PerryGolf
ArchitectureMarginWise site analysis · July 2026
links are shareable
Working session · Aug 26–27, 2026 · concluded

Working-session agenda — outcomes

The workshop was held Aug 26–27, 2026. Each block below keeps the original agenda and the target outcomes it went in with, and now carries what the room actually decided and what stayed open. The detailed record — every decision, parked item and owner, with links into the Fireflies recording — is in the engagement session notes (engagement/context/sessions/2026-08-26… and …2026-08-27…).

Note: the Day-1 recording ended mid-cruise, so Small group tour and What’s missing: Trip Finder & supplier pages did not run as distinct blocks — their topics were absorbed into PYOT and into Day 2 (supplier pages were retired).

How the session ran
  • Each block is time-boxed.
  • Every block ends with named outcomes.
  • The point of the two days is a written record: decisions documented, open questions captured.
  • Anything we can't resolve on the day goes on the parking lot. Fireflies is recording, so nothing said out loud gets lost. Open questions should be captured in writing and shared with Davis and Ryan.
  • Most of the day is hands-on in the CMS.
Read first:How the data worksA plain-English tour of how suppliers, tours, cruises and destinations link together. Skim it before the hands-on blocks so we’re building, not explaining.
Day 1Wednesday, Aug 26· Build the core pages, hands-on9:00 – 4:30 · 1-hour lunch
9:00 – 9:15

Kickoff & how we'll work

Ownership & launch
  • The one goal the site serves: enquiries from qualified buyers.
  • Website objectives (1. CMS, 2. Speed and Performance, 3. Design)
  • How we run the day — time-boxes, one owner on the parking lot, outcomes out of every block.
What we hoped to leave with
  • A named owner for the parking lot.
  • Agreement that design tweaks get parked, not debated in the build blocks.
What we decided
  • The site's one job, reaffirmed: enquiries from qualified buyers — destination pages are where it under-converts today.
  • Objectives ordering held: (1) CMS, (2) speed & performance, (3) design. Sanity confirmed as the CMS.
  • Format ran as agreed — time-boxed blocks, hands-on in the CMS, outcomes out of each block.
Still open / carried forward
  • The 'park design tweaks' rule held loosely — the room repeatedly dipped into design; a full design block ran on Day 2.
9:15 – 10:15

Destination page

Pages & CMSHands-on · in Studio
Pre-readData ModelDesign
  • Build a destination page: draft the first version from a ChatGPT skill (using AI to speed the first pass), then edit it by hand in the Studio.
  • Why the order of the page matters — what a visitor to a destination needs to see, and when.
  • What a destination page must always show, and what is optional.
  • Do we highlight the golf courses and the accommodations (suppliers) on it, and how.
What we hoped to leave with
  • An understanding of the sections a destination page must have
  • Yes/no on featuring courses and accommodations — and if yes, curated or automatic or additional filtering
  • One destination page built in the Studio we can keep as the template.
What we decided
  • Full rework — depth is paramount (for SEO/AI-GEO and customer experience). Tiered: ~5–7 deep country pages; everything else a higher-level region page. Keep both country and region pages.
  • Section set agreed: exec-summary strip, ways-to-travel (Learn more + View trips), featured courses + accommodations (curated, behind a modal), testimonials (destination + tour-type), best-time-to-travel, non-golfer, FAQ, brand/USP module, enquiry form.
  • Featuring is curated, not automatic (automatic-with-rules is for Trip Finder). Detail opens in a modal, never a new page.
  • Maps only where sub-regions matter — Ireland and Scotland. Dropped: 'Unique Access' (it's a brand USP, not destination content).
Still open / carried forward
  • Section order per page — PerryGolf drafts the template; MarginWise builds the country/region × component matrix.
  • First-visit personalisation and the interactive map — Phase 2 / later.
  • Product×destination pages ('cruises in Spain') — wanted, to be separate landing pages; not built.
10:15 – 10:30
Break
10:30 – 11:45

PYOT (Price Your Own Tour)

PYOT
Pre-readData Model
  • What do we want PYOT to be? Today it's 31 hand-built itinerary pages that bring in enquiries. The bigger version lets a customer build their own tour from our destinations, courses and accommodations: pick a place, the golf and where to stay, and it becomes a priced enquiry.
  • Those 31 pages already pull about 250 “Price this Tour” enquiries a quarter from roughly 240 organic clicks. High value per page, almost no search visibility.
  • A PYOT is really a destination plus chosen courses and hotels — the pieces we just built. Do we reuse them for build-your-own, or keep it a separate PGGO thing?
  • For now, structure and pricing stay in PGGO. Marketing content can move to the CMS through a one-way import that only creates a draft and never overwrites what's live. What's the smallest first step that doesn't mean rebuilding the quoting engine?
  • Two calls to make once the ambition is set: show a price or not (per-supplier, or a single figure), and whether to run paid search.
  • Name what keeps PYOT content from staying current today — the organizational blockers, honestly.
What we hoped to leave with
  • The ambition, decided: is PYOT a lead-gen itinerary page, or a self-service build-your-own-tour product — and is that launch, phase 2, or a direction we're rejecting.
  • How PYOT relates to the destination page just built — reuse the destination / course / accommodation model, or stay separate.
  • The smallest first step named: what content moves to the CMS now, what stays in PGGO.
  • The tactical calls resolved with owners: price on the page, paid search.
  • The blockers stopping PYOT content from staying current, each with an owner.
What we decided
  • PYOT is the custom-tour product page — not a separate product. It reuses the destination / course / accommodation model.
  • Sanity = marketing source of truth for the website; the legacy console is retired. Price-on-page is a Sanity marketing attribute. No sync to PGGO is built for now — Sanity and PGGO stay separate (any future sync would be one-way only).
  • Ground transportation re-merchandised: a self-drive 'from' price, then the VIP concierge driver as the recommended upsell (modal + reviews).
  • Self-service build-your-own = Phase 2, not rejected (history: ~85/15 preset-vs-build). Universal rule: Day 1 = arrival day. Golf-course carousel → list + modal.
Still open / carried forward
  • Move PYOT / contact-us emails into the marketing system (engagement tracking) — Phase 2.
  • Nav merchandising of top tours — not changing yet; a hero-over-results belongs on a separate landing page, not in Trip Finder.
  • Couples / non-golfer sample itineraries — Phase 2 (structure supports it).
11:45 – 12:45
Lunch
12:45 – 2:00

Cruise

Tours & cruisesHands-on · in Studio
Pre-readData Model
  • Build a cruise: draft, then modify by hand.
  • Walk the structure — how a cruise differs from a tour (ship, cabins, ports).
  • The lifecycle of a cruise, start to finish — and the website updates it triggers: destination page, navbar, and the rest.
  • Waitlist on a cruise, and photos after it runs.
  • Given cruise growth, should we think about dedicated cruise destination pages?
What we hoped to leave with
  • A cruise lifecycle, with the points where it differs from a tour called out.
  • The waitlist rule for cruises.
  • One cruise built in the Studio as the reference.
What we decided
  • Cruises auto-generate from a supplier spreadsheet: golf-course & hotel names matched to Sanity supplier records (lat/long pulled from them), AI drafts highlights & brand copy.
  • Configurable navbar — group by vendor / year / location, with a waitlist sub-filter.
  • Ship onboard/dining tiles are Sanity-owned (hand-refreshed from operator sites); Silversea cabins/pricing sync via API, other vendors static.
  • No auto-publish — human review (with auto SEO meta + JSON-LD) before a cruise goes live.
Still open / carried forward
  • Waitlist display rule — show as-is / hide / show-differently — raised, not decided.
  • Gallery/tile layout, image sizing, ship-tile column count — not finalised.
  • Azamara import and its nuances — dedicated session; not shown. Dedicated cruise destination pages — raised, unresolved (recording ended mid-cruise).
2:00 – 2:15
Break
2:15 – 3:30

Small group tour

Tours & cruisesHands-on · in Studio
Pre-readData Model
  • Build a small group tour the same way: draft using AI, then modify by hand.
  • Walk the structure of the tour page section by section.
  • The lifecycle of a small group tour — from creating it, waitlisting it, retiring it, and every place it appears on the site in between.
  • Photos after the tour runs — where they go and who adds them.
  • Roles and the review gate: who drafts, who approves, who publishes — and confirm a draft never auto-publishes.
  • Time the build start to finish — a benchmark for how fast publishing a tour should be.
  • Discuss the business value of self-service purchasing for small group tours.
What we hoped to leave with
  • A one-page lifecycle: the stages a tour moves through, draft to retired.
  • The rule for waitlisted / nearly-departed departures — what shows, what's hidden.
  • The checklist of places that update when a tour launches (nav, destination page, featured lists).
  • The named publish workflow — drafts → approve → publish — and a time-to-publish benchmark.
  • A decision on post-tour photos: who adds them, and where.
What we decided
  • Did NOT run as a distinct block — the Day-1 recording ended mid-cruise. Lifecycle, waitlist and the time-to-publish benchmark were not decided here.
  • No auto-publish was affirmed generally (on cruise) and carries to tours: AI/spreadsheet drafts get a human review before going live.
  • Post-tour photos WERE addressed on Day 2: remove the website gallery; deliver via a third-party tool; build an automated ingest scoped to Azamara cruises + SGT host photos.
Still open / carried forward
  • SGT lifecycle (draft→retired), waitlisted / nearly-departed display, and the publish workflow (draft→approve→publish) — still to be worked.
  • Self-service purchasing for SGT — subsumed by the build-your-own decision (Phase 2; see PYOT).
3:30 – 4:30

What's missing: Trip Finder & supplier pages

Data & suppliers
Pre-readSearch Performance
  • By now we've built a destination, a PYOT, a tour and a cruise. This closing block is where we decide what's still missing — every item should end in a decision, not a status update.
  • Name the pages that don't exist yet. Cruises in Spain, for example, currently falls through to the Trip Finder.
  • Decide how the Trip Finder should index. Today every permutation is separately searchable, and it probably shouldn't be.
  • Should itinerary pages be independently indexable and linked from their destination page, or is that out of scope? Heritage Golf Travel is a good content-rich comparison.
  • Which supplier pages do we keep at launch, and which retire? It's its own conversation, separate from the Day-2 supplier-data reconciliation.
What we hoped to leave with
  • A named "what's missing" page list.
  • The Trip Finder indexing rule, decided.
  • Itinerary-page indexability in / out of scope, with an owner.
  • A supplier-pages keep / retire direction.
What we decided
  • Did NOT run as a distinct block — its topics were absorbed elsewhere.
  • Trip Finder (covered inside PYOT): stays THE browsing / listing hub (curated + automatic + filtered); 'see all' deep-links in with filters.
  • Supplier-pages keep/retire (decided on Day 2): RETIRE ~1,500–2,000, 301-redirect, rebuild the sitemap; surface courses on destination pages instead.
Still open / carried forward
  • 'Cruises in Spain'-type product×destination pages — wanted, to be separate landing pages (not Trip Finder permutations); not built.
  • Trip Finder indexing rule (every permutation separately searchable) and itinerary-page indexability — not explicitly decided.
Day 2Thursday, Aug 27· Supplier data, images, documentation & design9:00 – 1:00 · hard stop for travel
9:00 – 10:15

Supplier data

Data & suppliers
Pre-readData Model
  • Reconcile PGGO and Console — where the same venue reads differently between the two systems, and which system is the source of truth for each field.
  • Multi-course venues (e.g. Gleneagles) — one property with several courses: how that is modelled and shown.
  • What a destination / tour / cruise page actually pulls from a supplier record, and which fields are mandatory vs optional.
  • Where supplier data is edited going forward — one home, not the same venue kept in three places that drift apart.
  • Description drift — what needs a manual pass before launch, named record by record.
What we hoped to leave with
  • The source-of-truth rule, field by field: what's authoritative in PGGO, in Console, in the CMS.
  • A decision on how multi-course venues are handled.
  • Where supplier data is edited going forward, and who owns it.
  • A named, owned list of supplier records to clean up before launch.
What we decided
  • Field-level split: Sanity owns marketing copy (brief + long description, short ≤25 / long ≤70 words); PGGO owns operational.
  • Decided: NO sync for now. Sanity and PGGO stay separate and are allowed to drift — a deliberate 'Chinese wall' (only the PYOT base-quote id + form submissions cross). Any future sync would be one-way Sanity→PGGO only, never bidirectional.
  • The console is retired, consumed into Sanity — but once an MGC quote is created it becomes the master for that quote. Multi-course venues = thin master record + sub-suppliers; Fancourt = SPLIT (Manor House distinct).
  • An AI cleanup/audit skill standardises descriptions (human-in-the-loop). PYOT itineraries stay in Sanity, linked to their PGGO base-quote id (~28); cross-reference + lock.
Still open / carried forward
  • The 45-group merge worklist — the model is decided, but per-record MERGE/SEPARATE/SPLIT calls are deferred to a data session (SMEs Gordon/Cameron/GP).
  • Whether a formal 'Phase 2' deep operational sync ever happens.
  • Name-collision suppliers (course vs hotel sharing a name, e.g. Dundonald Links) — disambiguation UX.
10:15 – 10:30
Break
10:30 – 11:25

Images & licensing

Data & suppliers
Pre-readData Model
  • The single home for images: everything staged in Sanity's asset library (or one agreed source), not scattered across PGGO, drives and email threads.
  • What we're actually allowed to use. Every image staged with its rights — owned, supplier-provided, licensed, or off-limits. Nothing goes live without knowing where it came from.
  • Post-trip photos — the flow for getting them onto the right pages after a tour or cruise runs, and who uploads them.
  • Generic vs specific supplier images — where a wrong or generic photo (e.g. the coach image) has to be replaced.
  • Tagging and reuse — enough metadata that the same image is found and reused, not re-uploaded three times.
What we hoped to leave with
  • The single source of truth for images, decided — Sanity asset library or a named alternative.
  • A licensing rule: every image carries its source and usage rights before it's used.
  • The post-trip photo workflow, start to finish, with an owner.
  • The must-fix / must-replace image list before launch.
What we decided
  • Sanity is the image source of truth; its CDN handles compress/resize/focal-crop; the importer takes the highest-resolution originals.
  • Rights system: a dropdown (owned / licensed / open-source / unknown) + note + date + attribution (shown on hover). No historical backfill.
  • Net-new images need licensed status or an approved 'unknown', signed off by a named leader (Alex or Brett).
  • Remove the website photo gallery; deliver post-trip photos via a third-party tool; automated ingest (Azamara + SGT) runs in parallel; MarginWise runs a ~2h feasibility test.
Still open / carried forward
  • Shutterstock audit (expiry; is AI-enhancement permitted?), Sanity paid Media Library need, and Sanity–Dropbox attribution import — to research.
  • Exact licensing dropdown values, and the guest-photo delivery vendor.
  • Silversea image permission re-audit (permission since resolved via a display agreement; attribution still open).
11:25 – 11:45

Documentation, ownership & launch

Ownership & launch
Pre-readDelivery Plan
  • The parking-lot readout: every open question from the two days in one place, each with who answers it and by when.
  • Everything decided gets written down — SOPs anyone can follow, not knowledge stuck in one person's head.
  • Who owns content once we launch — editing, images, approvals.
  • The launch plan: the mid-September MVP target and the milestones to it.
  • Check the “PerryGolf inputs needed” list on the Delivery Plan — the schedule that came out of the site analysis. Most of it is already in hand (PGGO Dev and Search Console access granted; in-flight product list and JonnieO requirements received); the one open item is the pages-to-retire call — come to the room ready to make it.
  • The response windows everyone is signing up to, from the contract: system/data access within 3 business days, sign-offs within 5, content decisions within 7 — a missed window slips the Day 120 (Dec 1) go-live day-for-day, and Scope Lock (Day 45 · Sep 17) closes new scope.
What we hoped to leave with
  • One owned parking-lot list of open questions, each assigned and dated.
  • Named owners for content — editing, images, approvals.
  • Confirmed launch milestones to the mid-September MVP.
  • The pages-to-retire call made in the room, and the rest of the Delivery Plan's PerryGolf inputs confirmed in hand — so the Day 120 go-live and Scope Lock hold.
What we decided
  • Retire ~1,500–2,000 supplier pages, 301-redirect, rebuild the sitemap; page-inventory workbook goes to PerryGolf (Harrison) to confirm keep/retire.
  • Promote The Open + St Andrews into the nav; About Us / brochure move to the footer (after a content refresh).
  • Delivery by rolling module sign-off via the twice-weekly steering call; the bulk targeted for end of September.
  • 'MGC retirement' means the name only (its role persists); the outdated Azamara ship-info page is kept as-is. Content owners: Stacy, Harrison (+ the room) for Phase 1.
Still open / carried forward
  • Mid-October plans — the standing timeline unknown after the end-September push.
  • A formal content-ownership RACI — high-level only for now.
  • SOPs for supplier-data ownership going forward — to write.
11:45 – 1:00

Design discussion

Design
Pre-readDesign
  • Work through the design items raised across both days — nav/dropdown, grid layouts, mobile, and anything parked.
  • Revisit the held convictions — noticeable CTAs, incentivise the enquiry form, remove carousels, higher-quality images — do they still hold?
  • Rank it honestly: what changes for launch, what waits.
What we hoped to leave with
  • A ranked design list, split into launch vs later.
  • The design convictions confirmed or revised.
What we decided
  • Golf courses & lodging: a card + 'View Details' modal — one course image at a time, arrows advance to the next course; drop the repeated per-course dates (redundant with the itinerary).
  • Gallery becomes a mosaic below the highlights; reject two-column highlights; bullets over check-marks, applied consistently.
  • Keep the yellow Trip Finder buttons (clarity over aesthetics); all-caps is CMS-configurable. Navbar keeps its large font + blue boxes; test it WITHOUT the nav image.
  • The held convictions confirmed — noticeable CTAs, incentivise the enquiry, remove carousels, higher-quality images. CMS is Lego blocks on a locked brand design system (no free-draw).
Still open / carried forward
  • MarginWise to mock 4–5 golf-course display options + the mosaic gallery for reaction.
  • Mobile course-card interaction (stack vs swipe); course content appearing in both the modal and the course section (SEO duplication) — re-evaluate against the mockups.
Second record · Ryan’s written notes · partial capture

Workshop notes — Ryan’s capture

Everything above is digested from the two Fireflies transcripts. This section is a separate, independent record — Ryan’s handwritten notes of the same two days, transcribed unchanged. Items carry stable ids (D decision, A action, P parking lot, F Phase 2) so they can be cited without quoting. The two records are compared in the section that follows; neither is edited to suit the other.

⚠️ These notes carry no day attribution, and none should be inferred (Ryan, 2026-08-28): they span both days, and any day reference in the original was incidental. Items are listed by kind, not by day. Where a day is named on this page it refers to the transcript record above, which is day-structured because the recordings are.

A partial capture — Ryan’s words were “here are some of the actions, parking lot, decisions and other stuff”. The absence of an item means nothing.

Recorded as decided

Decisions

No owner column — a decision needs an implementer only once it becomes an action, and several have not been turned into one yet.

  • D1Destination nav shows the countries that matter, then groups the rest into regions that matter.
  • D2On a destination page: yes to featured courses, yes to featured accommodations, yes to "when to play".
    Ryan’s annotation: "likely these decisions recorded clearly in transcript".
  • D3The "Unique Access" section is eliminated across the site.
  • D4aSuite selection table — remove suite location on ship.
  • D4bSuite selection table — put suite and veranda sq ft range in the suite details row.
  • D4cSuite selection table — tighten horizontal spacing.
  • D5Customise Mapbox to Perry colours.
  • D6Add image attributions where we have them, and audit where Perry actually holds permission — Dropbox and Shutterstock.
    The notes carry a "(?)" against Shutterstock; the second source is unconfirmed.
  • D7A short description should appear everywhere a golf course is shown.
  • D8Source of truth: Sanity owns any content on the website. PG Go retains source of truth for operationally related data.
Owned work

Actions

All on MarginWise unless noted. The notes use "MarginWise" and "MW" interchangeably; normalised to one owner.

  • A1Build a matrix of all destinations × all destination-page sections, marking which sections belong on each destination.
    MarginWise
    Guidance: Scotland and Ireland get all sections; others only where there is appropriate content depth and value.
  • A2Identify and develop content/data review sets for reconciliation and finalisation, and develop an AI data-audit skill.
    MarginWise
  • A2aGive Perry review opportunities framed as "do nothing and this will happen", to minimise project blocking.
    MarginWise
  • A2bProvide AI analysis alongside Davis/Ryan recommendations.
    MarginWise
  • A3Mock 4–5 golf-course display options + the mosaic gallery.done
    MarginWise (Davis/Ryan)
    Confirmed by Ryan 2026-08-28 — the notes recalled “2–3 variations”; the transcript gives 4–5 options and the subject. See the comparison below.
  • A4Revert Silversea images to the ones in the console — the Silversea CDN images are not authorised.
    MarginWise
  • A5Add suite selection to the cruise pages.
    MarginWise
  • A6Update the navbar: year columns, left-side blue shading, buttons for brochures.open
    MarginWise
    The note says "see attached whiteboard photo"; the photo did not come with the notes. Which side is shaded is open — see the comparison below.
  • A7Produce a list of courses with no short description.
    MarginWise
    Ryan’s note: "likely already done with Sanity folder" — check before building. Pairs with D7.
  • A9Revisit the multi-course venue model — how to better group courses and hotels that are part of the same complex.
    MarginWise (Ryan)
    Sat indented under the Trip Finder list in the source notes; promoted here as it is not a Trip Finder item.
  • A10Add permission tracking to image metadata: a flag for "is Perry authorised to use this image", permission status, permission notes.
    MarginWise
    Purpose is traceability back to the written confirmation. Pairs with D6.
  • A10aGet access to the Dropbox and Shutterstock accounts to audit the images and their metadata.
    MarginWise
A8, in detail

Trip Finder changes

  • A8aRemove the ability to filter by ship.
  • A8bChange "cruise vendor" → "cruise partner".done
  • A8cChange "nights" → "trip length".done
    Incomplete by Ryan’s own note: "nights still appears in filter selection and elsewhere".
  • A8dWhen custom tours are selected, make nights/rounds criteria unselectable.
  • A8eChange the date range to an explicit month list grouped by year — 2026: July–Dec; 2027: Jan, Feb, Mar, … and so on.
  • A8fWhen no criteria are selected, randomise the order trips appear in.open
    The notes record "a tier structure" alongside this and the context was lost. See the comparison below.
Open, and explicitly not decided

Parking lot

  • P1Did we decide completely on the included sections for the big destinations (Scotland, Ireland, UK/Isles)?
    Bears on A1 and D2 — the destination matrix may be blocked on this for the big three.
  • P2Order of the sections on the destination pages.
  • P3What does it mean — and where should it be done — to highlight non-golfing activities, to enable and encourage couples trips?
  • P4Mobile display of course/lodging details and photos.
    Davis looking into
  • P5Enhanced trip-builder functionality?
    Davis asked; Alex responded that they used to have it and it was deprioritised in the 2024 rebuild. Could be done again, but needs prioritising, examining for value, and/or checking whether analytics show people looking for it.
  • P6Should the emails sent after a PYOT enquiry carry the same information and text as the website?
  • P7Post-trip photo albums — need to be far more efficient.
    Use the wedding/event photographer approach: automate album creation and link distribution. Worry less about tagging people and photo locations.
  • P8The related trips/cruises section need not link to a trip or cruise — a card could link to a featured destination or a thematic grouping.
Later

Phase 2

  • F1Potentially differ content order, CTAs, etc. based on prior visits and the customer email journey.
  • F2Potentially add additional example PYOT itineraries, themed "perfect for couples", "boys trip", and so on.
    Relates to P3.
Third section · comparison

The two records compared

Both records above are left exactly as written. Nothing here edits either. Where the two overlap, prefer the transcript-fed record — it is timestamped, deep-linked and carries named owners. But it is not a superset, which is why both are kept.

Only in Ryan’s notes

The Day-1 recording ended mid-cruise. Four groups of items appear in no transcript note at all — for these, the written notes are the only record.

  • D4a–cThe whole suite-table detail — remove suite location, veranda sq ft range, tighten spacing.
  • A8a–fAll six Trip Finder changes, including two already done and one still ambiguous.
  • D2“When to play” on destination pages.
  • P8The related-trips card need not link to a trip or cruise.
Only in the transcript record

Effectively everything else. It is the far larger capture — two days with owners and Fireflies deep-links — and covers whole blocks the written notes do not touch: the source-of-truth split field by field, retiring the ~1,500–2,000 supplier/partner pages, the console sunset, the Phase-1 “Chinese wall”, Azamara onboarding, and SOPs for supplier-data ownership.

Read that record first. Ryan’s is a supplement to it, not an alternative.

Gap resolution

The three things Ryan’s notes flagged as lost

A3Confirmed — Ryan, 2026-08-28

The missing subject is golf-course display options, and the count is 4–5, not the “2–3” the notes recall. Davis’s Day 2: “mock 4–5 golf-course display options + the mosaic gallery; switch highlights to bullets; revisit arrow contrast” [3:00:13]. A3 should be read as that action.

A6Open — for Davis and Ryan to discuss

The two records disagree on which side is shaded. D25 says the blue boxes distinguish “functional right-side links from inspirational left-side text”. Ryan’s notes say “left-side blue shading”, and Ryan confirms his whiteboard photo of the drawings he and Alex did shows them discussing shading the LEFT side. So Ryan’s recollection and his notes agree with each other and not with D25. To be settled against the whiteboard photo.

A8fOpen — for Davis and Ryan to discuss

Davis’s Day-1 D6 confirms the randomisation half — “automatic-with-rules is for Trip Finder (curated + automatic + filtered, with some randomisation)” [1:07:44] — but does not explain the “tier structure” the notes record alongside it. The only tier language in either transcript note is D2, about tiered destination-page depth, a different subject. Either the tier note belongs with D2, or Trip Finder tiering was said off-recording.

Where the two agree

Worth stating, because agreement between an independent recollection and a transcript is stronger evidence than either alone.

  • Multi-course venue model — Ryan’s A9 (“revisit”) matches Davis’s own parked item at [1:03:23], alongside his D6 which sets the thin-master model (Fancourt = SPLIT). Decided and flagged for re-examination; both records say so.
  • Image rights — Ryan’s D6/A10/A10a line up with Davis’s image-licensing action and PerryGolf’s action to provide Shutterstock credentials.
  • “Unique Access” eliminated, Mapbox in Perry colours, short description wherever a course is shown, and the Sanity / PG Go source-of-truth split all appear in both.
Still not done

Neither record has been reconciled against the engagement’s open-questions.md, parked-work.md or pyot-open-items.md. The transcript capture did annotate open-questions and supplier-merge-review; Ryan’s capture has not been through that pass at all, and the overlaps are unmapped. Separate piece of work.

Markdown sources of record: engagement/context/workshop-outcomes.md (Ryan’s notes) and engagement/context/workshop-capture-comparison.md (this comparison). Change those first; this page mirrors them.

Post-workshop record. Each block shows the target outcomes it went in with, what the room decided, and what stayed open. Full detail with Fireflies deep-links: the engagement session notes for 2026-08-26 (Day 1) and 2026-08-27 (Day 2).