Bloom does not have a marketing problem. The nights are good, the creative lands, the room fills. What Bloom has is a handoff problem — decisions get made and then stop travelling halfway.

Every item below is from the platform audits run between 2 and 3 August 2026. None of them is a failure of taste or effort. Each one is a decision that was made correctly and then didn't finish moving.

Everything marketing does runs through the same six stages, whether it's a standing Tuesday or a tentpole weekend. The stages don't change. The lead time does.

Two kinds of thing run through the line, and they run on different clocks. Confusing them is why tentpoles get announced late and standing nights get announced twice.

Anything that is not one of those two is an interrupt — a last-minute guest, a weather call, a viral moment worth chasing. Interrupts are allowed. They skip the calendar but they do not skip the gates, and the promise test in section 07 still applies.

Ten seats. Positions only — names go on a roster, not in this document. At Bloom's current size one person holds several of these; that is fine and expected. What is not fine is a seat nobody holds, because the work still exists and simply stops happening.

Ten surfaces face the guest. Each gets exactly one owning seat. "Everyone can post" is how three surfaces end up disagreeing about what Wednesday is called.

There is one document that says what is true — the confirmed brand fact sheet. Names, hours, prices, addresses, which night routes to which booking platform. Every surface reads from it. Nothing overrides it, including a flyer that already went to print.

The propagation order

When anything changes, it moves in this order and only this order. The order exists because the surfaces at the end are the ones that get forgotten, so they go on a list instead of being remembered.

Before anything goes public — a flyer, a reel, an email, a story, an ad — it passes eight checks. Any single failure holds the post. This is the gate between stage 2 and stage 4, and it is the whole point of this document.

One page, per night, issued to the floor before the week starts. This is the artifact that crosses the line from marketing to operations. Without it the floor finds out what was promised from the guest who shows up holding a screenshot.

Stage 6 is where most hospitality marketing quietly stops. The night happened, everyone is tired, and the only record is a camera roll nobody edits. Three things happen after every night, and they are small.

Within 12 hours

• Content off the phone. Whoever held the shot list dumps it to the shared folder before the next shift. Footage that stays on a personal phone is footage that does not exist.
• The review ask goes out. Post-visit SMS or email to anyone on the book, pointing at the Google review link. This is the single highest-yield retention action Bloom is not doing systematically.
• Story reposts. Tagged content and story mentions get reshared while the night is still live in people's feeds.

Within 48 hours

• The recap post. Proof the night was good, aimed at everyone who didn't come. This is stage 3 of the funnel and it is what makes next week's decision easier for a group chat.
• Reviews answered. Every one, good and bad, inside the playbook.

By Monday

• The numbers filed. Covers, sections sold, spend per head, comps, DMs, new followers, new list subscribers, reviews. Against the same night last month.
• One line on what to change. Not a report. One sentence that goes into the next Decide.

Eight funnel stages. Each one gets a seat and one number. If a stage has no number, nobody can tell whether it is working, and it quietly becomes optional.

The line only works if it has a clock. This is the proposed week — fixed days, fixed outputs, so nobody has to ask what happens next.

Note what Wednesday is. It is not a content day — it is the day somebody walks every surface and refuses to green-light nights that are not actually bookable. That single recurring hour is what stops section 01 from happening again.

For Memorial Day Weekend, Summer Kickoff, Father's Day, Juneteenth, the anniversary, and any booked headline. Days before the night.

Eleven thousand eight hundred and fifty-one contacts. Instagram's reply window is 24 hours — after that you cannot message a person back without them writing again. The inbox is not admin, it is the most direct conversion surface Bloom has, and it is the one with the hardest deadline.

The standard

Rules

• A named person per shift. Not "whoever sees it." The shift ends when the inbox is at zero unassigned.
• The bot handles the top five questions, a human handles the rest. Automation that tries to answer everything answers nothing well and produces the placeholder problem.
• Nothing routes to a link that does not work. Every automated reply that names a night or a booking path is on the propagation list in section 06.
• Non-marketing inquiries have a destination. Today this is unresolved in the responsibility record and it is why things sit.

The goal is not a busy Friday. The goal is a person who says "I go to Bloom on Thursdays" without thinking about it. That sentence is worth more than any single campaign, and it is built deliberately.

What makes a night a habit

• It never moves. Same night, same name, same offer, indefinitely. A night that changes identity every few months cannot become a habit.
• The promise is identical every week. The variable is the guest, the host or the theme — never the format.
• Somebody notices they came. Recognition at the door, a held table, a name remembered. This is a floor job that marketing depends on.
• They get one message between visits. One. Timed to the night they already come to.

The segments worth building

Three systems hold pieces of the same person. Joined, they produce a list worth messaging.

Marketing decisions stall not because they are hard but because nobody knows who is allowed to make them. This is the shortest possible version.

Straight from the responsibility record and the platform audits. These are not proposals — they are jobs that exist, are being done badly or not at all, and have nobody's name against them.

This is a draft, and the people who know whether it's accurate are the people doing the work. If a seat doesn't match what you actually do, if a lead time is unrealistic, or if something you handle every week isn't on any page — say so here.