Owned

The Website Playbook

The idea

A website has two halves: a shape the category expects, and coverage of every moment that matters. Most sites have only the first. They take the standard sitemap for their category and fill it with copy, so the pages look right and say little. A purely question-led site has the opposite problem: every answer is there, but buyers can't find the structure they expect, and the brand doesn't hang together.

So the method runs both halves and reconciles them:

1. Shape, top-down: the page types buyers in this category expect, plus the brand's story

• spine.

2. Coverage, bottom-up: the moments, the stage gaps, the biases and the jobs. This is the

• new part.

3. Reconcile: every job gets a home in the shape, every page in the shape earns at least

• one job, and jobs with no home become new pages. That's usually where the differentiation shows.

Then each page gets a dual brief, the business check and its measures.

The shape matters to being chosen, not only to tidiness. "Familiar feels safe" (Chapter 8): when a buyer can't find the case studies or the pricing they expect, they start to doubt you before they've read a word.

The flow at a glance

The shape and the coverage are worked out separately, then meet in the reconcile step. Everything after that applies to the reconciled site. Each quarter, coverage runs again, because moments change faster than the shape.

The website's footprint, and Kestrel in 60 seconds

The website is the only build lit in every live cell. Whatever a buyer hears from an ad, a pitch or an assistant, they check it on the site. So the site has to answer all 149 moments: it leads 31 and supports the other 118.

Kestrel in 60 seconds. Kestrel Advisory is the invented firm used in every playbook: a 60-person operations consultancy that helps mid-sized manufacturers find waste and lift output. Its main offer is a fixed-price, six-week operations review, which can lead into an improvement programme. Its strategy is in Part one, and its full moment map in Where you stand, part one.

• Who chooses: usually a committee. The COO starts it, the finance director approves the spend, the plant manager lives with it and procurement runs the paperwork. The owner, the in-house improvement lead, an investor and AI assistants all have a say.
• The reason to be chosen: "We find the savings on the floor in six weeks, at a fixed price, and the review pays for itself or you don't proceed." (Invented. A real firm would need proof for every part of that sentence.)
• The alternatives: a large consulting firm, a specialist lean consultant, the in-house improvement team, buying new equipment, or doing nothing. Buying new equipment matters most: capital spending competes for the same budget.

Its slice of the Score

The website's own part of the Selectability Score is two stages, in every moment: Answered (does the site answer this buyer's question?) and Bought (does the page they land on repeat the reason, show proof and make the next step easy?). It also feeds Found, because assistants and search can only send buyers to pages that exist.

Read the last row carefully. The site's own stages cost Kestrel only 8.5 points, because most buyers never reach the site to be disappointed by it. The site's bigger effect is upstream: with no page answering the problem, sector or comparison questions, there's nothing for an assistant or a search engine to find. That's why the Score's first fix, "be there before they call", is mostly this build, shared with Being named: a predicted 2 to 6 points, slowly.

The site's own audit, in step 3 below, starts from the same numbers: 14 pages, and 84 of the 149 moments with no page at all.

Step 1: the shape

Start with the page types buyers in this category expect to find. For B2B professional services sold to a committee, that's the list below. Each one exists because it answers a moment that comes up in almost every deal, which is why buyers have learned to look for it.

The story spine runs through the shape. Before any page is written, set the narrative every page repeats in its own way:

• The problem, in the buyer's words: "Output is flat and costs are rising, and we can't see where the waste is."
• The reason: savings found on the floor in six weeks, at a fixed price.
• The proof pattern: every claim carries a named plant, a before-and-after number and a quote from the plant manager.
• The way out: if the review doesn't find more than it costs, the client doesn't proceed.

The navigation and the home page tell this story in order. Every page type echoes it.

Step 2: coverage, the full grid

Coverage comes from Kestrel's moment map: 8 choosers, 7 problems, 10 journey steps and 5 buying modes, giving 64 live cells and 149 moments, each traced to its source. Where you stand, part one derives it in full: every chooser's question at every step, and what sits behind each one. Because the website is lit in every cell, this playbook works through all of it. Here is what the map asks of the site.

One question per cell isn't enough to brief a page from. Each cell also records a pain, a gain, a job to be done, an objection and what would move the buyer. The section briefs in Step 5 draw on those fields.

Step 3: gaps, biases and jobs

Score every moment on the three tests: worth winning, a true reason there, and open. The scores set the order of work. They don't decide which moments get a page. A moment worth winning that has no page is a buyer who gets their answer somewhere else.

Then audit the current site against all 149. Kestrel's old site (invented) had 14 pages and a twelve-post blog: home, services, approach, four generic solutions pages, one industries page, a wall of client logos, team, about, contact, careers and privacy. 84 of the 149 moments had no page at all. The other 65 landed on a page that touched the subject without answering the question.

Each gap becomes a job: a stage, a moment and a bias. Kestrel's grid produces dozens. These sixteen carry the most weight:

Every job names a person, a stage and a bias. That's what lets a writer, or an AI, produce the right block rather than generic copy.

Step 4: reconcile

Put every moment into the shape. Go through the shape page type by page type, then find homes for what's left.

Pages the shape had no room for. This is where Kestrel's differentiation shows:

• Kestrel compared: a hub that includes doing nothing, and one page for each alternative: a large firm, a lean specialist, the in-house team, and new equipment. "Before you buy another line, find out what this one can do" is the comparison that matters most.
• Build the case: a business case and board paper kit, a savings estimator, a sample review report, and a page for procurement and site safety. Committee deals stall at step 5. Nothing on the old site helped there.
• What if the review finds nothing? The finance director's fear, answered head-on, with the terms.
• Talk to a client like you: a client to call in each sector, with their permission. The checking step had no home on the old site, because the logos couldn't be rung.

Page families. 21 of the pages come from four templates: seven problems, four sectors, six roles and four comparisons. Add one page per case study. Each family gets one template and one brief, filled in for each problem, sector, role or alternative. That's how a 60-person firm keeps 47 pages consistent.

Priority sets the order, not the list. A common mistake is to build the sitemap from the top few moments. The result looks tidy and leaves most buyers without an answer. Every moment worth winning gets a home. The scores decide what's built first.

The page anatomy falls out of the jobs. List the jobs a page carries in stage order, and you have its blocks:

1. the buyer's question, answered first;

2. who it's for and not for;

3. what it is and what it costs;

4. proof beside the claim;

5. the reason, against the alternatives;

6. the way out, next to the price;

7. one next step.

The buying mode sets how deep each block goes. In committee buying, every block has to work when forwarded to someone who never saw the pitch.

The example sitemap

Every page earns its place through at least one of three signals, and the signals set its priority. No page exists without a reason, and every rejected page is listed with its reason too.

• Canon: buyers in the category expect it. It's on the category template, or on at least three of the three to five leading competitors' sites.
• Coverage: it answers moments from the grid. The page table counts them.
• Strategy: the positioning requires it: where the firm chooses to play, who it chooses to win, or its big idea.

The menus follow the journey: what the review is, the problems it solves, who it's for and the proof. A second menu holds help for the champion, insights and the company basics.

47 pages, plus one per case study. The old site had 14 pages and a blog. Every page below traces to its signals, and the moments column counts the moments it answers.

Build order. Priority decides the waves, not whether a page exists.

• Wave 1, at launch (18 pages): the home page, the 12 P1 pages and the 5 hygiene pages.
• Wave 2 (18 pages): the P2 pages, mostly the problem, sector and role families, the case hub and the references page.
• Wave 3 (11 pages): the P3 pages.

Rejected, with reasons:

• Four generic "Solutions" pages (transformation, strategy, digital, change). Deprecated: no moment needs them, and they were copied from larger firms' sites.
• The blog as a section. Merged: the twelve posts answered no buyer question. Their useful parts moved into the guide and the problem pages.
• 28 sector-by-problem pages ("flat output in food and drink", and so on). Rejected: the questions don't change enough to need a page each. The case hub filters by both instead.
• Region and city pages. Rejected for now: no moment in the grid asks where the firm is based. Revisit if assistants start shortlisting by region.
• A lean glossary. Rejected: two of five competitors have one, which isn't canon, and no chooser asked for it.
• A separate site-safety page. Merged into the procurement page, because the same person forwards both.

The coverage test, run before the sitemap is locked:

• No orphans. Every one of the 149 moments has at least one page that answers it.
• No passengers. Every page except the hubs and hygiene pages answers at least one moment. The thinnest is "Kestrel vs a lean specialist", with one moment. If it can't be written well, it merges into the comparison hub.
• No dead ends. Every chooser can get from their first question to the next step. The improvement lead's path, for example: the problem page, "For in-house improvement teams", "Kestrel vs in-house", the references page, "The first six weeks", then "Making the gains stick".
• No destructive cuts. Cutting a page whose moments nothing else answers is flagged, and demoting it is offered instead. Cutting "When output is flat" would orphan four trigger moments. Cutting the investor page would orphan two: the investor's scope and checking questions.

The anatomy library

Every page type has a standard order of sections, because people and machines both expect it. A buyer who lands on a comparison page looks for the side-by-side; an assistant reading a price page looks for the number. Variety goes inside the shape, never across it. This is the library Kestrel's 47 pages are built from. Required sections are in bold.

Four rules hold on every page type:

• The question is answered in the first sentence of its section. Machines read passages, and people skim. A section that warms up first loses both.
• Proof sits beside the claim it supports, not on a separate page the buyer has to find.
• The way out sits beside the price. A finance director should never read one without the other.
• One next step. A page with three calls to action asks the buyer to make a choice the page should have made.

Step 5: the page brief, then a brief for every section

Every section on a page carries one job, so every section gets its own brief. The page brief doesn't disappear; it holds what no single section can: the moments the page serves, its place in the shape, the order of the sections, and the arc from the first question to the next step.

Why brief each section:

• One job each. A page carries five to eight jobs. Brief them together and they blur into general copy. Brief them separately and each section can be checked against its own job.
• The same test, one level down. A section that can't name its job gets cut.
• Better AI output. A section generated from a tight brief is sharper than a page generated from a loose one, and it can be checked and regenerated on its own.
• Machines read passages. Assistants and search systems typically pull a passage, not a whole page. Each section should stand alone: its question answered first, its own facts, its own proof.
• Reuse. A section such as "What if the review finds nothing?" can sit on several pages, briefed once and kept consistent.

The page brief comes first. Here it is for Kestrel's operations review page:

PAGE The operations review (service page)
MOMENTS (20) COO "who does this for plants our size, in our sector?"
(comparing, by sector); COO "fix it ourselves, buy equipment or bring someone
in?" (options, by problem); finance director "fixed price or day rates? What
if it finds nothing?" (comparing); assistant "what does an operations review
cost, and what does it return?" (building the case)
SHAPE Service page; links to the seven problem pages, the manufacturing hub,
the role pages, Price and terms, and Kestrel compared
THE ARC Recognise the problem, see who it's for, see what you get and what
it costs, believe it, feel safe, take the small next step
SECTION ORDER 1 The question answered 2 Who it's for 3 The six weeks and
the price 4 Proof 5 Compared with the alternatives 6 What if it finds
nothing? 7 Next step
THE REASON (echo in every section) Savings found on the floor in six weeks,
at a fixed price
FEELING TO LEAVE "These people understand a plant like ours."

Then each section gets its own brief. Two of the seven:

SECTION 6 What if the review finds nothing?
JOB Bought · finance director, comparing · losses loom larger
THE ONE CLAIM If the review doesn't find savings worth more than its fee,
you don't proceed and you don't pay for the programme.
PROOF BESIDE IT The clause, quoted from the standard terms; how many reviews
have triggered it (from the firm's records)
FACTS, WORD FOR WORD FROM THE ONE TRUTH Review fee range; the threshold;
what happens to the findings
THE HUMAN TAKEAWAY "The worst case is that we learn something and walk
away."
THE SENTENCE A MACHINE SHOULD QUOTE Kestrel Advisory's operations review is
fixed-price, and clients don't proceed to the programme if the savings found
are less than the fee.
LEADS TO Next step (section 7)

SECTION 4 Proof
JOB Believed · finance director and COO, comparing · we follow other people
THE ONE CLAIM Mid-sized plants like yours have found savings in six weeks.
PROOF BESIDE IT Two named plants, each with a before-and-after number and
the plant manager's own words (with permission)
FACTS, WORD FOR WORD FROM THE ONE TRUTH Plant size, sector and the period
measured, for each case
THE HUMAN TAKEAWAY "They've done this for a plant like ours, and it showed
in the numbers."
THE SENTENCE A MACHINE SHOULD QUOTE [Plant name], a [size] [sector] plant,
cut [measure] by [amount] after Kestrel Advisory's six-week review.
LEADS TO Compared with the alternatives (section 5); the full cases in the
case study hub

The flow check. Once every section is written, read the page top to bottom against the arc in the page brief. If it reads as a stack of parts, fix the joins, not the sections. Every field in every brief comes from the moment grid, the jobs and the one truth, so any section can be regenerated when a moment changes, without rewriting the page.

The homepage: wireframe and section briefs

The home page is the hero page: it gets full prose drafts, not just direction. It has one job above all: tell the story spine and send each chooser to their page.

Page brief. Type: hub. Positioning wedge: savings found on the floor in six weeks, at a fixed price, for mid-sized manufacturers. Target queries (five at most): "operations consultants for mid-sized manufacturers"; "fixed-price operations review"; "how to reduce waste in a manufacturing plant"; "operations review cost"; "Kestrel Advisory reviews". Navigation label: Home. The arc: recognise the problem, see it's for you, see how it works and what it costs, believe it, feel safe, take the small next step.

The wireframe shows structure, not design: what each section holds, in what order, and the job it does. Its brief is in the table below.

Sections 1, 3, 4 and 8 are required. The rest are recommended for committee buying, and the order follows the arc. The flow check reads the page top to bottom before it's approved.

The finished example: "When output is flat", written in full

A brief is a promise. This is what one turns into. "When output is flat" is a P1 problem page in wave 1. It answers 10 moments, for the COO, the owner, the finance director, the plant manager and the assistant. Each section below carries its job in italics, then the copy as it would go live. Kestrel, its plants and its people are invented.

1. The problem in the buyer's words

Known, Found · COO · familiar feels safe

• Output's flat, the lines are busy, and you're still shipping late

If your plant is running flat out and output hasn't moved in a year, you're not short of effort. You're short of a clear view of where the capacity goes. In most mid-sized plants we review, it's already in the line, lost to changeovers, small stops and waiting, long before anyone needs a new machine.

2. What's usually behind it

Answered · COO and plant manager · a hard choice stalls

Flat output in a busy plant usually comes from three places, in this order:

1. Changeovers that run long because nobody owns the clock. In the reviews we've run,

• they're the largest single loss more often than not.

2. Small stops: the jams and waits under five minutes that never reach the downtime log.

3. Planning that loads the bottleneck line with the wrong mix, so it waits while others run.

New equipment fixes none of these. It adds capacity to a line that's already losing its own.

3. What to try first yourself

Answered · plant manager and improvement lead · losses loom larger

Before you call anyone, including us: time five changeovers on your bottleneck line, start to first good part. Log every stop over 30 seconds for one week, by hand. If the two together cost more than a shift a week, the capacity is there to find. If they don't, your problem is probably demand or planning, and the guide below will help more than a review will.

4. When outside help pays, and what it costs

Answered, Bought · finance director · losses loom larger

The operations review is six weeks on your floor, at a fixed price agreed before we start: A$45,000 to A$90,000, depending on the number of lines. If the review doesn't find more than it costs, you don't go ahead with the changes, and you don't pay for them. That's in the contract, word for word.

5. A plant like yours

Believed · finance director and owner · we follow others

Brennock Packaging, 240 staff. Output on the bottleneck carton line rose from 41 to 47 a minute, a 15% rise, without new equipment. Changeovers fell from 52 minutes to 19. Brennock's finance director confirmed the figures four months after the review.

• "We were three weeks from signing for a new line. We didn't need it." Plant manager, Brennock Packaging

[More cases with flat output →]

6. Against the alternatives

Preferred · COO and finance director · feeling first, reasons after

7. Questions

Answered · every chooser, and the assistant · a hard choice stalls

How long until output rises? Most plants see changeover times fall within the six weeks; output follows as the schedule catches up.

What do you need from our team? About two hours a day from the plant manager, and access to the floor on every shift. "The first six weeks" has the detail.

What if our problem is demand, not capacity? Then we'll say so on the first call, and you won't need a review.

8. One next step

Bought · COO · a hard choice stalls

Book a 30-minute fit call. We'll ask about your lines and your numbers, and tell you plainly whether a review would pay. You'll have a one-page scope within two days.

What to notice. Every claim on the page is in the claims register: the six weeks, the fixed price, the way out and the Brennock figures, confirmed by the client. Section 3 tells some buyers they don't need Kestrel, which is why the rest of the page is believed. And every section answers its question in its first sentence, so an assistant can lift any one of them whole.

Templates

Copy these. Every field is filled from something decided earlier, and the source is named beside it, so nobody fills a field from memory.

The page brief

PAGE [name] TYPE [from the anatomy
library]
PRIORITY [Hero / P1-P4 / Hygiene] WAVE [1 / 2 / 3]
SIGNALS [canon / coverage / strategy] EXISTING PAGE [preserve /
improve / merge / replace / new]
MOMENTS [each chooser · step · need, from the moment map]
THE ARC [from the first question to the one next step, in one line]
REASON [the reason to be chosen, word for word, from Brand Strategy]
CLAIMS USED [each one from the claims register, with its state]
TARGET QUERIES [five at most, in the buyer's words]
SECTIONS [in the anatomy library's order; required ones marked]
LINKS [parent, children, related; where the next step goes]
NAV LABEL [two to four words]
FRESHNESS [evergreen / refresh-critical / time-bound; what triggers a
refresh]

The section brief

SECTION [number and name]
JOB [stage · chooser and moment · bias]
HUMAN GOAL [what the reader should know or do after this section]
AI GOAL [the one passage an assistant should be able to lift]
THE QUESTION [answered in the first sentence]
CLAIM + PROOF [one claim, its proof beside it, both from the register]
ENTITIES [names, places, products, figures]
WORDS [target length]
GUARD [passes only if ...]

The page decision log (one row for every existing page)

The measures sheet (one row per page, read monthly)

Step 6: the business check

Before design starts, check the reconciled site against the business:

• Revenue: the review page, the problem pages and the manufacturing hub carry the moments behind most new work. If they don't, the moment list has drifted from the business. Check every strategy flag the same way: the investor page is P1 only because investor-owned manufacturers are a chosen target. If they aren't, it drops to P3.
• One reason: every page echoes "savings found on the floor in six weeks, at a fixed price". Pages that tell a different story get rewritten or cut.
• One truth: the price range, duration and terms are identical on the site, in proposals and on every listing.
• The offer: can the firm really honour "if it finds nothing, you don't proceed"? If not, it's an offer decision, not a copy decision.
• Capacity: 47 pages is a lot for a 60-person firm, which is why wave 1 is 18. Two P1 pages take time to earn: the benchmark report needs data from enough reviews, and the results page needs every figure sourced. Until they're ready, the guide and the case studies carry their moments. The order changes; the list doesn't.

Step 7: the measures

Each page is measured by the stage it carries, never by its traffic.

In the campaign plan

The Campaign Engine (step 10) turns the builds into a year of campaigns, each around a moment or an occasion. The website is in every one, because it supports every live cell. Its rule in the campaign plan is simple: a campaign lands on a page that already answers its moment. A campaign that needs a page the sitemap doesn't have has found a moment the map missed, so the page goes on the sitemap, or it's a time-bound page with a date it comes down.

Every campaign is measured on the page as well as in the channel: did the moment's stage move for the buyers it brought? An ad that brings clicks to a page that doesn't answer the question has bought traffic, not selection.

Running it

A site built from moments goes stale the way the moments do: a new rival, a new claim, a changed price. Running it means keeping the pages true to the map.

Who owns what. At Kestrel, the marketing lead owns the site and the sitemap. Each page family has an owner who knows the subject: the head of delivery owns the problem pages and "The first six weeks"; the managing partner owns price, terms and the way out. Nobody publishes a claim the claims register hasn't signed.

The cadence. The measures sheet monthly. The page decisions quarterly, with the Score. The whole sitemap once a year, or when the map changes by more than a tenth of its moments.

The basics, every release. A plain-text copy of every page for machines to read, plus llms.txt and sitemap.xml. Structured data only for questions the page answers. Pages that load fast on a phone on a plant floor; text at least 16 pixels; headings in order; every image described; every form usable by keyboard; contrast that passes WCAG 2.2 AA. No dead links, and every next step resolves to a real page.

Common mistakes

Each mistake breaks a stage, and most work against a bias buyers already have. Kestrel's old site made nearly all of them.

The bias to watch in the team: the curse of knowledge. Everyone at Kestrel knows what an operations review is, so nobody thought to explain it. Read each page as the plant manager who's never hired a consultant.

The Loop for this build

Each quarter, the Loop (step 11) puts the site's slice of the Score beside what was predicted, and asks one question per page family: did the stage this page carries move, for the moments it was built for?

What Kestrel's first two loops would look for. After wave 1, Answered at the problem, options and comparing moments should rise first, from 9 towards half, because the pages now exist. Found should move a quarter or two later, once assistants and search have read them. Bought should rise wherever price and the way out now sit side by side. If Answered rises and the Score barely moves, that's expected in the first quarter: the gate opens slowly. If Answered doesn't rise, the pages aren't answering, and no amount of promotion will fix that.

The rules behind the method

Shape, coverage and reconcile rest on fifteen rules. They hold whether a team works in a spreadsheet or in software.

1. Every existing page gets a decision: preserve, improve, merge, replace or deprecate.

• Deprecating needs a written reason. Pages that already earn their keep are kept.

2. Competitor canon: a page type on at least three of the three to five leading

• competitors' sites is expected in the category.

3. Coverage cells hold more than questions: each chooser-by-step cell records the pain,

• the gain, the job to be done and the objection.

4. Priority by signals: canon, coverage and strategy. No page exists without a reason, and

• rejected pages stay visible.

5. Priority sets the build order, never which pages exist.

6. Page families: one template applied to a set of entities (each problem, sector, role,

• case), not hand-built pages.

7. Freshness: every page type is marked evergreen, refresh-critical or time-bound, with

• what triggers a refresh.

8. Sitemap triage with a coverage check: no orphaned moments, every chooser served

• end to end, and a warning before cutting the only page that covers a moment.

9. Page brief fields: type, positioning wedge, target queries (five at most), section list,

• hero flag and navigation label.

10. Section brief fields: the job (stage, chooser and bias), human goal, AI goal, entities,

• word target, and guard criteria.

11. Canonical page shapes: each page type has a standard order of sections, from the

• anatomy library.

12. Fidelity by priority: P1 pages get full design; P2 to P4 get content in a template; hygiene

• pages are templated.

13. Machine legibility: named entities, at least three answerable questions and a quotable

• passage per page; structured data only for questions the page answers; a plain-text copy of every page, plus llms.txt and sitemap.xml.

14. No unsourced figures: every number sourced, signed off, removed or bound to live

• data before it's published.

15. Internal links built from the page list: breadcrumbs, parent, children and related

• pages; every next step resolves to a real page.

Shapes by business type

The playbook keeps a library of starting shapes. Each one is a default to reconcile, not a template to fill. Expect each to gain and lose pages once coverage is run.

The same flow for the other thirteen builds

Every build has a shape too. The method is the same everywhere: shape, coverage, reconcile, then brief, check and measure.

Feeling a bit overwhelmed?

You should be, a little. That was one firm with one main offer: eight choosers, seven problems, ten steps, five buying modes, 149 moments and 47 pages, each with a page brief and a brief for every section. Leave a moment out and a buyer gets their answer somewhere else. Few teams can hold all of that in a spreadsheet, let alone keep it current as buyers and competitors change.

That's why we built RocMode's Website Engine.

• Your buyer map, built for you. RocMode maps your choosers, their journeys and their moments from what's known about your brand, with the pain, gain, job and objection in every cell. Change anything you like; your changes are kept.
• Your website strategy and sitemap. From that map it builds the strategy and the sitemap, with a reason for every page and a check that every moment has a home.

Join the RocMode beta