence that requires the listener to read three other documents to understand, because no one will read them.
What advice do you need, and from whom? A reset that touches a contractual entitlement, a protected characteristic, or a regulatory threshold needs local advice before the reset is announced, not after. The bad answer is "we will get advice if it becomes a problem," because advice after the reset is a defence, not a design.
The Cost of Getting This Wrong
The cost of a badly handled reset is rarely the work of the reset itself. The reset is a memo, a meeting, a change to a document. The cost is the second-order effect on the next policy change, the next time the handbook needs to be updated, the next time a manager has to choose between the document and the practice. A reset that's handled badly trains the organisation that the handbook is a draft, that practice is the rule, and that the next policy update is the next round of the same fight. The work of the reset is paid once. The work of the next reset is paid forever.
The second cost is the manager who stops trusting the handbook. A manager who has been told that the policy is one thing, who has done it the other way for a year, and who has then been told that the policy is the rule, is a manager who will, in the next policy cycle, ask whether the new rule is real or whether it's another document. The question is fair. The answer, if the reset has been handled well, is that the new rule is real because the organisation has just been through a process to put it in place. The answer, if the reset has been handled badly, is that the new rule is real until it isn't, and the manager will plan accordingly.
The third cost is the one that surfaces in a year, when an employee who was terminated or who was not promoted points to the gap as evidence of what the organisation actually agreed. In states that recognise the implied-contract exception to at-will employment, courts weigh the policies themselves and the employment history alongside any disclaimer language, and an unenforced policy is part of that evidence. The cost of that moment isn't the work of the moment. It's the cost of the year of small decisions that produced the evidence.
So the question to put to the leadership team, before the reset is announced, isn't "what should the new rule be?" The question is "what do we want the next reset to look like?" Because the reset you're about to do is also the reset you're about to teach the organisation to do, and the organisation will learn from the example more than from the memo.
When You Are Ready to Go Further
If the gap you're looking at touches more than one policy, or if you're not sure which of the five responses fits, the next step is a structured review of the policies in question against the practice, with a record of the gaps and a recommendation for each. HROpsLab's independent comparison work covers the policy management platforms that organisations use to run that review, with the trade-offs between them written for the reader who has to make the decision rather than for the vendor who has to make the sale. The work is independent. HROpsLab is a review publication, not a vendor, and the comparisons are written to be used, not to be admired.
If the gap you're looking at is one of a kind, a single policy, a known group, a defined response, the decision is yours to make and the sections above are the working notes. The diagnostic questions, the decision table, and the questions to ask before you commit are the parts to copy into a working document. The reset announcement, the policy file, and the audit at the end of the reset period are the parts that turn the decision into a record.
If you're not sure which of the two you're looking at, that's also a useful answer, and the next conversation is the one to have with the leadership team before the reset is announced.
HROpsLab's promise is the same in either case: the work is independent, the comparisons are written for the reader, and the publications aren't a pitch for a vendor. The recommendations in this article aren't a pitch for a platform. The diagnostic questions aren't a pitch for a service. The decision is yours, and the record is yours, and the next reset will be easier if the record is in good shape.
Independent comparisons, written for the reader. HROpsLab is a review publication, not a vendor, and sells nothing.
Frequently Asked Questions
What do I do when practice has drifted from the handbook?
Pull the current policy text and the version history, then map the practice across the organisation, team by team, with a rough count of how long the gap has been in place. The decision you're about to make depends on three things: how big the gap is, how long it has been in place, and what the gap touches. A gap that's small, recent, and contained is a different decision from a gap that's large, old, and cross-functional, even if the rule in question is the same rule. The response that fits each is set out in the decision table above. Don't act on the individual case before you've a position on the policy.
Should I enforce the old rule or rewrite it to match practice?
Enforce if the rule is still the right rule and the practice was a drift. Rewrite if the rule is no longer the right rule and the practice is the new answer. The hardest case is the one in between, where the rule is broadly right but doesn't fit every case, and the response is usually a formal exception process that lets the rule stay and the practice become a written exception. The decision rule is whether you can name, in one sentence, the practice you would want to see in place six months from now. If you can, you know which way to go. If you can't, the reset is premature.
How far back can a change apply?
A change to a policy generally applies from a named effective date forward, not retroactively. Disciplining past breaches is a separate decision, and it's rarely the right one, because the people who followed the practice were following what they understood to be the rule at the time, and a disciplinary process that punishes people for following the practice will be read as a punishment for the gap itself. The exception is the case where the practice is so far from the rule that the people following it should have known, and that case is rarer than it looks. Take advice before acting on an individual case.
Does inconsistent enforcement create legal risk on its own?
In states that recognise the implied-contract exception to at-will employment, courts weigh the policies themselves and the employment history alongside any disclaimer language, and an unenforced policy can become part of the evidence of what the organisation actually agreed. The exposure isn't automatic, and it isn't the same in every state, and the size of the exposure depends on the rule, the gap, and the individual case. Describe the shape of the exposure to the leadership team, name the rule that's at risk, and take local advice before acting. Don't rely on a disclaimer alone, because disclaimer language is one of the things courts weigh, and a clear disclaimer combined with a signed employee acknowledgement is the combination generally recommended to support at-will status, while a disclaimer on its own isn't.
How do I reset a rule without implying that everyone who followed practice was in the wrong?
Write the reset announcement as a deliberate update in response to a changed environment, not as a correction of a past mistake. The first sentence says what is changing. The second says why. The third says what to do. The language is "we've reviewed the policy and are updating it to reflect current practice," not "we've become aware that the policy has not been consistently applied." The two sentences describe the same gap. The first one is the one you can defend. Read the announcement back as if you were the manager who has been doing it the other way for a year, and if the announcement reads as an admission that the organisation got it wrong, rewrite it.
What do I do about one manager who ignores policy?
Have the conversation, in writing, with the manager, and follow up at thirty days and at ninety. The conversation names the policy, the practice, the gap, and the expectation going forward. It doesn't name the manager's past breaches, because the past is the organisation's problem, not the manager's. The follow-ups are short notes that record whether the practice has been narrowed. If the practice has not been narrowed by the second follow-up, the conversation is no longer a policy conversation, it's a management one, and the next step is a performance conversation. Don't run the policy conversation and the performance conversation at the same time, because the policy conversation is harder to recover from if it doesn't land.
Should I discipline people for past breaches of the rule they were not enforcing?
Usually no, because the people who followed the practice were following what they understood to be the rule at the time, and a disciplinary process that punishes people for the gap will be read as a punishment for the gap itself. The exception is the case where the practice is so far from the rule that the people following it should have known, and that case is rarer than it looks. Take advice before acting on an individual case, and don't run a disciplinary process in the same week as a reset, because the two will be read as connected even if they're not.
What if the leadership team has informally approved the practice?
The gap is no longer between the handbook and the practice. The gap is between the handbook and the operating model, and the reset is a leadership conversation before it's a policy one. Bring the gap to the leadership team in writing, with the options and a recommendation, and let the leadership team make the call. Don't announce a reset before the leadership team has agreed to it, because the reset will surface the gap, and the leadership team won't enjoy learning about the gap from the same memo that's supposed to fix it.
Independent comparisons for HR operations, written for the reader.
