Uma Blu Coffee & Brunch · Atrium Complex, Kuta, Lombok, Indonesia. Food & beverage operations, one outlet trading, six in pipeline. Prepared for the shareholders of Blu Zone Group.
The month in the Director's words, followed by the numbers that back it up and the one place the story is more nuanced than the headline.
July was the best month Uma Blu has had. Rp 357.8M net, 2,474 covers, and the first month we have ever run under the 60% prime cost line. Two things drove it: more tourists in Kuta, and social media engagement that is finally translating into people walking through the door and ordering properly rather than just ordering a coffee.
June was a bad month and shareholders deserve the reason plainly. Kitchen service was unsatisfactory, and the person in the purchasing seat was not doing the job. Suppliers were not being paid every Monday as we had committed, which cost us the relationships we had been deliberately building. Both problems have been dealt with, one by removing a person, one by removing the role entirely and replacing it with a system. We are up to date with every supplier and rebuilding goodwill.
The Indonesian menu is launched. The fish curry needs a commercial microwave to reheat at service speed, particularly the bowl. That is a small piece of capex standing between us and a faster kitchen. The ABC alcohol licence is still sitting with Erick and we need it now; it is the single largest piece of revenue we are not yet earning. The display we put in for baked goods did lift bakery sales, and it is being retired in favour of three dome displays at the customer end of the bar. Chilled goods were also purchased, which was a mistake. That stock needed a refrigerated display. Nobody told me it was coming, and I would have stopped it. Most of it went in the bin because it was stored at room temperature. That is a control failure, not a bad luck event, and it is in the risk register this month for that reason.
On people: Devi is working so hard I have to send her home. I sit with her every Sunday morning to mentor her and to listen to where her week is at. Her development as a leader is gradual and real. Her one risk is how deeply she involves herself in everybody's personal lives. If that does not stop, it will be her undoing, and she knows it. Dede was given his marching orders at contract renewal; he was not the standard Atrium should be known for. Harianto has joined for three months to educate, mentor and organise the cafe, and will then build the Blu Zone Restaurant menu, staffing and kitchen operations. If that goes the way I expect, we will offer him a permanent contract.
Water pressure in the cafe is still a problem and on bad days we have to close the toilet. There remain numerous small snags and unfinished items across front of house, back of house and the public areas. It is not dramatic, but it is not finished, and it is visible to guests.
ILA is now engaged for tax reporting and compliance only. We are considering moving to Mayple as a complete CFO package covering BZG, AMG and KeenEye. The day to day financial control has moved to the Alchemist system, which is described in Section 07.
Revenue is up 37.6% on June while covers are up 24.5%, meaning we sold more and sold it at a higher price per guest at the same time. That combination is the hard one. Prime cost fell below the 60% line for the first time. Supplier arrears went from a hundred million rupiah problem to Rp 32.9M inside a month. Google reviews grew by 42 with the rating holding at 4.9.
The prime cost improvement is driven by revenue leverage on a fixed wage bill, not by getting food cheaper. Labour is Rp 88.5M every month regardless of trade; spread over Rp 357.8M instead of Rp 260.0M it falls from 34.0% to 24.7% of revenue. The wage figure is sound: the June payroll run on the Mandiri statement was Rp 90,730,769, 2.5% above the HR Master run rate of Rp 88.5M, so labour is bank confirmed rather than estimated. Food cost is the problem. It did not move as a percentage because it is held at 34.5% in the model, derived from recipe cost cards rather than from what we actually bought and consumed. Until we run monthly stock counts, we do not know our true food cost or our wastage. That is the single most important gap in this report and it is Ask 2.
The ABC alcohol licence has been pending since the May report. Every month without it is a month of bar revenue we do not earn, on a venue built with a bar. It sits with Erick. Shareholders are asked to agree an escalation route this month, see Ask 1.
Four full months of trading history. Every revenue and cover figure below is drawn from the ESB point of sale system, branch BLZ, and reconciles to 123 days of daily records since opening on 5 April.
April was a partial month, 26 trading days from the 5th. May, June and July are full months, so the June dip and the July recovery are genuine trading movements, not calendar artefacts.
| Month | Days | Net revenue | Covers | Bills | Avg / cover | Rev / day | Covers / day | Prime % |
|---|
Flat at ~Rp 130k for three months, then a decisive break upward in July. This is the metric that carried the month.
Prime cost is food and beverage cost plus labour. Under 60% is the industry line between a venue that works and one that does not. July is the first month across it.
Read top down. Each step is a percentage of net revenue. Gross margin is before labour; contribution is what is left after food cost and wages, and before rent, utilities, licences and overhead, none of which are yet captured in the system.
Cost of goods is modelled at 34.5% of net revenue from recipe cost cards covering 93% of revenue. Labour is the current wage run rate of Rp 88.5M applied to every month. Neither is a purchase ledger or payroll ledger actual, see the data integrity panel at the end of this section.
| Line item | Apr 2026 | May 2026 | Jun 2026 | Jul 2026 | Jul vs Jun |
|---|
Gross collected includes PB1 tax and service charge, so it is the cash the business actually took. Seventeen weeks from opening. The June trough and the late July peak are both clearly visible.
Complete and verified: revenue, covers, bills, average spend, service charge and PB1, all from ESB, 123 consecutive days, no gaps.
Modelled, not actual: cost of goods (recipe cards, 93% of revenue costed, 48 items still uncosted worth ~Rp 79M) and labour (current wage run rate). No stock counts are running, so wastage is unmeasured.
Lagging: bank balances are verified to 30 June, with Mandiri to 15 July. July and August statements have not been uploaded. Not captured at all: rent, utilities, licences, depreciation, capex and tips, so there is no EBITDA figure in this report and any number described as "contribution" is above those lines.
Where inside the month the growth actually came from, by day, by day of week, and by half. The daily chart is annotated with the two Instagram posts that launched the Indonesian menu, because the timing is the most useful fact in this report.
The rolling average strips out the day of week sawtooth. It climbs almost without interruption from Rp 8.5M a day at the start of the month to a peak above Rp 13M a day in the final week, a genuine trend rather than a few good Saturdays. The two orange markers are the Instagram teaser on 20 July and the Indonesian menu launch post on 22 July. The week they sit in, 20 to 26 July, is the highest revenue week the business has ever recorded.
Revenue per day rose 26.1% between the two halves of July. Covers per day rose only 2.5%. Almost the entire gain came from average spend, which jumped 23.1%, the same number of guests spending significantly more each.
The second half surge is a ticket story, not a footfall story. Whatever changed mid month, the Indonesian menu bedding in, better upselling, a shift toward higher value dishes, is working on the guests already coming through the door. That is the cheapest growth there is: it needs no additional marketing spend and no additional covers to serve. Protect it, and find out precisely which items moved so it can be repeated deliberately rather than accidentally.
The teaser post went out at 7:39pm on 20 July and the launch post at 3:37pm on 22 July, earning 6.97% engagement against an account average of 5.63%. The week those two posts sit in, 20 to 26 July, is the highest revenue week in the venue history at Rp 99.2M net, 23% clear of the best week outside July. Three of the four highest revenue days ever recorded fall inside it: 23 July at Rp 17.1M, 24 July at Rp 16.1M and 25 July at Rp 15.6M, ranking second, third and fourth across all 123 trading days. In fairness the single highest day on record is still 27 May at Rp 17.9M, so July did not take every crown. Average spend across the second half of the month sat 23.1% above the first half. Section 08 has the social numbers and the caveat that matters: the Instagram audience is 36% Mataram, so this reads as the local market responding to a local menu rather than a tourist wave. Either way, two posts preceded the best week the business has had, and that is a repeatable lever rather than a lucky month.
Saturday is the strongest day on both revenue and covers. Thursday is a close second on guests but has the lowest average spend of the week. Monday is the weakest day for covers; Tuesday is the weakest for revenue.
Darker cells are higher values, read across each row independently. Thursday is the anomaly: 90.8 covers a day, almost Saturday's 92.2, but the lowest average spend of the week at Rp 133k. A busy, low ticket daypart is worth investigating: it may be a promotion, a group booking pattern, or simply weaker upselling under volume. Monday and Wednesday carry the fewest guests and are the obvious targets for the catering and room service ideas in Section 09.
Every single day of the week improved. The largest gains are Thursday (+50.9%), Friday (+50.4%) and Saturday (+47.1%), the recovery is broad, not concentrated. Sunday, June's strongest day, grew least (+8.9%): it was already close to capacity.
The spread between the best and worst trading day in July was 2.4×. Understanding that gap is where the next margin gain lives, staffing to a Rp 17M day and a Rp 7M day should not look the same.
| Date | Net revenue | Covers | Avg spend |
|---|
July recorded 79 cancelled bills against 1,576 completed bills (5.0%), in line with June (4.9%) and May (6.9%). It is stable, but 5% is high enough to be worth a root cause: order errors, guests walking, or till handling mistakes each imply a different fix. Recommend the Alchemist bot start flagging cancellation reason at till closure.
Every dish on the menu, costed against its real ingredient prices and plotted against how often it sells. This is where the next Rp 23.6M a month is sitting.
Each bubble is a dish. Horizontal position is units sold; vertical position is gross margin percentage; bubble size is total revenue contributed. The dashed lines are the menu medians, which creates the four classic quadrants. Stars (top right) sell well at good margin, protect them and never discount them. Plowhorses (bottom right) are popular but thin, bundle or nudge the price. Puzzles (top left) are profitable but nobody orders them, merchandise them harder. Dogs (bottom left) do neither, cut, rework or reprice.
83 of the 145 menu items are plotted, those with a finished cost card, at least 20 units sold, and a margin inside 0 to 100%. Four otherwise eligible items are excluded because their recorded cost exceeds their sell price, a cost card error rather than a real negative margin dish (see the repricing note below). The 48 uncosted items cannot be plotted at all until their cost cards exist.
145 items. Well over a third are Dogs, a menu this long on a 25 person team is a wastage generator and a service speed tax. Culling the tail is a margin action and a kitchen action.
Bite Boards (food) and Refreshments (drinks) are close to even by revenue, but drinks move 40% more units at a much lower ticket. Drinks are the volume engine; food is the value engine.
The Pesto Chicken Toastie remains the single biggest revenue line in the business, and Hot Cappuccino remains the single biggest volume line. Both were top of the May report, the hero items are consistent, which is a good sign for menu stability.
| Item | Category | Units | Revenue | GM % | Class |
|---|
Sorted by how much margin each dish would return per month if brought to its cost of sales target, not by how far off target it is, which would put unsellable oddities at the top. These twelve recover Rp 15.8M of the Rp 23.6M total, from a single afternoon's work. Each row shows both levers: cut the cost to the figure in the final column, or raise the price. Six of the twelve (French Toast, Pastriami & Cheese Toastie, Oat Milk Substitute, Bonsoy Substitute, Egg Benedict Smoked Beef and Green Detox) run at food cost above 45%; they are sold at close to no margin. Note also that the two milk substitutes are Plowhorses selling 224 and 130 a month at 82% and 91% cost of sales: they are a giveaway, not a product.
| Item | Type | Sold / mo | Net price | Cost | CoS now | Target | Over by | → cut cost to | Recover / mo |
|---|
A further five items carry a recorded cost higher than their sell price. Chicken, Extra Hashbrown, Sparkling Water, Almond Milk and Bean Decaff. These are cost card errors, not real negative margin dishes, and they are excluded from the Rp 23.6M above rather than inflating it. They need correcting before the next cost review.
The Indonesian menu has launched, but the fish curry cannot be reheated at service speed, particularly the bowl format. A commercial microwave is a modest piece of capex that directly unblocks a dish on a menu we launched specifically to reach the Mataram local market. Recommend approving it in August rather than letting a signature dish underperform on ticket time. Included in Ask 3.
Three bank accounts, every one reconciled to a statement. Where the money is, where it went, and what is still owed.
Bank balances are verified from statements as at 30 June 2026, with the Mandiri day to day account carried forward to 15 July. July and August statements have not been uploaded, so the group cash figure of Rp 306.3M is a June position, not a today position. Given the business has since traded its strongest month, the true figure is very likely higher. Closing this lag is Ask 4.
Total group cash Rp 306,317,235 across the three accounts. Every account has been confirmed against the statement file itself and against the Director directly, bank identity is never inferred from an account number.
Every month reconciles: opening plus credits minus debits equals the verified statement closing. Note how thin the closing balance runs. Rp 9.1M at the end of May. This account is a conduit, not a reserve, and it has been close to the edge more than once.
| Period | Opening | Credits in | Closing | Reconciles |
|---|
July figures cover 1 to 15 July only, because that is where the statement ends. Do not read the July column as a full month.
| Flow type | Apr | May | Jun | Jul 1 to 15 |
|---|
Ordered Rp 554.5M against paid Rp 594.8M across the whole period since March, leaving Rp 32.9M still owed across 11 suppliers. Every one of these is a small residual within terms, a very different picture from the Rp 100M+ of unrecorded payables that the previous purchasing arrangement left behind.
Supplier accounts are paid every Monday, including monthly terms accounts, to build the relationships and the pricing leverage that come with being the customer who always pays early. That commitment lapsed under the previous purchasing arrangement; it is back in force and the outstanding balance proves it.
Two recurring bill payment references leave the Mandiri account every month. They have run without a gap since April. The May and June shareholder reports were wrong to describe statutory contributions as uncaptured, and that is corrected here.
| Paid | Reference ending 2350733 | Reference ending 6574000 | Month total |
|---|---|---|---|
| 20 April 2026 | Rp 5,975,000 | Rp 598,690 | Rp 6,573,690 |
| 4 May 2026 | Rp 4,485,000 | Rp 575,663 | Rp 5,060,663 |
| 9 June 2026 | Rp 4,350,000 | Rp 592,610 | Rp 4,942,610 |
| 7 July 2026 | Rp 4,550,000 | Rp 592,609 | Rp 5,142,609 |
| Four months | Rp 19,360,000 | Rp 2,359,572 | Rp 21,719,572 |
Settled: statutory contributions are being paid, monthly, on time, for four consecutive months. The pattern of one larger payment and one much smaller one, at these amounts against a headcount of 25, is consistent with the two national schemes. Not settled: which reference is which scheme, and whether every one of the 25 active staff is enrolled. The bank proves money left; it does not prove coverage. Producing the enrolment register against the payroll list is the remaining task, and it is Ask 5.
The account made 362 outbound payments totalling Rp 927,514,065 between March and 15 July. Only Rp 552,302,921 of that matches a supplier invoice. The table below accounts for the rest, and is honest about the part that is still unattributed.
| Category | Amount | Share of outbound | Evidence |
|---|---|---|---|
| Supplier invoices matched | Rp 552,302,921 | 59.6% | Reconciled to invoices |
| Payroll, 29 payments | Rp 94,810,769 | 10.2% | Bank narration |
| Service charge and tips, 38 payments | Rp 27,036,758 | 2.9% | Bank narration |
| Statutory bill payments, 8 payments | Rp 21,719,572 | 2.3% | Bank narration |
| Bank and card charges | Rp 36,000 | 0.0% | Bank narration |
| Not yet classified | Rp 231,608,045 | 25.0% | Needs attribution |
| Total outbound, March to 15 July | Rp 927,514,065 | 100% | n/a |
Rp 231,608,045 left the Mandiri account between March and 15 July without being classified. Much of it will be legitimate supplier payments whose invoices were never recorded, which is exactly the failure that broke June, and some will be payments to categories the workbook has no bucket for. This is not an allegation of anything; it is an accounting gap, and it is the largest single unexplained number in this report. Every line carries a date, an amount and a bank narration, so it is fully resolvable. The Alchemist system now prevents new payments from entering this state, but the historical Rp 231.6M still needs attributing. Recommend a one off classification pass on the 362 payment lines, before the CFO decision is made, so whoever takes the mandate starts from a clean ledger.
PB1 is a genuine liability collected on the group's behalf. Service charge is a pass through to staff; it is neither revenue nor a business cost, and it must be distributed in full.
| Obligation | July amount | Status |
|---|---|---|
| PB1 restaurant tax (10%) | Rp 35,777,175 | Confirm remittance with ILA |
| Service charge collected | Rp 39,354,895 | Distribute in full |
| Service charge distributed, to 15 Jul | Rp 8,800,126 | Bank confirmed |
| Statutory bill payments, July | Rp 5,142,609 | Bank confirmed |
| Payroll, June run | Rp 90,730,769 | Bank confirmed |
| Tax and accounting fees | Rp 2,500,000 | Paid, no invoice matched |
| PPh 21 and PPh 25 | Not separable | Needs the filings |
| PB1 plus service charge, July | Rp 75,132,070 | n/a |
Correction to the May and June reports. Those reports treated statutory contributions as uncaptured. They are not. The Mandiri statement carries two recurring bill payments every month from April through July without a gap, one of roughly Rp 4.35M to 5.98M and one of roughly Rp 575k to 599k, totalling Rp 21,719,572 over the four months. Payroll and service charge distribution are on the same statement. What is genuinely missing is not the money but the paperwork: the per employee enrolment register, the split between the two schemes, and the PPh 21 and PPh 25 filings that should sit alongside the payments. That is what Ask 5 now requests.
Every entry is photographed into Telegram and filed automatically with an attributed logger. Rp 7,129,100 spent across the last four weeks (13 Jul to 3 Aug), of which Rp 1,560,700 falls in August to date. Float on hand Rp 301,300.
Staff meals are 63% of all petty cash spend. Rp 4.52M across 94 entries. That is a real cost of employment sitting outside both the wage line and food cost, and it is running at roughly Rp 1.1M a week. Recommend it be budgeted explicitly rather than absorbed by the tin. Note also that top ups (Rp 5.76M) trailed spend (Rp 7.13M) over the four weeks, the float is being run down.
July was the month the team stopped being a cafe roster and started being the staffing plan for a multi outlet group. Most of the movements below are promotions into venues that have not opened yet.
Kitchen and floor carry 66% of the wage bill between them. Security at Rp 12M across four heads is 13.6% of payroll for a single outlet cafe, worth reviewing once the other venues open and the cost can be shared across the complex.
Working harder than the role asks, the Director has to send her home. Developing as a leader gradually and genuinely, with Sunday morning mentoring every week. Base raised from Rp 4.2M to Rp 5.5M and placed on an outlet sales bonus (below). Development risk: she involves herself too deeply in colleagues' personal lives. If that does not change it will limit her, and she has been told so directly.
Joined to educate, mentor and organise the cafe to a higher standard. On completion he moves to building the Blu Zone Restaurant menu, staffing and kitchen operations from scratch. A permanent contract will be offered subject to performance. This is now a key person dependency on the restaurant launch, see the risk register.
Developed the cocktail programme and showed strong leadership behind the bar. Salary increased by Rp 300,000 to Rp 4.5M monthly. Note that she has built a cocktail offer the venue is not yet licensed to sell, the ABC licence turns her work into revenue.
Both told they will be the new captains of the restaurant. Willing to learn, exceeding in role, and showing positive leadership signs. Their promotion creates one dining room vacancy at Uma Blu.
Both bar staff earmarked for the Pool Bar on opening. Imran is repeatedly named by guests in five star Google reviews, he is a genuine brand asset and the Pool Bar is the right stage for that. Their move creates one bar vacancy at Uma Blu.
Contract not renewed. Not the standard Atrium should be known for. Handled at the natural renewal point rather than mid term.
Internship ends mid August. Returns approximately one month later after university graduation to join the bar permanently. A planned, temporary gap rather than a loss.
Replaced, not by a person, but by a system. Errors and unreported payables under this role totalled over Rp 100,000,000 of unpaid product, which is the direct cause of June's supplier relationship damage. The function is now automated end to end (Section 07). All suppliers are current and goodwill is being rebuilt.
Tiered on monthly outlet sales, with an open ended top end. Aligns the Operations Lead directly to the number shareholders care about.
| Monthly outlet sales | Bonus | Status vs July |
|---|---|---|
| Rp 450,000,000 | Rp 1,000,000 | Rp 17M short on gross |
| Rp 500,000,000 | Rp 2,000,000 | Rp 67M short on gross |
| Rp 550,000,000 | Rp 2,500,000 | Stretch |
| Each further Rp 50,000,000 | + Rp 500,000 | Open ended |
July gross collected was Rp 432.9M and July net revenue was Rp 357.8M, a Rp 75M gap, because gross includes PB1 tax and service charge that never belong to the business. Even on the more generous gross basis, July fell Rp 17.1M short of the first tier; on net it was Rp 92.2M short. So no bonus is payable for July on either reading, but August is close enough on gross that the definition will matter within weeks. Shareholders should fix the basis before August closes so the first payment is not a dispute. Recommendation: net revenue, since that is the figure the business actually keeps.
The promotions and redeployments above take four experienced people out of Uma Blu into venues that have not opened. The cafe needs backfilling before, not after, those venues launch.
Two external hires and one or two internal transfers rebalance the team without growing the Rp 88.5M wage line materially. If the wage bill holds at ~Rp 88.5M and August revenue holds at July's level, labour stays at roughly 24.7%, inside the 25 to 30% target band.
The May and June reports recorded statutory contributions as uncaptured. That was wrong, and the Mandiri statement proves it: two recurring bill payments every month from April through July, Rp 21,719,572 in total, alongside the June payroll run of Rp 90,730,769 and Rp 27.0M of service charge distributed to staff. Section 05 sets out the evidence line by line. What is still missing is documentation, not money: the per employee enrolment register for all 25 staff, which reference belongs to which scheme, and the PPh 21 and PPh 25 filings that should accompany the payments. That is the narrower and much more tractable version of Ask 5.
The most consequential thing that happened in July was not a revenue number. It was replacing a failed control function with an automated one that scales to every outlet we open.
The purchasing role generated over Rp 100,000,000 in unreported payable product. Invoices were not recorded, so suppliers were not paid on the Monday cycle we had committed to. That is what broke June: not demand, not the menu, and not the team on the floor, a single unmonitored control point.
Staff photograph a document into Telegram. The bot classifies it, files it to the right ledger with the logger attributed, and the system reconciles the result against the bank statement looking for discrepancies. No human sits between the document and the ledger, so nothing can quietly fail to be recorded.
The system is already built to accept all remaining F&B areas as they open. When Blu Zone Restaurant, Pool Bar, Rooftop, Underground, Catering and Room Service come online, they attach to the same pipeline rather than each needing their own purchasing officer. That is the argument for the whole approach: the control cost of the seventh outlet is close to zero.
Nothing in this report is estimated where it could be measured. Where a gap exists it is shown as a gap.
| Domain | Source | Coverage | Status | What is missing |
|---|---|---|---|---|
| Sales | ESB POS, branch BLZ | 123 days | Complete | Six outlets not yet on POS |
| Food cost & margin | Recipe cost cards | 93% by revenue | Partial | 48 items uncosted, ~Rp 79M revenue |
| Suppliers | Audit master | 49 suppliers | 84% banked | 8 suppliers have no account number on file |
| Petty cash | Telegram bot | Live, attributed | Live | Opening float from outlets |
| Labour | HR Master, cross checked to bank | 25 staff, June run confirmed | Confirmed | Rostered hours, tips |
| Cash | Bank statements | To 30 Jun / 15 Jul | Lagging | July and August statements |
| Stock & wastage | , | Not running | Gap | Monthly counts, blocks true food cost |
| Overheads | , | Not captured | Gap | Rent, utilities, licences, capex, blocks EBITDA |
| Statutory contributions | Mandiri statement | Apr to Jul, no gaps | Paid, verified | Enrolment register and scheme split |
| Payroll tax filings | n/a | Not captured | Partial | PPh 21 and PPh 25 returns |
| Payment attribution | Mandiri statement | 75% classified | Gap | Rp 231.6M of outbound has no category |
| Social analytics | Metricool, Uma Blu brand | July, all three platforms | Complete | Nothing missing for Uma Blu |
Full July analytics from Metricool across Instagram, Facebook and TikTok, plus the live review position. This section answers the question the rest of the report cannot: where did the guests come from.
The teaser went out on Instagram at 7:39pm on 20 July, captioned "coming this week". The launch post went out at 3:37pm on 22 July and earned 6.97% engagement on 1,030 views, against an account average of 5.63% and a reach of 445 accounts. The week those posts sit in, 20 to 26 July, produced Rp 99,197,517 net, the highest weekly revenue in the venue history, 23% above the best week outside July. Three of the four highest single days across all 123 trading days fall inside that week: 23 July Rp 17.1M, 24 July Rp 16.1M and 25 July Rp 15.6M, ranked second, third and fourth. The all time single day record is still held by 27 May at Rp 17.9M. The chart in Section 03 marks both posts against daily revenue. This is correlation rather than proof, but the alignment is exact and the mechanism is obvious.
Instagram follower location for July. This is the single most commercially useful number in the section, and it reframes the month.
Over 36% of the Instagram audience is in Mataram, and eight of the top ten cities are West Nusa Tenggara: Sengkol, Midang, Praya, Narmada, Kediri, Kuripan, Masbagik and Pancor. Jakarta is 1.1%. Australia, the United Kingdom, the Netherlands, Spain, France and the United States all appear, but far down the list. The social channel that drove July is reaching locals, not tourists, and the menu it promoted was built for the local market. The Director read the month as tourist led; the audience data reads as Mataram led. Both can be partly true, and ESB does not record guest origin, so this cannot be settled from the data we hold. Adding a single origin question at the till would settle it permanently and costs nothing. Recommended for August.
Reels reach 2.5 times as many accounts per item as a static post. Stories reach few accounts each but there were 113 of them, so they still delivered 10,140 impressions.
Six reels produced 7,557 views and 168 interactions. Eight posts produced 6,494 views and 153 interactions. Reels do more with fewer items. Shifting the ratio toward reels is free and the July data already argues for it. The top reel, the team favourite picks on 17 July, reached 1,730 accounts at 30.4% retention, five times the average post.
Pieces published in July against interactions earned. The disparity is the point.
| Platform | Published | Followers | Views | Interactions |
|---|---|---|---|---|
| 127 | 1,172 | 52,520 | 321 | |
| TikTok | 13 | 37 | 4,454 | 70 |
| 82 | 17 | 1,302 | 5 | |
| Total | 222 | 1,226 | 58,276 | 396 |
82 pieces of content went to Facebook in July and earned 5 interactions. Average reach per post was 8 accounts. The page has 17 followers. TikTok, with 13 posts, earned 70 interactions and 4,454 views off a 37 follower base. Recommendation: stop cross posting to Facebook, keep the page live for search and check ins only, and move that production time to reels and TikTok. This is a free gain in output quality.
| Metric | Posts | Reels | Stories | Note |
|---|---|---|---|---|
| Published in July | 8 | 6 | 113 | 127 pieces on Instagram alone |
| Views | 6,494 | 7,557 | 10,140 | Stories counted as impressions |
| Average reach per item | 340 | 843 | 85 | Reels lead by 2.5 times |
| Engagement rate | 5.63% | 3.32% | n/a | Posts above the 5% target |
| Interactions | 153 | 168 | n/a | 321 combined |
| Likes | 135 | 123 | n/a | 16.88 likes per post |
| Comments | 1 | 3 | n/a | Almost no conversation, see below |
| Saves | 5 | 7 | n/a | Saves signal intent to visit |
| Shares | 12 | 35 | n/a | Reels are shared three times as often |
| Account level, July | 52,520 views, 355 accounts reached per day, 515 accounts engaged, 51 new followers | |||
The eight posts and six reels together drew 4 comments all month against 258 likes. Reach and likes are healthy; conversation is not happening. Likes are passive and cheap. Comments and saves are the signals that precede a visit. A single question in each caption, and replying to every comment the way the Google reviews are handled, is the obvious fix and costs nothing.
The May report showed 5.0 stars from 91 reviews. Today it is 4.9 from 133. That is not a decline in quality; it is the arithmetic of scale. A perfect score on low volume reads as untested. A 4.9 across 133 reviews reads as consistently excellent and is far more persuasive to a traveller comparing options. 42 new reviews in roughly two months while holding 4.9 is the strongest brand signal in this report.
Three themes dominate the review summary: food quality and portion size, the atmosphere and interior, and above all the people. Named staff appear repeatedly: Imran, Awan, Bruce. That is a service culture showing up in public, and it is worth more than any campaign.
"Good place in the centre of Kuta. It is good for breakfast. Coffee is a bit pricier than other places but delicious. Staff are super friendly and kind."
Pedro Pareja, Google Local Guide, 98 reviews, two months ago. Owner replied
"All the staff took amazing care of me, the service was truly excellent! Huge thanks to Imran and the rest of the team."
PIA, Google, two months ago. Owner replied
"The place is beautiful, the food is good, but the best is the people who work there. Thank you all for being so kind and helpful, especially to Bruce."
Júlia P, TripAdvisor, August 2026.
"Amazing little brunch and coffee place we stopped by after the beach, amazing service. The burrito was so good and the french toast was heavenly! So much time and effort is put into every dish."
Naairah Ali, Google, three months ago. Owner replied
TripAdvisor volume. Three reviews at 5.0 puts the venue number 4 of 8 coffee spots and number 560 of 1,078 restaurants in Kuta. Google momentum has not transferred. A single prompt at bill presentation, directed at TripAdvisor rather than Google for one month, would move the ranking materially at almost no cost. Guest origin. The Instagram audience says Mataram, the Director says tourists, and ESB records neither. One question at the till, logged through the Alchemist bot the same way petty cash is, ends the ambiguity and lets every future report attribute revenue to a market.
The July figures above are pulled directly from Metricool for the Uma Blu Cafe brand, period 1 to 31 July 2026, covering Instagram, Facebook and TikTok. Separately, the Goss Group brand connection in the same Metricool account has been disconnected since approximately 23 June and still needs reconnecting. That is a different brand and does not affect any Uma Blu number in this report.
One venue trading, six to come. July's people decisions have already allocated staff to three of them, the constraint has moved from labour to licensing.
Bars show current readiness, not a committed opening date. No outlet has a confirmed launch date, because licensing has not been confirmed. That is stated as a fact rather than filled with an estimate.
Three of the six pipeline venues are bars. All three are gated on the same document.
People are allocated. A consultant chef is engaged. The system scales. Licensing is the only thing standing between the group and three additional revenue lines, and it has been open since the May report without a stated deadline or an escalation path.
Catering and Room Service appear last in the workbook and have had no attention, but they are the only two pipeline venues that need no new licence, no new kitchen and no new fit out.
While the alcohol licence sits with a third party, sequence Room Service and Catering ahead of the bars. They are the only outlets whose timeline the group fully controls.
Positioned by likelihood and impact. Every risk carries a named owner and a mitigation.
| Risk | L | I | Owner | Mitigation and status |
|---|---|---|---|---|
| ABC alcohol licence delay | High | High | Erick / Wolf | Open since May report. Blocks three bar venues and caps Uma Blu bar revenue. Escalation requested. Ask 1 |
| True food cost and wastage unknown | High | High | Harianto / Wolf | COGS is modelled from recipe cards at 34.5%; no stock counts running. Real wastage unmeasured. Ask 2 |
| Rp 231.6M of outbound payments unclassified | High | High | Wolf / CFO | 25% of all Mandiri outbound March to 15 July has no category. Every line is dated and narrated, so it is resolvable. New this month, Ask 5 |
| Statutory enrolment register missing | Medium | Medium | Tisa / ILA | Payments are running monthly and bank confirmed. What is absent is the per employee register, the scheme split, and the PPh 21 and PPh 25 filings. Downgraded, payments verified |
| Overheads absent from reporting | Medium | High | ILA / Mayple | Rent, utilities, licences, depreciation and capex uncaptured, no group EBITDA can be stated. CFO decision pending |
| Construction snags visible to guests | High | Medium | Wolf / contractor | Numerous unfinished items across FOH, BOH and public areas. Snag list to be formalised with dates |
| Water pressure, toilet closures | High | Low to Med | Atrium / Wolf | Recurring. Toilet closed on bad days, a direct guest experience hit in a 4.9 rated venue. Unresolved since May |
| Procurement without approval | Medium | Medium | Alchemist / Devi | Chilled goods bought with no refrigerated display; stock kept at room temperature and largely binned. Director was not informed and would have refused. New this month. PR approval gate now mandatory |
| Mandiri operating balance runs thin | Medium | Medium | Wolf | Closed at Rp 9.1M end of May against ~Rp 250M monthly throughput. BNI backstop; minimum float to be set |
| Key person: Harianto | Low | High | Wolf | Restaurant menu, staffing and kitchen operations all depend on one three month consultant. Require written documentation as a contract deliverable |
| Bonus basis dispute | Low | Medium | Shareholders | Gross vs net definition unresolved on Devi's sales bonus, a Rp 75M gap. Settle before August close |
| Devi, boundary risk | Low to Med | Medium | Wolf | Over involvement in colleagues' personal lives. Addressed directly; weekly Sunday mentoring in place. Monitored |
| 48 uncosted menu items | Medium | Low | Harianto | ~Rp 79M of revenue with unknown margin, largest being Egg Benedict Salmon at Rp 19M. Cost cards required |
| Competitor opens nearby | Low | Low to Med | Wolf | 4.9 rating across 133 reviews, named staff, local menu ownership. Well defended |
One month, one quarter, one year. Targets are stated so that next month's report can be marked against them.
The first five days of August have already delivered Rp 59.4M at an average spend of Rp 162,844, 12.6% above July's average and the highest run rate the venue has recorded. On five days it is a signal, not a trend, but the direction is right.
Five items. Three need a decision, two need acknowledgement. Please respond using the buttons at the end of this report.
The licence has been with Erick since before the May report. It blocks three pipeline venues and caps Uma Blu's own bar revenue, with a cocktail programme already built and unsellable. The ask is not for more effort; it is for shareholders to agree a dated deadline and a named alternative route (a second consultant, a direct application, or a legal escalation) if the licence is not issued by 31 August 2026.
Every food cost figure in this report is modelled from recipe cards. Without a physical count we cannot state true food cost or measure wastage, and wastage was the exact mechanism by which the chilled goods purchase became a total loss. The ask is to mandate a month end count from August, owned by Harianto during his engagement and by the Operations Lead thereafter.
The commercial microwave unblocks the fish curry at service speed on a menu launched specifically for the local market. The three dome displays replace the retired baked goods unit at the customer end of the bar, where the uplift in bakery sales has already been proven. Both are modest, both have a direct revenue line, and both are held pending shareholder approval. Indicative costs to be circulated with the purchase requests.
Group cash of Rp 306.3M is verified only to 30 June. July and August statements have not been uploaded, so the reported position understates the likely truth after a record month. Separately, ILA is now engaged for tax and compliance only, and a move to Mayple as a complete CFO package covering BZG, AMG and KeenEye is under consideration. Shareholder views on that consolidation are welcome.
Rp 231,608,045 of Mandiri outbound payments between March and 15 July has no category. That is 25% of everything the account paid out. Each line has a date, an amount and a bank narration, so it is resolvable rather than lost, and most of it is likely supplier payments whose invoices were never recorded, the same failure that broke June. The ask is to approve a one off classification pass over the 362 payment lines and to have it finished before the CFO mandate is awarded, so whoever takes it inherits a clean ledger.
Two smaller confirmations in the same breath. Statutory contributions are being paid, monthly and bank confirmed, which corrects the May and June reports; what remains is the enrolment register and the PPh 21 and PPh 25 filings, owned by Tisa and ILA. And Devi's bonus basis needs fixing: gross collected, Rp 432.9M in July, or net revenue, Rp 357.8M. The recommendation is net revenue. Please confirm so the first payment is not a dispute.
| May ask | Outcome | Status |
|---|---|---|
| Note and acknowledge ILA engagement for accounting and monthly close | Partly delivered, then changed. ILA is now scoped to tax reporting and compliance only. Day to day financial control moved to the Alchemist system. A full CFO package via Mayple is under consideration. | Superseded |
| Note and acknowledge the AI daily reporting agent going live within 2 weeks | Delivered and exceeded. The system refreshes every morning at 05:47 WITA and has gone well beyond daily metrics, it now runs petty cash, purchase requests, goods received, nightly till closure and bank reconciliation, and is already built for all remaining outlets. | Delivered |
| Input requested: priority metrics each shareholder wants in their daily report | No shareholder responses received. The daily board pack was built to a default specification, revenue, covers, margin, cash and payables. The request stands and is easier to act on now that the pipeline exists. | No response |
| May action: reduce food COGS with FIFO and waste protocol | Prime cost is under target, but through revenue leverage rather than measured cost reduction. FIFO discipline cannot be verified without stock counts. | Unverifiable |
| May action: BPJS enrolment audit, all staff, due 20 June | Partly done, and previously misreported. The contributions themselves have been paid every month since April, evidenced on the Mandiri statement, so the May and June reports were wrong to call this uncaptured. The enrolment audit itself, matching all 25 staff to a register, has still not been produced. | Payments yes, register no |
| May action: launch Indonesian menu, 15 June | Launched. Fish curry constrained by reheat speed pending a commercial microwave. | Delivered |
| May action: install baked goods display, 10 June | Delivered and already superseded. Bakery sales lifted; the unit is being retired for three dome displays at the customer end of the bar. | Delivered |
| May action: launch alcohol menu, June, licence pending | Not delivered. Licence still with Erick. Cocktail programme built by Yanti in the meantime. | Blocked |
| Action | Owner | Due | Status |
|---|---|---|---|
| ABC alcohol licence, dated escalation path agreed | Wolf / Erick | 15 Aug | Critical |
| First month end physical stock count | Harianto / Devi | 31 Aug | Pending approval |
| Execute top 12 repricing rows, measure margin effect | Harianto / Devi | 20 Aug | Ready |
| Cost cards for all 48 uncosted items | Harianto | 31 Aug | Not started |
| Hire 1 dining room, 1 bar | Devi / Tisa | 25 Aug | Open |
| Transfer one or two kitchen heads to main kitchen | Harianto | 31 Aug | Planned |
| Upload July and August bank statements | Wolf | 10 Aug | Overdue |
| Produce the statutory enrolment register for all 25 staff | Tisa / ILA | 31 Aug | Payments verified, register open |
| Classify the unattributed Rp 231.6M of Mandiri outbound | Wolf / ILA | 15 Sep | Critical |
| PPh 21 and PPh 25 filings into the reporting system | ILA | 31 Aug | Gap |
| Install 3 dome displays; retire baked goods unit | Wolf / Devi | 20 Aug | Pending approval |
| Commercial microwave installed | Wolf | 20 Aug | Pending approval |
| Snag list formalised with owners and dates | Wolf / Devi | 15 Aug | Open |
| Water pressure escalated to Atrium in writing | Wolf | 12 Aug | Open |
| Confirm bonus basis, gross or net | Shareholders | 31 Aug | Awaiting |
| Scope Room Service and Catering launches | Wolf / Harianto | 31 Aug | Recommended |
| Reconnect the Goss Group brand in Metricool | Wolf | 15 Aug | Open, does not affect Uma Blu |
| Add a guest origin question at the till, logged via the bot | Devi | 20 Aug | Recommended |
| Stop cross posting to Facebook, move the time to reels and TikTok | Wolf | 15 Aug | Recommended |
| CFO decision. ILA vs Mayple package | Shareholders / Wolf | Sept | Under review |
Click your name below. It will open a already addressed email to mrwolf@wolfcommand.net with a short structure already filled in, acknowledgement, your vote on each of the five asks, and space for anything else. Edit freely and send.
If the button does not open your mail app, email mrwolf@wolfcommand.net directly with the subject BZG Shareholder Report. July 2026, [your name].