Bloom Year One
Fifteen months of trading, 106 dated events and 8,995 reservations, matched night by night against what an ordinary night of the same kind actually does.
The year in numbers
Every figure on this page comes from Toast, Tock and the venue’s own Instagram archive. Nothing is modelled or estimated except where it says so.
Each event is measured against the median of the same weekday within six weeks either side, excluding the event date itself and any non-trading day. Bloom’s monthly volume swings by more than 40%, so comparing an August Thursday to a February Thursday measures the season, not the event.
Where the money came from
Vice is separated out here and everywhere else on this page, so no Vice sale is ever counted as a Bloom sale.
Vice did $3,459 on 90 guests in 2025 — it effectively did not trade. It did $108,592 in 2026 to date, 4.9% of the building; the entire ramp is this year. Vice’s check average is $47.84 against the Dining Room’s $63.46.
One thing to watch: on 15 September, Vice was 35 of 81 guest checks — 43% of the night by count while being small in dollars. On a quiet night Vice is a large share of the room, so any per-night read that ignores it will misattribute where the traffic is.
The shape of the year
Weekly net sales across the whole period. Use the arrow keys or hover to read any week.
The three biggest weeks of the year are the three biggest tentpoles. The week of 3 May took $106,474 — Cinco de Mayo, Nurses Week and After the Bell inside five days. The week of 23 November took $106,298 around Bloomsgiving Eve, and the week of 28 December took $103,689 around New Year’s Eve. Halloween week is nowhere near them.
The other feature is the collapse that begins on Saturday 10 January, shaded here and dealt with directly in section 08. Weeks run Sunday to Saturday; the partial week ending 17 September is excluded.
What actually moves a day
64.8% of the variation in daily sales is which day of the week it is. Add which month, and it is 85.1%. Everything a marketing calendar can influence competes for the remaining 14.9%.
That is not an argument against programming. It is the scale against which programming should be judged — and it is why every number on this page is measured against a matched weekday rather than against last week.
How each night has moved
Median net sales per night, opening months against the last twelve weeks.
Sunday has fallen from a $10,588 median to $3,671 — down 65%. Saturday is down 27% and Friday 25%. Monday is the only night that has grown, and it grew because a format was put on it.
Which formats work
This is the finding that matters most, and it only became visible once every event was sorted by what it was rather than what night it fell on.
With the archive complete, 97 of 106 events were promoted. Promotion is table stakes, not the variable. What separates a good night from a bad one is which of these nine things it is.
A holiday that sends people out is +194% at the median, six for six. A one-off themed Ladies Night is −34% at the median, zero for seven. Those are not close calls, and both are decisions the calendar makes every month.
The standing offers
Events are one-offs. The weekly food and drink offers run every week whether anyone is watching, and they are the other half of the year-two decision.
Each one is judged two ways: whether its night gained ground against the rest of the venue while the offer ran — which strips out the venue-wide decline — and whether the item itself makes money.
What the offers did to the average item
The share of everything sold that is a $1–$3 promo item, against the average price of an item sold.
In September, 35.7% of every item Bloom sold was a $1–$3 promo item — and those items produced 6.2% of net sales. The average item sold has fallen from $16.61 to $10.05, a 40% drop that tracks the promo rollout almost exactly.
More than a third of the kitchen’s and bar’s output, a third of the tickets, a third of the labour — for one dollar in sixteen.
What they did to the full-price versions
Full-price lamb chops fell from $302 a day to $83, and from 3.00% of an average trading day to 1.49% — so this is not simply the venue getting smaller around it. Full-price wings fell from 40.3 a day to 17.2 in five months.
Cannibalisation is shown as a trend rather than a dollar loss on purpose: some of those guests would never have bought at full price, and that share cannot be measured from the POS.
Every marketing promotion in the building comes to $22,539, or 9.4% of all discounting. Manager and owner comps are 48.8%. The $3 lamb chop is a margin problem worth fixing; it is not where the money is going.
Every event
All 106 dated events, each against its matched baseline. Sort any column; filter by night or result.
| Date | Event | Net sales | Baseline | Result | Covers |
|---|
Promoted — a flyer or post exists Planned only Not promoted — something was on and Bloom promoted nothing
Halloween week, and the cost of stacking
Halloween 2025 was not an unprogrammed night, which is what an earlier draft of this analysis assumed. It was a seven-night themed series with six flyers and a week-at-a-glance card.
| Date | Event | Net sales | Baseline | Result |
|---|
The only two that held were the two bolted onto formats that already existed — the Monday happy hour and Live Music Tuesday. All four brand-new concepts lost. By Thursday the week had run out of people: Bunny Mansion stopped selling at 11PM with $431 in the late block.
The Saturday is the sharpest lesson. Nightmare in Bloom was the one night of the week carrying a cover charge for men, and it took $14,195 after 10PM against $21,313 on an ordinary Saturday a week earlier. The late block lost $7,118 — 69% of the entire week’s shortfall.
Halloween 2026 falls on a Saturday. One night, one concept, and a dinner reason: Halloween night itself lost dinner, not the party — the evening block fell 53% while late trade held within 19%.
January
The meeting notes record January as a marketing miss — “MLK weekend missed.” The data does not support that, and the correction matters because it puts effort against the wrong problem.
A woman was shot and killed in Bloom’s parking lot, after hours, early on Saturday 10 January 2026. Friday 9 January had traded a full, strong night — $22,953, up 55%, last sale at 2AM. Every night from Saturday 10 January onward falls away at once.
Saturday 10 January had 237 covers on the book — an ordinary-to-strong Saturday — and took $11,839, or $50 a cover against a venue average of $108. Those reservations were made before that morning. The daytime traded normally; the late-night peak simply never arrived, at $4,623 after 10PM against roughly $20,000 on a normal Saturday.
MLK weekend is a week later. MLK Saturday was in fact the best of the four Saturdays in that stretch. Weekly sales went $104,570 → $57,709 → $55,562 → $39,883 at the trough, then took about six weeks to reach a stable level below where December had run. Weeknights were hit hardest: Tuesdays and Wednesdays ran −62% to −77% for a fortnight.
Thursday 15 January — Alpha Kappa Alpha, Eta Iota Omega Chapter, Founders’ Day. Members only 7–8PM, public social 8PM–midnight. It did $13,471, up 163%, five days after the shooting, in the emptiest fortnight of the year — on fourteen covers on the Tock book. The people who already had a reason to be there still came.
The two record Saturdays
21 and 28 March 2026 are the two biggest days in Bloom’s history, eight days apart, and they happened for completely different reasons. An earlier version of this analysis credited both to the NCAA tournament. That claim is withdrawn — nothing was promoted on either date, and the hour-by-hour data contradicts it.
| Saturday | Net sales | Covers | vs normal Sat | Avg party | $ per cover |
|---|---|---|---|---|---|
| 7 Marcontrol | $29,653 | 219 | −1% | 5.5 | $135 |
| 14 Marcontrol | $28,872 | 194 | −13% | 5.9 | $149 |
| 21 Marrecord | $41,933 | 392 | +76% | 12.2 | $107 |
| 28 Marrecord | $41,331 | 227 | +1% | 6.5 | $182 |
| 25 Aprpromoted Pistons brunch | $24,956 | 186 | −17% | 6.9 | $134 |
The day carried one reservation for 200 people at 6:00PM. That single booking is 51% of the day’s covers and the largest party in the file by a distance. The 6PM hour took $6,200 against $861 and $682 on the two control Saturdays.
It booked an entirely ordinary 227 covers and took $41,331 anyway — $182 a cover, with $12,742 of it before 6PM. Bloom’s own promoted Saturday daytime attempt, the Pistons playoff brunch, produced $1,427 before 6PM. This was walk-in spend at double the normal rate, and nothing explains it.
When the money arrives
Twenty-nine dates pulled hour by hour from Toast. Day is open to 5:59PM, evening 6–9:59PM, late 10PM to close. The best nights win two blocks, and they do it by running two separate programmes rather than one long one.
The shapes that win
- Both blocks — Bloomsgiving Eve ($9,017 evening + $12,748 late) and Home for the Holidays ($5,969 + $15,876). The two highest-lift nights of the year, and both ran a dinner attraction until 11 and a DJ after it.
- Late only — New Year’s Eve, 87% after 10PM with $9,711 in the 11PM hour alone.
- Dinner led — Cinco de Mayo, $10,564 in the 6–10PM block, the largest evening figure in the set.
- Daytime only — the Lions Home Opener took every dollar between noon and 5PM and nothing after.
And the shape that loses
Mics and Martinis sold nothing after 9PM. Bunny Mansion stopped at 11PM with $431 late. The Super Bowl closed at 9PM. An event that closes before the late block on a nightlife night has not underperformed — it has skipped the shift that pays.
The books, the discounts and the platforms
Reservations
$107.80 of net sales per booked cover, with covers and sales correlating at 0.844, is now a forecast anyone can apply on the morning of service. A night with 100 covers on the book should take about $10,800 — and where it does not, the gap is itself the finding.
Organisation nights book almost nothing and sell enormously. AKA Founders’ Day: 14 covers, $13,471, $962 a cover. Founders Day with the Ques: 25 covers, $10,486, $419 a cover. Judged on the Tock book the night before, every one of them looked like a failure — and those are precisely the nights that most need bar and security cover.
The reverse also shows. Valentine’s Day booked 282 covers, the second-highest of the year, and lost 22% — a dining room of two-tops crowding out the biggest nightlife night of the week. The offer-led nights do the same: the Galentine’s oyster flight drew +571% covers for +3% sales.
Where the discounting actually goes
$238,757 given away across 7,934 uses. The common assumption is that the promotions are the problem.
Marketing promotions are 9.4% of all discounting. Manager and owner comps are 48.8% — $116,483 — and open free-form discounts another 28.2%. The $3 lamb chop is not where the margin goes.
The platform stack
| Source | State | What it holds |
|---|---|---|
| Toast | Complete | 473 days of sales, 29 dates hour by hour |
| Instagram archive | Complete | 106 dated events, Jun 2025 – Sep 2026 |
| Tock | Complete | 8,995 reservations across 392 dates |
| Tablelist Pro | $0.00 | 43 events, none before 17 Jul 2026, no bookings and no revenue in fifteen months |
| Instagram Stories | Unavailable | Would not paginate — story-only promotions are missing from the record |
Live, public, configured, carrying artwork — and $0.00 revenue, zero reservations and zero admissions across all 43 events. It holds nothing at all before 17 July 2026. August was named properly (Teachers Mixer, the 2 Goated takeover, DJ Boof’s residency); September reverted to bare weekly shells, so the two worst promoted nights of the month are listed as if nothing was on.
Separately: the 1 Year Anniversary sold Yung Joc at $20 early-bird tickets and appears in Tablelist as “Bloom 1 Year Anniversary Weekend” with $0.00. Wherever that ticketing ran, there is no record of what it sold.
What year two does differently
Nine decisions, each one carried by a number above rather than an opinion.
