Chapter 25: Timeline Construction — Making Fifty Vendors Move as One

"Time is what we want most, but what we use worst." — William Penn, Some Fruits of Solitude (1693)

Chapter Overview

The Reyes–Whitfield master timeline is four pages long, and one number on it is load-bearing.

Not the ceremony at 4:25. Not the first dance at 8:45. Not the 10:30 cut-off, which is contractual and immovable and is therefore not a decision at all.

It is 3:15 p.m. — the moment Chapter 13's build completes.

Everything before it is a dependency chain that has already been drawn. Everything after it is a wedding. And there is no slack in it at all: the tent is up at 11:00, the flooring at 12:15, the tables at 1:00, the linen at 1:30, the floral at 1:30–2:30, the place settings at 2:30, the lighting focused at 3:00, the work lights off at 3:30 — and the first guest arrives at 3:55.

Move any one of those and the wedding is being built while guests watch.

This chapter is where twenty-four chapters of documents become one document.

And it opens with something less satisfying than construction: an audit. This project is carrying eleven open findings, a contingency at 2.8%, and at least three numbers that were true in month four and have not been restated since. A timeline built on top of those is a timeline built on top of those.

In this chapter, you will learn to:

• Audit what you are carrying before you commit a timeline
• Build backward from fixed anchors rather than forward from a start
• Find the critical path and compute its float
• Put buffers where variance actually accumulates, and remove them where they are waste
• Sequence arrivals against a dependency chain that already exists
• Version and distribute one document to fifty parties without ambiguity

🏃 Fast Track: If you have production, project-management, or construction-scheduling experience, skim §25.3 and §25.4 and read §25.1 (the audit), §25.6 (buffers, and why most of them are in the wrong place), and §25.8 (versioning) in full. §25.6 is the one that differs from general project practice: a wedding's variance is not evenly distributed and three hours of it carry almost all the risk. Then do exercises B.3, B.5, and D.2.

📖 Standard: Everything, with a calculator and the project's accumulated documents to hand.

🔬 Deep Dive: Work case-study-02.md, which builds the complete Reyes–Whitfield master timeline — and starts by writing off two of the eleven open findings as unfixable.

25.1 The Audit

Before a timeline, an accounting. This is the section that does not appear in any other book about wedding planning and it is the one that should come first.

🚪 Threshold Concept: a timeline inherits everything

A master timeline is not a fresh document. It is the point at which every unresolved thing in the project becomes a time, or fails to.

An unfunded item becomes a moment when somebody discovers there is no money. An unassigned task becomes a minute when nobody does it. An estimate that was true in month four becomes a block that is the wrong length.

Which means the first act of timeline construction is not construction. It is: list everything the project is carrying, and decide, item by item, whether it is resolved, resolvable, or something you are going to live with.

⚡ Quick Reference: the audit, in four columns

Finding	Where it came from	Status	What the timeline does about it
(the item)	(chapter)	Resolved · Resolvable · Living with it	(a time, a name, a buffer, or nothing)

The third column is the one that matters, and "living with it" is a legitimate entry. A project with eleven open findings and five months to go will not close all eleven, and pretending otherwise produces a timeline that assumes resolutions that have not happened.

The three kinds of open finding

🎤 From the Field

The audit takes about two hours and it is the most uncomfortable two hours of any engagement.

It is uncomfortable because it is a list of everything you have not done, written by you, in one place, in month eleven. Every planner I know avoids it, does it late, or does a partial version.

And every one of them who has done it properly has found something they had genuinely forgotten — an unfunded item, a person who was never confirmed, a number that stopped being true in the spring.

Two hours, once, before the timeline. Not after.

🔄 Check Your Understanding

1. Why is "living with it" a legitimate audit status rather than an admission of failure?
2. A finding is classified "resolved but not transferred." What does the timeline do about it?

Verify

1. Because the alternative is a timeline that assumes resolutions which have not happened.

A project with eleven open findings and five months to go will not close all eleven. A planner who marks them all "to be resolved" has produced a plan that depends on eleven future events, and the plan will be wrong in whichever ones do not occur.

"Living with it" is a decision, and it produces different behaviour: a structural constraint you have accepted gets a buffer, a contingency line, or an explicit conversation with the couple — none of which happen if it is filed as pending.

And it is honest. A contingency at 2.8% five months out is not going to become 7% by September. Saying so lets you plan against it rather than hoping past it. 2. Puts it in — and this is most of them.

"Resolved but not transferred" means the decision was made and it exists in somebody's head, an email, or a case study, and not in any document a person will be holding on the day.

The timeline is where it becomes a time, and Chapter 26's run sheet is where it becomes a person. It is an afternoon's work and it is the cheapest afternoon in the project — Chapter 21's photography restrictions sat "resolved" for two chapters before Chapter 23 moved them into a contract.

25.2 Fixed Anchors

A timeline is built backward from things that cannot move, and there are fewer of them than people think.

⚡ Quick Reference: hard versus soft

	Hard — cannot move	Soft — can move, at a cost
Contract	A curfew · a licence hour · an access time · a removal deadline	A vendor's arrival, usually
Physical	Sunset · a tide · a fixed religious hour (Ch.21)	The length of a walk
Structural	The build's completion, if there is no slack in it	The build's start, if access allows
Human	A person who must leave · a child's bedtime · a flight	Almost everything else

The test for "hard": what happens if it moves by twenty minutes? If the answer is a cost, it is soft. If the answer is a breach, an impossibility, or a person who is not there, it is hard.

At Reyes–Whitfield there are exactly five hard anchors:

Everything else on four pages is soft, and knowing which five are which is what makes the document usable under pressure.

⚠️ Common Pitfall: treating the ceremony time as an anchor when it is not

A ceremony time is frequently soft — the officiant is available for a window, the light is workable across an hour, and moving it fifteen minutes costs nothing.

And planners treat it as the hardest point in the day, because it is what the invitation says.

The invitation is a real constraint and it is not the same constraint. Once printed, the arrival time is fixed for the guests. The ceremony can still start at 4:31, and frequently does, and the fifteen minutes of float there is one of the most useful in the day — as long as you know it exists.

25.3 Building Backward

Forward from a start time produces a wish. Backward from an anchor produces a schedule.

🕐 The method, in five passes

Pass 1 — Place the hard anchors. Five of them. Nothing else on the page.

Pass 2 — Work backward from each anchor through everything that must precede it, using real durations (Chapters 13, 14, 20, 23) rather than estimates.

Pass 3 — Work forward from the access time through the build's dependency chain (Chapter 13).

Pass 4 — Look at where the two meet. That collision is the critical path and §25.4 is about it.

Pass 5 — Add buffers, only where §25.6 says they belong.

Why backward

Because forward planning hides the failure.

A forward timeline says "tent 9:00–11:00, flooring 11:00–1:00, tables 1:00–2:00" — and every block looks reasonable, and the document ends at some time, and if that time is 4:10 the planner adjusts the last block by twenty minutes and moves on.

A backward timeline says: guests arrive at 3:55, so everything is complete at 3:15, so place settings start at 2:30, so floral starts at 1:30, so linen is on by 1:30, so tables are placed by 1:00 — and when it runs out of morning, it runs out visibly, at the top, where a change is still possible.

🧩 Productive Struggle

A venue permits access at 10:00 a.m. Guests arrive at 2:30 p.m.

The build: marquee 3 hrs · flooring 2 hrs (after the marquee) · tables 45 min (after flooring) · linen 30 min (after tables) · floral 45 min (after linen) · place settings 40 min (after floral) · lighting rig 2 hrs (can run parallel to flooring and tables; must finish before floral).

Does it fit? Show your working, and if not, name every option.

Worked

Backward from 2:30, less Chapter 13's 45-minute rule (everything complete 45 min before guests) = complete by 1:45.

	Duration	Must finish by
Place settings	40	1:45
Floral	45	1:05
Linen	30	12:20
Tables	45	11:50
Flooring	120	11:05
Marquee	180	9:05

The marquee must start at 9:05 and access is at 10:00.

It does not fit. It is 55 minutes short.

---

Every option, in order of what it costs:

	Saves	Cost
Negotiate earlier access — 9:00 instead of 10:00	60 min	Usually a fee. Ch.13 paid $300 for exactly this
Marquee installed the day before	180	A day's hire, and permission
Run lighting parallel and compress nothing else	0	Already assumed
Two crews on flooring	30–45	A labour line
Drop the 45-minute rule to 20	25	Guests arrive to a wedding being finished. Ch.13
Move guest arrival to 3:00	30	The invitation is printed (Ch.19). Only available before −10 weeks
Reduce the floral install	15	Ch.17 — and it is already fourth in a four-step chain

The right answer is almost always the first one, and the reason is that it is the only option that buys time without taking it from something else.

And the finding that matters more than the arithmetic: this had to be discovered in month six. Two of the seven options — the marquee the day before, and moving the arrival time — stop being available at −10 weeks when the invitations print.

25.4 The Critical Path

The longest chain of dependent tasks. Everything else has slack; this does not.

💰 Run the Numbers: finding it

A task is on the critical path if delaying it by ten minutes delays the whole day by ten minutes.

A task has float if delaying it by ten minutes delays nothing.

At Reyes–Whitfield	On the path?	Float
Tent erection	Yes	0
Flooring	Yes	0
Tables → linen → floral → place settings	Yes	0
Generator delivery	No	~4 hrs
Restroom trailer	No	~5 hrs
Bar setup	No	~2 hrs
Ice, first delivery (1:00)	No — but see below	45 min
Hair and makeup	Separate chain	See §25.7
Ceremony chairs	No	3 hrs

The critical path at this wedding is six consecutive tasks with zero float between them, and it occupies 11:00 a.m. to 3:15 p.m.

Which means: any delay before 11:00 is absorbable. Any delay after 11:00 is not.

🚪 Threshold Concept: the fragile hour

Most of a wedding day is robust. A vendor arriving twenty minutes late at 9:00 a.m. is a non-event; the same vendor twenty minutes late at 1:00 p.m. delays the wedding.

Which means the planner's attention should not be evenly distributed — and it usually is.

Find the hours where float is zero, and be physically present for them. At Reyes–Whitfield that is 1:00 to 3:15, and it is also, not coincidentally, the two hours when a planner is most tempted to go and deal with something else.

The compression test

Run once, before the timeline is issued.

⚡ The compression test

Take the critical path and ask, of each task: if this ran 25% long, what would you do?

Task	+25%	Response
Tent, 3 hrs	+45 min	Nothing available. This is the exposure
Flooring, 2 hrs	+30	Add a person, if the company has one
Tables, 45 min	+11	Absorbable
Linen, 30 min	+8	Absorbable
Floral, 60 min	+15	Reduce the install — Ch.17's order is designed to be placeable, not built, on site
Place settings, 40 min	+10	More hands. Anybody can lay a place setting

The output is not a plan. It is a list of which tasks you can do something about and which you cannot — and at almost every wedding the tent is the one you cannot.

Which is why Chapter 13's wind clause and the tent company's arrival time matter more than anything else on the page.

25.4a The Cascade

What happens when something slips — and the useful distinction is between a delay that absorbs and one that propagates.

⚡ Quick Reference: absorb or propagate

A delay absorbs when	A delay propagates when
The next task has float	The next task has none
The following block ends at a fixed anchor and can start late	The following block is an anchor
Somebody can add resource	Nothing can be added
It happens before 11:00 a.m. at Reyes–Whitfield	It happens after

The single most useful thing a planner knows on the day: which of the two is happening.

The three cascades

Every wedding has three, and they behave completely differently.

💰 Run the Numbers: a twenty-minute delay, in three places

1. In the build, at 9:00 a.m. The tent company arrives twenty minutes late.

Propagates?	Yes — fully. The chain has zero float
Everything shifts	Build completes 3:35, not 3:15
Guests arrive	3:55
Result	Twenty minutes of margin gone; the 45-minute rule becomes 20
Recoverable?	Only by adding people to a later task. More hands on place settings

2. In the people chain, at 1:00 p.m. Hair and makeup runs twenty minutes over.

Propagates?	No — that is what the hidden buffer is for
Ready at	3:20, against a published 3:00 and a held 3:30
Result	Nothing. Nobody knows
And this is the whole reason §25.6 exists

3. In the evening, at 8:15. The toasts run twenty minutes over.

Propagates?	Yes, and it hits a hard anchor
Cake, coffee, changeover shift	First dance 9:05, not 8:45
Dancing	10:30 − 9:05 = 85 minutes, not 105
Result	Twenty minutes taken directly from the party, which is ranked second of twelve
Recoverable?	Partially — compress the cake and coffee block, which has 8 min of float

Same twenty minutes. Three completely different consequences, and only one of them is invisible.

🚪 Threshold Concept: at a wedding, delay is not additive — it is directional

A build delay eats the margin before the guests. A people delay is absorbed by a buffer that exists precisely for it. An evening delay comes out of the end of the party, because the end is contractual.

Which means the same twenty minutes costs nothing, something, or the thing the couple ranked second — depending entirely on when it happens.

And it produces the day's operating rule: before you spend attention on a delay, ask which chain it is in and whether the next thing has float. Most delays at a wedding do not matter and a planner who treats them all identically will be exhausted by two o'clock.

The cascade brief

Give it to the couple, once, in month eleven, and it changes how the day feels for them.

📋 The Planner's Script

"One thing about tomorrow that I'd like you to know, because it'll make the day easier.

Things will run late. They always do, at every wedding, and most of it doesn't matter.

If we're ten minutes behind at two o'clock, that's nothing — there's slack in the afternoon and I'll have it back by four.

If we're ten minutes behind at eight fifteen, that is ten minutes of dancing, and I'll be quietly firm about the speeches for that reason.

So if you see me looking relaxed about something at lunchtime and less relaxed about something at half eight — that's why, and it isn't that anything's wrong."

25.5 The Two Chains

A wedding day is not one timeline. It is two, and they meet once.

⚡ Quick Reference: the build chain and the people chain

	The build	The people
Runs	7:00 a.m. – 3:15 p.m.	8:00 a.m. – 10:30 p.m.
Contains	Deliveries, installs, focus, setting	Hair, makeup, dressing, travel, photographs, the ceremony
Governed by	Dependencies (Ch.13)	Durations and people
Fails by	A task not finishing	A person not being ready
Location	The venue	Frequently somewhere else entirely
They meet		When the couple arrive at the venue

Planners build the first chain well and the second badly, because the first is a logistics problem and the second is eleven people in a house fifteen minutes away.

The people chain

🕐 From the Run Sheet: the morning, backward

Working back from: couple at the venue by 3:30
3:15	Depart the rental house	15-min drive
3:00	Ready. Photographs of the two of them getting ready are done
2:15	Dressing — 45 min, and it is always longer than anyone thinks
1:45	Buffer — 30 min. See §25.6
11:15–1:45	Hair and makeup, 5 people, 30 min each	Ch.23: this runs late structurally
11:00	Artist arrives and sets up
10:00	Detail items gathered in one place (Ch.23)

And the rule from Chapter 23 that applies here: build backward from a "ready" time that is 45 minutes before it needs to be, and tell nobody except the artist that the buffer exists. It will be consumed.

25.6 Buffers

The section that differs most from general project practice.

A wedding's variance is not evenly distributed. Padding every block by ten per cent produces a timeline that is 40 minutes long and still fails, because the variance is concentrated in about three places.

⚡ Quick Reference: where variance actually lives

	Typical overrun	Buffer?
Hair and makeup	+20 to +50 min	Yes. 30–45 min, hidden
Family photographs	+10 to +25	Yes. 10 min (Ch.23)
The ceremony	+5 to +12	Yes. 10 min
Toasts	+5 to +15	Yes. 10 min
Dinner service	±5	No — it is a staffing ratio and it is predictable (Ch.14)
Tent erection	±30	No buffer available. This is exposure, not variance
Cocktail hour	0	It is defined by its end
The first dance	0
Load-in tasks	±10 each	A single buffer at the end of the chain, not per task

Four buffers, totalling about 65 minutes, and they go in four specific places.

Everything else is either predictable or unbufferable, and padding it costs time that the four real buffers need.

The hidden buffer

A buffer that people know about is not a buffer.

🚨 When It Goes Wrong: the visible buffer

A timeline says "2:15–3:00 — dressing (buffer to 3:30)."

Dressing now takes until 3:30. Not through bad faith — because a stated end time is what people work to, and a buffer that is written down is an end time.

What to do instead: publish the earlier time and hold the later one.

The distributed timeline says "ready at 3:00." The planner's copy says 3:30 and nobody else has it.

And there is one person who must know: whoever is delivering the thing. The makeup artist is told the truth — "I've told everyone 3:00; I actually need you done by 3:30, and I'd rather you were calm than fast" — because an artist who thinks they are 30 minutes late will rush the last face.

Where buffers do not belong

After a fixed anchor. A buffer between the ceremony ending and cocktail hour beginning is not a buffer; it is a gap, and a hundred people will stand in it.

Between load-in tasks. Chapter 13's chain is dependencies, and a buffer inside it just moves the finish. One buffer at the end of the chain — the 45-minute rule — does the whole job.

And anywhere the couple can see it. A visible buffer is a licence.

25.6a Float, and How to Spend It

Float is the opposite of a buffer: a buffer is time you added, float is time the structure already contains. And most planners do not know where theirs is.

⚡ Quick Reference: where the float is at Reyes–Whitfield

	Float	Usable for
Before 11:00 a.m.	Up to 4 hrs on non-critical tasks	Absorbing a late arrival, a wrong delivery, a missing item
Generator delivery	~4 hrs	A supplier problem
Restroom trailer	~5 hrs
Ceremony chairs	~3 hrs
The ceremony start	~15 min	The single most useful float in the day — see below
Cake and coffee block, 8:25–8:40	~8 min	Recovering an overrun toast
The portrait session, 5:36	~10 min	Compressing if family photos overrun (Ch.23)
Cocktail hour	0 by design — it is defined by its end
Dinner service	0
8:45 first dance to 10:30	0. It is the party

The fifteen minutes at the ceremony

§25.2 said the ceremony time is frequently soft and planners treat it as the hardest point in the day.

Here is what that float is actually for:

And what it is not for: absorbing your own lateness. A ceremony that starts at 4:40 because the morning ran over has spent float that was there for the guests.

⚠️ Common Pitfall: spending float before you need it

The most common error with float is using it as slack in the plan — building the timeline so that the ceremony starts at 4:40 because that is easier.

Float that is built into the plan is not float. It is a later start time.

Plan to 4:25 and hold 4:40 in reserve, exactly as §25.6 holds the dressing buffer. The distributed timeline says 4:25.

🔄 Check Your Understanding

1. What is the difference between float and a buffer, and why does it matter operationally?
2. A vendor is forty minutes late at 8:30 a.m. and another is ten minutes late at 1:20 p.m. Which do you deal with first?

Verify

1. A buffer is time you added. Float is time the structure already contains because a task is not on the critical path.

Operationally the difference is that float is discoverable and a buffer is a decision. You find float by drawing the dependency chain and asking which tasks can slip without moving anything; you create a buffer by deciding to.

And the practical consequence: float is free and buffers cost. Every minute of buffer is a minute the day is longer or a block is tighter. Which is why §25.6 has only four of them and §25.6a's list is much longer.

Plus the trap: float that gets built into the published plan stops being float and becomes a later start. 2. The second one, immediately, and probably not the first at all.

At 8:30 a.m. the build has hours of float on non-critical tasks and the critical path does not start until 11:00. A vendor forty minutes late at 8:30 is a non-event unless they are the tent company — and if they are, it is the most important thing happening.

At 1:20 p.m. the critical path has zero float. Ten minutes there is ten minutes off the 45-minute rule, and it is the fragile hour (§25.4).

The general rule: the question is never how late somebody is. It is what the next thing is waiting for.

25.7 Arrivals

Fifty parties, and about eleven of them arrive on the day.

⚡ Quick Reference: the arrival schedule

	Governed by	Common error
Rental company	The dependency chain. They are first and everything waits on them	Booking a four-hour delivery window (Ch.13)
Tent company	Frequently a separate vendor from rentals, and first of all	Assuming they are the same call
Caterer	Their own prep chain (Ch.14)	Arriving before there is a kitchen to arrive at
Bar	After the bar is placed	Ch.15's ice, which is a separate delivery
Florist	Fourth in a four-step chain (Ch.17)	Scheduled first, because it is what the client is thinking about
DJ / band	Needs power and a clear floor	A band's 3-hour load-in (Ch.22)
Photographer	The people chain, not the build	§25.5
Officiant	Ceremony −30
Ice	Two deliveries (Ch.15)	One
Cake	Whenever there is somewhere to put it	It is somebody's aunt and she is not a vendor

⚠️ Common Pitfall: the arrival that has nowhere to go

A vendor arrives on time, at a venue where the thing they need does not exist yet.

A caterer at a raw site at 11:00 has no kitchen — Chapter 13's marquee goes up at 10:00 and is not usable until the flooring under the adjoining area is down.

A florist at 12:00 has no linen. A DJ at 1:00 has no power if the generator is at 2:00.

The check: for every arrival, name the thing that must already exist, and confirm its completion time from the dependency chain. It is twenty minutes of work and it finds two or three every time.

25.8 Versioning and Distribution

Fifty parties hold a document that changes eleven times.

🚨 When It Goes Wrong: two versions in one field

The caterer is working from the version sent at −6 weeks. The DJ has the one from −3 weeks. The photographer printed hers at −10 days and it is the current one.

Nothing has gone wrong yet. Then dinner moves twelve minutes, and three people are working from three different assumptions about when the toasts are.

This is the single most common cause of a wedding running late for no identifiable reason.

⚡ Quick Reference: the versioning discipline

Every version has a number and a date, in the header, in large type	"MASTER TIMELINE v6 — 28 August"
A version is never edited in place	v6 is a file; v7 is a different file
The change is stated at the top	"CHANGED FROM v5: dinner service 6:40 (was 6:35). Nothing else."
Distribution is logged	Who got which version, when
And the final version has a different name	"FINAL — 10 September" — not v9
Re-issue is confirmed	A vendor who has not acknowledged has not received it

The distribution matrix

Nobody gets the whole timeline.

⚡ Quick Reference: who gets what

	Receives
The couple	A simplified version — 12 lines, no load-in, no vendor names
The wedding party	Only the moments they are in (Ch.24)
Each vendor	Their extract — their arrival, their tasks, their dependencies, and the three things around them
The venue	The full build and strike
The planner	Everything, plus the hidden buffers

A vendor extract is not a shortened timeline. It is their part plus context — because a caterer needs to know when the toasts are even though they are not doing them.

And the couple's version is the one most often got wrong. A couple handed four pages will read it, worry about it, and ask about page three. Twelve lines is a service.

🔄 Check Your Understanding

1. Why does a vendor extract include three things the vendor is not doing?
2. Why is the final version called "FINAL" rather than "v9"?

Verify

1. Because a vendor who knows only their own tasks cannot tell whether a delay matters.

A caterer who knows dinner is at 6:40 will serve at 6:40. A caterer who also knows the toasts are at 8:15 and the first dance is at 8:45 knows what their overrun costs, and will tell you when they are running behind rather than absorbing it silently.

Context converts a vendor from an executor into a participant, and it costs three extra lines.

And the specific three: what happens immediately before them (so they know what they are waiting for), what happens immediately after (so they know what is waiting for them), and the next hard anchor. 2. Because version numbers do not tell you which is current and a word does.

At 2 p.m. on a Saturday, a vendor holding "v8" and one holding "v9" both believe they have the latest. Neither can tell without checking, and neither will check.

"FINAL — 10 September" is unambiguous from across a field, and anybody holding anything else knows immediately that they are holding the wrong thing.

And the corollary: if FINAL changes, it becomes "FINAL v2 — SUPERSEDES 10 SEPTEMBER" in large type, and it is re-issued with confirmation — because a vendor who has not acknowledged has not received it.

25.9 The Estimate That Expired

Chapter 20 established it and this is where it is enforced.

⚠️ Any estimate given to a client has an expiry, and the planner owns restating it.

"About half an hour" was true in month four. Three vendors planned on it. Nobody withdrew it.

The practice: before issuing a master timeline, re-derive every duration from its current source — not from the last timeline.

	Current source
Ceremony length	Chapter 24's rehearsal: 36:25, not the 35:15 assumed
Dinner service	Chapter 14's staffing ratio, against the current guarantee
Family photographs	Chapter 23's twelve combinations, not the couple's original list
Processional	Chapter 20's 45 sec/unit, not the rehearsal's 30
Hair and makeup	The artist's schedule, re-confirmed
Caterer's power draw	Chapter 18's finding — stated in month six, against a menu that changed

Six re-derivations, about forty minutes, and at least two of them will have moved.

25.9a The Variants

A master timeline is not one document. It is two or three, and the others have to exist before they are needed.

⚡ Quick Reference: which variants a wedding actually needs

	When it is needed	What differs
The rain version	An outdoor ceremony, or an outdoor anything	A different ceremony location, a flip, a decision point, and a decision-maker. Chapter 28
The compressed version	A build that starts late	Which tasks lose time and which do not
The extended version	Rarely — a delayed guest arrival at a venue with no curfew
The variant nobody builds	Somebody essential is not there	§"the person variant"

Two of these are worth building in full and two are worth a paragraph.

The rain version

Chapter 28 owns the decision. This chapter owns the schedule that follows it.

What differs is not just the location:

And the finding: the rain version needs a crew that the fair-weather version does not, at a specific time, and that crew has to be identified in advance even though they will probably not be used.

Which is a real cost. Chapter 13's principle: labour costed at $0 is charged to somebody's attention. Naming four people who might have twenty-five minutes of work at 1:00 p.m. is the cheapest form of insurance available, and it is free right up until the moment nobody did it.

The compressed version

One page, and it is a priority list rather than a timeline.

⚡ If the build is running 45 minutes late, what goes

Order		Saves
1	More hands on place settings — anybody can lay a place setting	15 min
2	Floral placed rather than adjusted — Ch.17's order is designed for this	10
3	Lighting focus reduced to a check — the uplighters are pre-angled	10
4	The 45-minute rule to 25	20
5	Ceremony held to 4:40 — spend the float (§25.6a)	15
6	Signage placed by whoever is free	5
7	Detail photographs of the empty room	Cut entirely

Seventy-five minutes available, in order, and the planner is not deciding this at 1:40 p.m. — it was written in month eleven, on one page, and it is in the production book.

The person variant

The one nobody builds and the one that comes up.

Somebody essential does not arrive. A vendor, an officiant, a person with a job.

Chapter 28 owns the response. What the timeline owes it is one line per essential person:

"If [person] is not here by [time], then [action], and [name] does it."

Six lines, and writing them takes fifteen minutes.

25.10 Summary

The timeline is the product, and it is the point at which every unresolved thing in the project becomes a time or fails to.

So the first act is an audit, not construction: everything the project is carrying, in four columns, with "living with it" as a legitimate status.

Build backward from hard anchors, of which there are fewer than people think — and the test for hard is: what happens if it moves twenty minutes? A cost is soft; a breach or an absence is hard.

Forward planning hides the failure at the end. Backward planning shows it at the top, where a change is still possible.

Find the critical path and note that it is short. At most weddings it is six consecutive tasks with zero float, occupying about four hours — and the planner's attention should be concentrated there rather than spread evenly.

Run the compression test: what would you do if each critical task ran 25% long? The output is a list of what you can act on and what you cannot, and the tent is almost always the second.

A wedding is two chains — the build and the people — and they meet once. Planners build the first well and the second badly.

Four buffers, about 65 minutes, in four specific places: hair and makeup · family photographs · the ceremony · the toasts. Everything else is predictable or unbufferable, and padding it steals from the four that matter.

A buffer that people know about is not a buffer. Publish the earlier time and hold the later one — and tell the one person who is delivering the thing.

For every arrival, name the thing that must already exist. Twenty minutes, and it finds two or three every time.

Version everything, number it, state the change at the top, log the distribution, and confirm receipt — because a vendor who has not acknowledged has not received it.

And re-derive every duration from its current source before issuing. An estimate given in month four has expired, and the planner owns restating it.

Spaced Review

From Chapter 13 — the load-in as a dependency graph. Why does that matter more here than a list of times?

From Chapter 14 — service duration. Why is dinner not buffered?

From Chapter 20 — the estimate that expired. State the practice.

📐 Project Checkpoint

Build the Reyes–Whitfield master timeline.

Start with the audit. This project is carrying eleven open findings from Chapters 14 through 24, a contingency at 2.8%, and at least three numbers that were true in month four. List them, classify them, and decide what the timeline does about each.

Then build it. You have: five hard anchors · Chapter 13's dependency chain completing at 3:15 with no slack · Chapter 24's ceremony at 36:25 · Chapter 14's dinner block · Chapter 23's photo timeline · Chapter 22's music arc · Chapter 18's cue sheet · Chapter 15's two ice deliveries · the rental house fifteen minutes away · and a drainage channel with two cable ramps on it.

Produce:

1. The audit, four columns, every open finding
2. The five hard anchors, and one thing everybody thinks is hard that is not
3. The master timeline, built backward, with the critical path marked
4. The compression test on every critical task
5. The four buffers, and where the hidden one is
6. The arrival schedule, each with the thing that must already exist
7. The distribution matrix and the version discipline

case-study-02.md builds all of it — and begins by writing off two of the eleven findings as unfixable.

Looking Ahead

Chapter 26 is the run sheet and the production book — the operational form of everything this chapter has just constructed.

A timeline says what happens when. A run sheet says who does it, what they need, and what happens if they cannot — and the production book is the object a planner is physically holding at 4:26 p.m.

Bring the master timeline and the distribution matrix. Chapter 26's first job is to turn one document into eleven.