You already know the week is wrong. That is not the problem. The problem is that when someone asks you to justify changing it, “I have too many meetings” is not an argument — it is a mood, and moods lose to a manager who has a roadmap.
A calendar audit converts the mood into an inventory. It takes about 90 minutes and it produces four numbers and a table. Neither the numbers nor the table are impressive on their own. What they do is make the next conversation about a specific recurring meeting rather than about whether you are coping.
Do this once properly. Then do it once a quarter, in twenty minutes, because the second one is mostly comparison.
What the calendar is actually a record of
Your calendar is not a plan. It is a ledger of claims other people have made on you, plus a few blocks you put there yourself and have already lost twice.
That framing matters because it tells you what to look for. You are not auditing your discipline. You are auditing who has standing access to your time, how they got it, and whether anyone ever revisited the decision. Most of what is on your week was agreed once, by someone who has since changed roles, for a project that shipped.
An audit that produces the conclusion “I should be more focused” has failed. An audit that produces the sentence “eleven hours a week goes to four recurring meetings, three of which predate my current role” has worked.
Before you start
You need three things.
Four weeks of history, not one. One week is always atypical — you were on holiday, or it was quarter close, or the offsite ate Thursday. Four weeks smooths that out and still fits in an afternoon. Skip a week that contained a holiday or a company-wide event and take the next one instead.
A spreadsheet. Any one. You need columns you can sort, and a calendar app cannot sort.
A decision to count honestly. This is the part that fails. You will be tempted to leave out the meeting you volunteered for, the one you enjoy, and the one where you are the reason it exists. Those are precisely the ones that determine what you can change. Count them.
You do not need a time-tracking tool. If you already run one, use its export. If you do not, do not install one for this — the calendar you already have is enough for the first pass, and the manual version teaches you something an automatic categoriser cannot.
Step 1 — Get the four weeks out of the calendar
Both major platforms will export, and both exports are more annoying than they should be.
Google Calendar. Settings → Import & export → Export produces a .ics file per calendar. If you would rather not deal with .ics, switch to the four-week view, screenshot it, and type the entries in. For a typical week that is 20 to 30 rows and takes ten minutes.
Outlook. File → Open & Export → Import/Export → Export to a file → Comma Separated Values, then pick the calendar and the date range. This one lands in a spreadsheet cleanly.
Either way, you want one row per meeting occurrence, with: date, title, duration, organiser, and number of attendees. Recurring meetings get one row per occurrence, not one row per series — the whole point is to see the aggregate.
Add three empty columns: Category, Role, and Would I notice. You will fill them in Step 2.
Step 2 — Categorise every block
This is the audit. Everything else is arithmetic.
Go row by row. Each meeting gets exactly one category. Resist inventing a new category — the value is in the forced choice, and the eighth category always turns out to be “important” wearing a hat.
| Category | What belongs in it | The question it answers |
|---|---|---|
| Decide | A decision gets made and someone leaves with authority to act | Could this have been made async, by the owner, in a document? |
| Produce | Work actually happens in the room — pairing, review, drafting, planning together | Is the work real, or is it a status meeting with a whiteboard? |
| Inform | Someone tells people things | Why is this not written down? |
| Align | Two or more parties reconcile positions that had drifted | Is this recurring because alignment keeps breaking, or out of habit? |
| Relate | 1:1s, skip-levels, onboarding, the coffee you owe someone | These are load-bearing. Do not cut them first. |
| Attend | You are in the room for visibility, coverage, or because you were never removed | This is the harvest. Almost all of your recoverable time is here. |
| Own | You called it, you run it, it exists because of you | Nobody else can kill this one. That is a feature. |
Then fill Role with one of: run (you own the agenda), contribute (you are expected to speak), listen (you are not).
Then fill Would I notice with yes, no, or someone would. Answer it as though you had been unavoidably absent — not as though you had skipped it, which is a different and more loaded question.
Three notes from doing this a lot:
- The Attend column fills up fastest for people two levels below where a decision gets made. That is not a personal failing; it is what happens when meeting invites propagate by role rather than by need.
- If a meeting is genuinely two categories, it is genuinely two meetings, and one of them is usually the reason it runs long. Note it and move on.
- A 1:1 you dread still goes in Relate. The audit measures shape, not sentiment.
Step 3 — Count what the calendar does not show
The export undercounts your week in four specific ways, and the gap between the export and reality is where most people’s sense of being buried actually lives.
Preparation and follow-up. Add a column and fill it honestly for each meeting: minutes spent before and after. A weekly review you prepare a deck for is not a one-hour meeting. Do not estimate this as a blanket percentage — go through the rows. You will find that four or five meetings carry almost all of the hidden load.
Ad-hoc interruptions. Nothing in the export shows the unscheduled call, the “quick question”, or the thread that ate forty minutes. For the next week only, keep a tally mark somewhere for anything that pulled you out of what you were doing for more than five minutes. You want the count, not the duration.
Fragmentation. Two hours of meetings arranged as four thirty-minute blocks costs far more than two hours arranged as one. The export shows the two hours. Step 4 measures the rest, and the fragmentation problem gets a post of its own because it is the number most people have never computed.
Time you are technically free but effectively on call. A 25-minute gap between two meetings is not 25 minutes of work. Mark those gaps; you will use them in the second number below.
Step 4 — The four numbers
Compute these from the four weeks, then divide by four to get a weekly average. Every one of them is a ratio, because absolute hours mean nothing without the denominator.
1. Meeting load. Total meeting hours ÷ your contracted hours. Include preparation and follow-up.
2. Longest unbroken block. For each day, the longest continuous stretch with no meeting in it, counting a gap under 30 minutes as no gap at all. Take the median across the twenty days, not the mean — one clear Friday will otherwise flatter you badly.
3. Listen share. Hours where Role is listen, as a proportion of total meeting hours. This is your recoverable time. It is also the number that survives being shown to a manager, because nobody has to concede that the meeting is bad — only that you specifically do not need to be in it.
4. Origination. Proportion of your meeting hours where Role is run. If this is near zero, your week is entirely composed of other people’s decisions, and every fix in the declining playbook applies to you before any of the meeting-design ones do.
Resist the urge to compare these to a benchmark. There is no credible public number for what a normal meeting load is — the figures circulating are vendor surveys with self-selected samples, and the honest answer is that the right load depends on your role, your seniority, and whether your job is coordination. Reading your own numbers is a whole post about why the comparison you want does not exist and what to use instead.
The comparison that does work is against yourself, one quarter later.
Step 5 — Read the result
Sort by category, then by total hours descending. Three patterns turn up, and each one has a different first move.
The recurring block. A handful of series, usually three to five, account for most of your meeting hours. This is the most common result by a wide margin, and it is the good one, because a small number of series is a small number of conversations. Start with the largest one where Role is listen.
The fragmentation pattern. Your total load is unremarkable but your median unbroken block is under 90 minutes. You do not have a meeting problem, you have a placement problem, and declining things will not fix it. Batching and buffers will.
The ad-hoc pattern. The calendar looks fine and the week does not. Your load lives in the interruption tally from Step 3, which means the fix is in norms and response expectations rather than in the calendar at all. This one is the hardest to argue, because you have no artifact — which is exactly why you kept the tally.
Most weeks are a mix. Pick the dominant one and ignore the others for now; fixing two things at once means you cannot tell which one worked.
Turning the audit into one specific move
An audit that stays in a spreadsheet has cost you 90 minutes and returned nothing. The output is one sentence you say to one person this week.
Take the single largest Attend + listen row. Write down:
- the series name and its weekly hours to you
- who owns it
- what you would do instead
- what the room loses if you are not there
Then say it. The version that works is not a complaint about volume:
“I’ve been in the Thursday partner sync for eight months and I haven’t had an input in the last two. I’d like to drop off and read the notes — if something comes up that needs me, tag me and I’ll join for that item. Happy to keep going if you’d rather; I just wanted to check whether it’s still useful.”
That last sentence is not weakness. It is what makes the request cheap to grant, and it is the difference between a request and an announcement. The full set of scripts, with what each one costs, is the next thing to read.
If the answer is no, you have still gained something: you now know that meeting is not negotiable, which means you stop spending energy on it and move to the next row.
Re-running the calendar audit
Put a 20-minute block in the calendar for the first week of next quarter, titled with what it produces rather than what it is. “Q4 meeting inventory” gets kept; “calendar audit” gets booked over.
The re-run is not a fresh audit. You already have the categories. You are answering three questions:
- Did the series you removed stay gone, or did it come back with a new name? Recurring meetings are remarkably good at reincarnating.
- Did the hours you freed get absorbed by something else within a month? They usually do, and the useful response is to defend the freed time explicitly rather than to conclude the audit did not work.
- Which category grew? A rise in Align almost always means an organisational change upstream that nobody told you about, and it is worth naming.
Keep the spreadsheets. Four quarters of these is a document that makes a case no single week ever can, and it is the only version of this argument that survives a reorg.
Common questions
How is this different from time tracking? A time log records what you did; a calendar audit records what was claimed. They answer different questions, and the audit is the one that names a person you can negotiate with. If you want both, the two-week log is the version that survives contact with an actual working week — most time-tracking attempts die on day four.
Four weeks of history feels like a lot of typing. Can I audit one week? You can, and it is better than nothing, but do not draw conclusions about recurring meetings from it. One week cannot distinguish a fortnightly series from a cancelled one, and fortnightly series are where a surprising amount of load hides.
My manager schedules most of my meetings. Does this still work? It works better, because the entire result points at one person and one conversation instead of at a diffuse sense of overload. The output changes: instead of declining, you bring the inventory to your 1:1 and ask which of these you should be in. Managers are usually surprised by the total, because they see their share of it and not the sum.
Should I include commute, lunch, and personal blocks? Include them in the export so the day’s shape is accurate, but exclude them from the meeting-load ratio. They matter for number 2 — a lunch block genuinely does break an unbroken stretch, and pretending otherwise gives you a number you cannot use.
What if the audit says my load is fine and I still feel buried? Then the load is not the problem and you have learned something worth 90 minutes. Look at number 2 and the interruption tally. Feeling buried at a moderate meeting load is nearly always fragmentation or unbounded availability, and both have different fixes than declining does.
What to read after the audit
- How Many Meetings Is Too Many? Read Your Own Numbers — How many meetings is too many has no credible public answer. Two ratios from your own calendar tell you more than any industry benchmark can.
- The Two-Week Time Log That Survives Contact With Work — Most time logs die on day four. The column set, the recording rule, and the two-week limit that make a time audit produce something you can act on.
- Your Calendar's Fragmentation Problem — A moderate meeting load spread across the day costs more than a heavy one stacked together. How to measure calendar fragmentation and what actually fixes it.
- What a Meeting Actually Costs Your Team — The meeting cost formula, why the dollar figure a calculator gives you persuades almost nobody, and the two numbers that land better in the conversation.
- Six Meeting Types, and Which Ones You Can Drop — Six types of work meetings, what each one is actually for, and a decision table that says whether to drop it, shrink it, make it async, or leave it alone.