How a monthly Business Review actually reaches an owner, now drawn from six GM interviews (Accra 20 Jul, Dar es Salaam 21 Jul, Morocco 21 Jul, Abha 23 Jul, Addis Ababa 27 Jul, Dubai 27 Jul) and the first corporate-seat session (finance and commercial, 6 Aug), on top of the kickoff call and the sample packs. The estate is not one process with local variants: it runs four ledger stacks, two artefact architectures, six owner models and two clocks, and at one site the owners deck reaches owner and corporate in a single email. Nothing here is a fix yet; this is the diagnosis, sharpened.
The shared spine of the monthly cycle, five lanes now instead of four. Every property crosses the same manual bridge into the same Excel template; after the property artefact is built the estate splits into two architectures. Solid arrows are the intended flow, dashed arrows are rework, each burst marks a friction from the register below.
The apparent contradiction in the testimony resolved. Where the question was answered, the P&L reaches corporate by the 6th or 7th, and every artefact date we hold sits at day 10 or later; on 6 August the corporate seat confirmed the second clock from the receiving end. Both are true: they are different clocks. Hollow marks are self-reported, solid marks are artefact-dated or observed.
Six sites, four ledger stacks, one shared shape: whatever the stack, the figures leave the ledger through a person re-keying them into Excel, and every site where the bridge has been seen then feeds the same single Aleph P&L template; at the Dubai cluster it runs out of the GM's sight. Because the same mismatch appears on three different ledgers, the defect is template-side: it is Aleph's own file to correct, and the change depends on nobody else.
One reporting obligation on six different sets of rails. The matrix is the estate side by side; the cards below it are the six owner models the future state has to span. The constant across everything: the last check before a number leaves is one named person, at every site, in a different role each time.
| Accra · The Pelican | Dar es Salaam · ONOMO | Morocco · ONOMO cluster | Abha · Grand Iva (Saudi) | Addis · Ramada (Wyndham) | Dubai · JOC cluster | |
|---|---|---|---|---|---|---|
| Ledger + feeders | Sun; two P&Ls (Rooms + F&B); PMS/POS unnamed; statistics keyed by hand | Sage (COA at ONOMO HO); Opera + Simphony summary feed; Materials Control with no interface | Sage; 'Accuracy' reporting layer (unconfirmed) bypassed by the pipeline | Oracle NetSuite (no dept numbers); other two cluster hotels on Sun | Sun, with Opera feeding statistics; mapping suspected, under review by a new finance director | Sun (self-reported); the GM receives the finished Excel and has no visibility of the Sun-to-Excel step |
| Finance org | On-property FC; team of ~10, ~5 on the pack; no regional layer | Cluster finance manager across Dar + Kampala; two in accounts; payroll lands last | No property FMs; finance at HQ Casablanca, outside the country manager's line | Cluster Director of Finance over 3 hotels; property feeders are a supervisor and two others | On-property Director of Finance, one month in post; the GM personally builds the deck | Cluster finance team; account structures and formats are being aligned across the three hotels by the cluster DOF |
| Property artefact | Business review document: 17 blocks + 3-month forecast + appendix, rolled forward monthly | 29-slide pack (already cut once from ~50 pages); tables pasted as pictures | P&L Excel from HQ; deck per hotel; regional consolidated deck on top | Excel workbook, 7 to 8 sheets, zero written commentary; no PowerPoint | 32-slide deck built by the GM, plus the P&L workbook, plus a forecast file sent twice a month | One consolidated deck for two branded hotels plus the marina and retail estate: about 80 slides, grown from about 20 by accretion |
| Deck built at | Property | Property | Property + regional | Regional (never seen by DH) | Property, by the GM himself | Property: departments feed one marketing consolidator; the GM types all commentary |
| Owner model | First-time single-asset owner (best reading Devtraco Group, unconfirmed); weekly informal visits; ~100 fractional owners unscoped | ONOMO owner reps; sub-30-minute monthly meetings; standing questions GOP, cost ratios, net impact | Professional in-house asset manager (owner best reading AHD, unconfirmed); formal presentation per hotel | Local family, only hotel, no asset manager; Arabic summary written personally by the Hotel Manager | Monthly physical board meeting: owner, an advisor and sometimes his son. Interested in revenue, receivables and cash, explicitly not in expenses | Three-principal lessee group; no owners' review for six to eight months; informal contact two or three times a week instead |
| Clock 1 (to corporate) | Close 30th to 3rd, ready 4th-5th, submit 7th SELF-REPORTED | ~5 days end to end, sent ~7th SELF-REPORTED | Not stated; 5th-7th pattern INFERRED | Processing from the 1st, sent 6th-7th SELF-REPORTED | Not asked. Forecast file goes twice monthly SELF-REPORTED | Not asked for the Excel P&L; the owners deck goes out 15th to 20th with corporate copied SELF-REPORTED |
| Clock 2 (owner-facing) | May pack file dated 17 June ARTEFACT; meeting date negotiated fresh monthly | May pack file dated 10 June ARTEFACT | Cluster pack V160626 ARTEFACT; June cycle presented day 22 OBSERVED | No owner meeting date established | Monthly, physical, date not captured | Emailed by the 15th, latest the 20th SELF-REPORTED; June pack held ARTEFACT |
| Single-person gate | The Financial Controller | The cluster finance manager | The country manager (checks every table personally) | The cluster Director of Finance; the Arabic summary is a second single point | Whoever updates the statistics tab: an error there propagates into every report | The GM himself: he writes the commentary and issues the pack; no corporate step before the owner on the deck |
The owner attends most Fridays and messages the GM directly; this is the owner's preferred way of working, not a pack failure. No minutes are taken anywhere, so nothing next month can be checked against what was asked. The meeting date is negotiated fresh each month.
Under 30 minutes every month this year, on GOP, cost ratios and net impact. The calm is explicitly conditional: the GM notes the full pack becomes relevant when performance turns. ONOMO separately checks the net result matches.
The most formal model, and the only one with an observed date: a per-hotel presentation to an in-house asset manager, June cycle presented 22 July, day 22. The country manager presents alongside each GM.
A family owner with many businesses and one hotel, no asset manager, no hospitality background. The working artefact is a personally written Arabic summary of revenue, expenses, profit and bank balance. Whether anyone else reads it was not established.
Owner, one advisor and sometimes his son, facing the Director of Finance and the GM, walking the deck side by side. Content with the pack and asking nothing repetitive. Interested in revenue, receivables and cash, and explicitly not in the expense lines. The clearest materiality signal any owner has given.
The deck is emailed on the 15th to the 20th and no meeting follows: no owners' review in six to eight months, no feedback, no questions. The owners were deeply involved two years ago; since the properties turned around they have stepped back, and the GM sees them informally two or three times a week instead. The author's own verdict on the 80 slides: death by information.
The month as it actually runs at each interviewed site, step by step, with each step assigned to its lane. Timing tags carry the evidence class. Rework loops and per-site friction verdicts follow each map. All six of the introduced sites are now mapped.
| F1 | Present, reframed as a manual re-key: Sun out, Excel offline, back in; two ledgers merged. The chart-of-accounts question was not raised and remains open. |
| F2 | Not raised; open. Nobody described how a number physically reaches a slide. |
| F3 | Not applicable: single property, no regional layer. |
| F4 | Mechanism confirmed (roll-forward by several hands); no defect reported here. |
| F5 | Submission reported on the 7th, self-reported; the May pack on file is dated 17 June. Owner-facing interval unmeasured. |
| F6 | Confirmed but relocated: the cost is 1 to 2 days of waiting on sections, not authoring hours. No hours figure exists. |
| F7 | Confirmed as an event; content commercial, not corrections. Turnaround to be confirmed against email timestamps. |
| F8 | One accuracy issue in two years reported, caught at corporate review; the direct question was not put. |
| F9 | Reframed: weekly informal owner contact is this owner's preferred mode, not evidence the pack failed. |
| F10 | Volume confirmed from the owner's side (the pack was once called too long); readership contradicted (one wrong figure among thousands was caught at corporate). |
| F11 | Not raised; the question was not put. |
| F1 | Confirmed with the mechanism: the template and the Sage chart of accounts do not match; Sage accepts only predetermined codes; the fix at source has never been requested of ONOMO in six-plus years. |
| F2 | Confirmed on the call and in the artefact: the pack's KPI and summary tables are pasted images. |
| F3 | Partly confirmed: two properties hand-built in parallel by one person; no consolidation step witnessed here. |
| F4 | No evidence here; the error class is added-lines-break-totals, found by ticking back through the trial balance. |
| F5 | About five days claimed, pack sent around the 7th self-reported; the May pack file is dated 10 June. |
| F6 | Not confirmed here: commentary is distributed, and every time figure volunteered related to data mechanics, not words. |
| F7 | Confirmed in shape: email, business-why, on the P&L not the pack; no turnaround or round-trip count captured. |
| F8 | Open in both directions: testimony describes internal catches before release; the May pack we hold contains presentation defects that were issued. The direct question was not put. |
| F9 | Dormant: meetings under 30 minutes all year; the GM notes the full pack becomes relevant when performance turns. |
| F10 | Contradicted for today's pack, confirmed historically: the predecessor ran ~50 pages and was already cut once to 29 slides. |
| F11 | No deferral described; the pattern is chasing a broken total down before release. |
| F1 | Present: Sage figures hand-keyed into Excel before the P&L issues; the GMs see only the result. The mapping question cannot be answered from this seat. |
| F2 | Confirmed at the regional tier: tables cross into the presentation as screenshots and copy-pastes; a Morocco number is flattened into an image at least once before the owner sees it. |
| F3 | Confirmed, the programme's only direct observation: historic solo consolidation of four hotels at one week to 10 days per cycle; June moved the drafting to finance, which moved the labour rather than removing it. |
| F4 | Not put as a question; untested here rather than absent. |
| F5 | Two clocks: the owner-facing pack lands day 22, observed and dated. Property arrival dates into the region were not captured. |
| F6 | Confirmed: all commentary, interpretation and harmonisation sits with one person, who also presents. |
| F7 | Confirmed, but the loop runs inside the region, to a finance team outside his reporting line. |
| F8 | No owner-facing error reported, prevented by one personal check: insurance bought monthly with senior hours, resting on one person. |
| F9 | Not evidenced; post-meeting follow-up was not raised. |
| F10 | Confirmed from the consumer seat: the time spent preparing a presentation on data from the past was questioned to Aleph leadership before this programme existed. |
| F11 | Not evidenced: June's incorrect figures were bounced back, not carried forward. |
| F1 | Confirmed on a third ledger, resolving the design fork: the template is built for the Sun users; NetSuite maps by hand; the same code mismatch is reported on the cluster's Sun hotels. Template provenance: developed from one of the African hotels. |
| F2 | Untested here, not absent: the deck is built above the property tier and has not been seen. |
| F3 | Present at regional: three hotels consolidated above the property; build not observed, output not seen. |
| F4 | Not evidenced: no written commentary to carry forward, no prior year; the shared template file is reused monthly. |
| F5 | Month-end wait, then a six-day build to deadline; workbook to regional finance by the 6th or 7th, self-reported. |
| F6 | Written narrative nil, and corporate does not ask; the verbal narrative costs about two hours with the VP plus HOD explanations plus the owner sit-down, and none of it leaves an artefact. |
| F7 | Not reported: corporate accepts where the figures align. |
| F8 | Not tested: the question was not put. The Arabic summary is written by one person; whether anyone else reads it was not established. |
| F9 | Not reported: the owners focus on the bottom line and bank balance. |
| F10 | Live in its purest form: seven or eight sheets of figures carrying no commentary at all. |
| F11 | New: some errors met at the end of the reporting window may be left for the next month. Hedged, no named instance; test elsewhere in identical words. The manual workaround holds only while occupancy is low. |
| F1 | Suspected at a fifth site, on Sun. The incoming finance director is not sure the back-office mapping is correct and is reviewing it as part of his own onboarding. Suspected, not yet demonstrated. |
| F2 | Confirmed by the author of the deck himself: screenshots of the P&L pasted into PowerPoint, commentary typed by hand. |
| F3 | Not applicable today. Single property, though an area cluster may be extended to this GM once the owners activate further projects. |
| F4 | A new and sharper mechanism: not a stale sentence but a stale comparator. Forget the forecast-versus-actual filter and the whole comparison is against an un-actualised forecast. |
| F5 | Not established. The clock was not asked, for the fifth time. One new datum: the forecast file goes twice a month. |
| F6 | Contradicted, deliberately. The GM writes all commentary himself and does not want it automated. The cost he names sits in the back office, not in the narrative. |
| F7 | Not raised. No corporate correction loop was described for this property. |
| F8 | Reframed and sharpened. No owner-facing error reported, but the finance director states there is no way of proving the numbers, so an absence of known errors carries little information. |
| F9 | Dormant. The owner asks no repetitive questions and is content with the deck. |
| F10 | Contradicted. 32 slides, not felt as tedious, and expectations described as light next to a fully branded operator's. |
| F11 | Not raised. The submission deadline was never discussed. |
| F1 | Untested from this seat: the GM receives the finished Excel, the bridge runs inside cluster finance, and he declined to guess at what feeds it. No mapping complaint recalled. |
| F2 | Present by construction: a PowerPoint assembled from department slides. Image-versus-native not yet measured; the June pack is on file to check. |
| F3 | Present in a form no other site has: two branded hotels plus a marina and retail estate consolidated into one deck at the property tier, with no regional lane. |
| F4 | Accretion rather than roll-forward evidenced: about 20 slides became about 80 and additions were never removed. Roll-forward mechanics were not asked. |
| F5 | The deck goes to the owners and to corporate in the same email, the 15th latest the 20th, reporting the previous month; on the deck, corporate cannot catch anything before the owner sees it. Whether cluster finance submits the Excel P&L to corporate separately was not asked. |
| F6 | Confirmed at its heaviest: about 10 to 20 hours of management time a month across contributors, one consolidator, and every comment typed by the GM. The author says the report tells him nothing he has not already been through. |
| F7 | Absent by design: corporate is copied at issue and nothing comes back. The correction loop other sites describe does not exist here. |
| F8 | The author is the only gate. The one catch on record is his: a first-page food cost of 10 per cent at one cluster property, a formula error, caught on sight and stopped before issue; finance traced it immediately once challenged. |
| F9 | Dormant by withdrawal: no review meeting in six to eight months and no feedback; two or three informal meetings a week carry everything. Monthly reviews previously ended near slide 15 of 80. |
| F10 | Confirmed in its most extreme form, by the author: about 80 slides he calls death by information, about 10 marketing slides of which three or four matter, a shortening request declined, and a standing doubt that anyone opens the file. His fix is a short template of the crucial information everybody needs. |
| F11 | Not raised. |
Twelve items: eleven re-cut against six interviews, and a twelfth from the corporate seat (6 Aug). The severity column is a future-state priority read. Statuses are honest: confirmed only where the evidence is there, contested or open where it is not.
| # | Friction | Status | Severity | Evidence across the six sites and the corporate seat |
|---|---|---|---|---|
| F1 | The P&L arrives wrong. One Aleph Excel P&L template serves the estate and matches none of the ledgers the properties run. The template is formula-driven on Sun department codes, so every property examined rebuilds the mapping by hand each month: adding lines, changing codes, re-keying the statistics the ledger does not hold. The defect is template-side, not chart-of-accounts side, because the same failure appears on three different ledgers. Who owns the template file, and whether it can change, is not yet established.The single most consistent finding in the programme, and Aleph's own file to fix. | Confirmed and reframed | High | Accra (Sun): statistics hand-keyed into offline Excel so the ratios calculate; two P&Ls merged. Dar (Sage): the template carries cost centres that do not exist in Sage; ~2 of 5 production days; the source fix never requested of ONOMO in 6+ years. Morocco (Sage): HQ finance keys Sage figures into Excel, bypassing the existing reporting layer. Abha (NetSuite): the format is built for the Sun users, so the majority of the job is manual; the same mismatch reported on the cluster's Sun hotels. Addis (Sun): the incoming finance director is not sure the back-office mapping is correct and is reviewing it. Provenance: developed from one of the African hotels. Dubai (Sun): untested from the GM seat; the bridge runs inside cluster finance, the GM declined to guess at it, and no mapping complaint is recalled. Corporate seat (6 Aug): a corporate-issued, formula-driven results workbook exists with a named owner (whether it is the same file as the P&L template is not established), and Fairmas is dated to replace the Excel reporting for a 17-hotel first wave by about September, ONOMO signed for about year end. |
| F2 | Numbers travel as pictures. Once a number leaves Excel it is carried into the pack as a screenshot or pasted table and stops being data. Any figure that changes after that point means copying the table in again by hand. The friction belongs to whoever builds the deck, not to the property as such: where the deck is built above property level, the defect sits at that tier, and that tier has not yet been seen.Kills any automated read of the pack downstream. | Confirmed and reframed | Medium | Dar: KPI and summary tables in the May pack are pasted images, confirmed on the call and in the artefact. Morocco: tables cross into the regional deck as screenshots. Abha: untested, not absent; the deck is built at regional and unseen. Addis: confirmed by the author of the deck himself, who pastes screenshots of the P&L into PowerPoint. Accra: not raised. Sample packs: one 82-slide pack carries 107 images and no native charts; another carries 37 native charts. Fidelity varies pack to pack. Dubai: a PowerPoint assembled from department slides; image-versus-native not yet measured, the June pack is on file to check. |
| F3 | Manual regional rollup. Wherever more than one property feeds a single review, the consolidation is done by hand: per-hotel sheets pulled into a regional workbook, tables carried into a regional deck. It is the most senior time observed anywhere in the process and the tier where the owner-facing lag accumulates. The load grows: Morocco consolidates four hotels today with three more opening.The tier the programme has seen least of. | Observed at one tier | High | Morocco: the only rollup observed; historic solo consolidation at one week to 10 days per cycle; June's finance-drafted cycle moved the labour rather than removing it. Abha: three Saudi hotels feed a regional deck neither observed nor seen. Dar: one cluster FM hand-builds two countries in the same window. Accra: not applicable. Addis: not applicable today, single property. Sample packs: two regional cluster decks re-aggregated above property level. Dubai: a property-tier consolidation nobody else has: two branded hotels plus a marina and retail estate in one deck, with no regional lane. Corporate seat (6 Aug): the top of the rollup is three individuals with no teams; about 50 forecast files arrive in different currencies with no means to consolidate them, though a US-dollar number is expected; about 21 hours per month per executive as computed on the call (17 hotel sessions plus about four regional; the round figure offered was 20); the stated future is regional VPs owning the review. |
| F4 | Copy forward from a cloned pack. Packs are built by rolling the previous month forward and overwriting it, at every site that builds a pack. The mechanism is confirmed and not itself a defect; the exposure is a value nobody overwrites travelling into the new month, in a document worked on by several hands.The error class the sample set already exhibits. | Confirmed | Low | Accra: last month's document is the shell, updated by several hands; no defect reported. Dar: roll-forward did not arise; the error class is added-lines-break-totals. Morocco: untested. Abha: no written commentary to carry, no prior year. Addis: a sharper mechanism than roll-forward, and a named one. A forecast-versus-actual filter in the workbook: if it is not switched, the comparison silently runs against an un-actualised forecast, so the stale item is the comparator rather than a sentence. Artefact: the Michlifen May pack in the sample set carries a cover reading Avril, a copy-forward defect that reached issue. Dubai: accretion rather than roll-forward: about 20 slides became about 80, additions never removed. Corporate seat (6 Aug): about 90 per cent of hotels struggle to hold the template three months running, and an attempt to machine-read packs across months failed because no two months are identical, down to the page numbers. |
| F5 | The lag after month end: two clocks. The property-to-corporate clock runs close-by-the-5th, submit 6th-7th, reported consistently and in every case self-reported. The owner-facing clock runs to day 10-22, artefact-dated, with the only observed owner event at day 22. The lag sits in the middle tier, not at property close. Any speed claim must say which clock it addresses. At the Dubai cluster the deck runs one clock, owner and corporate in the same email; a separate finance-side P&L submission was not asked about.Both clocks are true; they measure different things. | Split | High | Accra: close 30th-3rd, submit 7th (self-reported); May pack file dated 17 June. Dar: ~5 days, sent ~7th (self-reported); May pack file dated 10 June. Abha: cut-off 25th, processing from the 1st, sent 6th-7th (self-reported). Morocco: June pack presented to the owner asset manager Wednesday 22 July, day 22, observed and dated. Cluster pack V160626 = day 16. Addis: not established, the clock was not asked; a forecast file goes to Aleph twice a month. Dubai: the deck emailed to the owners 15th latest 20th with corporate copied on the same send (self-reported); a separate P&L clock not asked; June pack held. Corporate seat (6 Aug): a handful of BRMs ready by the 10th, the vast majority 15th to 20th; the agreed target is now the 10th; every hotel also files a three-month rolling forecast on the 25th. |
| F6 | The narrative eats the hours, in two forms. Written narrative is heaviest where each HOD populates a section, and the cost there is queue time (the FC waits 1-2 days), not authoring. At the other extreme one site writes no commentary at all, yet the same month absorbs about two hours of verbal review plus HOD explanations plus an owner conversation, none of it written, so none of it reusable, auditable or capable of being rolled up. An hours split, numbers versus words, has not been captured at any site.The product target widens: draft the text somebody wants, and capture the judgement that today leaves no artefact. | Split | High | Accra: 17 blocks, each HOD a section, GM writes the summary; 1-2 days of chasing. Dar: commentary genuinely distributed; every time figure volunteered was data mechanics. Morocco: all commentary, interpretation and harmonisation with one person, who also presents. Abha: written commentary nil and not requested; verbal narrative real and unrecorded. Addis: contradicted deliberately. The GM writes every comment himself and declines to delegate it, because preparing the deck is how he comes to own the figures. The cost he names is in the back office, not the narrative. Dubai: 10 to 20 hours of management time monthly (his estimate); every comment typed by the GM; the author calls the content already known to him. Corporate seat (6 Aug): the same person can end up preparing the presentation, reviewing it and presenting it; about 20 hours a month each at the corporate tier. |
| F7 | The correction loop with corporate. Corporate comes back on the numbers, and where it does, the questions are commercial rather than error correction; answering means returning to the source documents, principally the trial balance. Not universal: one site reports no come-back where figures align, the Dubai cluster has no loop at all, and where a regional tier exists the loop runs inside the region instead. Turnarounds are self-reported everywhere.To be measured from email timestamps, not memory. | Confirmed | Medium | Accra: a numbered list by email, commercial (A&G, flow-through). Dar: 3-4 business-why questions on the P&L, answered from ~4 source documents. Morocco: the loop runs to HQ finance, outside the country manager's line; June tables returned for correction. Abha: no corporate loop reported; corporate accepts where figures align; routing unresolved. Addis: not raised, no correction loop described. Dubai: no loop exists: corporate is copied at the moment the owner is, and nothing comes back. Corporate seat (6 Aug): the loop is real but under-resourced (three individuals, no teams), sometimes runs blind (the deck arrives hours before, or at, the meeting), and a pre-BRM means the same deck is reviewed twice; owner follow-up questions route to corporate even where a regional team exists. |
| F8 | Errors reach the owner. Open in both directions. No site reported in testimony that an error reached an owner, and four described internal catches before release. Against that, the one owner-facing pack we hold and have read line by line contains presentation defects that were issued. The direct question was not put at most sites. The structural constant: the last check before the owner is one named person, never a process.Effective and unscaleable at the same time. | Contested | Medium | Accra: one accuracy issue in two years, caught at corporate. Dar: internal catches described; the May pack artefact contains issued defects. Morocco: no owner-facing error, prevented by the country manager's personal check of every table. Abha: not tested; the Arabic summary has a single author and possibly a single reader. Addis: no owner-facing error reported, but the finance director states there is no way of proving the numbers, so an absence of known errors carries little information. Dubai: at one cluster property a first-page food cost of 10 per cent, a formula error, was caught on sight by the GM and stopped before issue; the author is the only gate. |
| F9 | Owners come back afterwards. Owner follow-up tracks the owner model, not the quality of the pack. Six sites, six pictures: continuous informal contact by preference; calm conditional on green figures; a live professional review; a bottom-line conversation; a monthly board walk-through; and a review that stopped happening once results turned. No single review design survives contact with all six.Depth of review should follow the owner, not only the asset classification. | Split | Low | Accra: the owner visits most Fridays and messages the GM directly, by preference; no minutes are taken. Dar: meetings under 30 minutes all year; the full pack becomes relevant when performance turns. Morocco: formal per-hotel review by an asset manager, observed day 22. Abha: bottom line and bank balance; early purchase-request explanations were onboarding, not a reporting loop. Addis: a monthly physical board meeting; the owner asks nothing repetitive and is content with the deck. Dubai: withdrawn: no review meeting in six to eight months; informal contact carries everything; monthly reviews previously ended near slide 15 of 80. |
| F10 | Volume without analysis. Authors defend the content and would cut little (one author breaks the pattern: the Dubai cluster GM would cut his own deck, and was declined); the one consumer in the programme questions the time spent presenting data from the past. The purest form ships seven or eight sheets of figures with no commentary at all: volume travels while the analysis is held verbally or not at all. Length and readership are separate questions; the packs are read closely enough for a single wrong figure to be caught.Authors defend, consumers cut; weight consumer testimony in the future state. | Confirmed and reframed | High | Accra: 17 blocks + forecast + appendix; the owner rep once called the pack too long; corporate caught one wrong figure among thousands. Dar: nothing in the 29 slides seen as redundant, but the predecessor ran ~50 pages and was cut once already. Morocco: one week to 10 days preparing a presentation on the past, questioned to Aleph leadership before this programme. Abha: 7-8 sheets, zero commentary. Addis: contradicted. 32 slides, not felt as tedious, and expectations described as light beside a fully branded operator's. Sample packs: 29 to 82 slides. Dubai: the extreme case, from the author: about 80 slides grown from about 20, a shortening request declined, about 10 marketing slides of which three or four matter, and a standing doubt that anyone opens it. Corporate seat (6 Aug): the review hours are called "not at all adding a value" (one reviewer's words, generally endorsed by the second); deep-dive content (recipe cost) is mixed into a strategic review; the corporate wish is 25 per cent post-mortem, the rest forward-looking, with the three to five initiatives that move the needle and the bottlenecks to clear. |
| F11 | Deadline-driven error deferral. New since v2, surfaced at one site: where an error is found late in the reporting window, the submission date can win, and the item is carried into the following month rather than resolved. Hedged, no named instance; to be tested elsewhere in the same words. Carries a scaling warning: the manual workaround was described as sustainable only while transaction volumes are low, at a hotel that opened in January.The clock beating the fix, admitted for the first time. | New | Medium | Abha: a 6th-or-7th deadline, and some errors met at the end of that window may be left for next month; problems expected as occupancy grows. Accra: not asked. Dar: no deferral; broken totals are chased down before release, up to two days. Morocco: incorrect June figures were bounced back, not carried forward. Addis: not raised. Dubai: not raised. |
| F12 | Formulas found altered at property level. A corporate-issued results workbook computes the pack's numbers from dumped data; corporate testimony is that its backend formulas are altered at property level so that the end result moves. Steering rather than error, single-witness testimony from the corporate seat, no property named, and untested from the property side. It is a different class from every other item on this register: mapping fixes and automation do not close it; only a locked or centrally regenerated master does, plus a formula-integrity check each cycle.The first testimony of steering rather than error. Verify before it is ever quoted onward. | New, corporate seat | High | Corporate seat (6 Aug): single-witness testimony from one of the two corporate reviewers; the workbook's builder did not speak to it on the call. The motive given was to move the end result; how it is detected today was not described. GM interviews: never surfaced from the property seat; the question was not put. |
One Excel file, owned by Aleph, failing against three ledgers. Correcting it needs no AI, no software and nobody's permission, and it removes the biggest recurring manual cost at every site. It should reach Bani and Jad before build scope is fixed. First question: who owns the file.
Close-to-corporate already runs on a 6-to-7-day rhythm by every account. The owner sees the month on day 10 to 22. Improvements to clock 1 are invisible to an owner living on clock 2; the middle tier (consolidation, commentary, correction, scheduling) is where the days go.
At one extreme, written sections nobody reads; at the other, two hours of verbal judgement that leaves no artefact at all. The engine should draft what deserves writing and record what today evaporates: the explanations, the commitments, the owner's questions.
FC, cluster FM, country manager, cluster DOF, at Addis whoever updates the statistics tab, and at the Dubai cluster the author himself: at every site one named individual stands between a manual pipeline and the owner. It works, it does not scale, it vanishes on leave or departure. Future-state checks must be reproducible by a second person, not just faster for the first.
Morocco's June cycle (finance drafts the tables, the senior person validates and adds judgement) is exactly the operating model the engine needs, emerged organically, and it was asked for unprompted at Accra. The build does not have to sell the workflow, only make the draft instant and right.
Two artefact architectures, six owner models and at least five finance org designs. A future state designed only for property-built decks will not deploy to the Saudi cluster; one designed only for formal reviews will not fit a Friday conversation or an Arabic one-pager. Build the data layer once; let the presentation flex.
Occupancy, ADR, covers and segmentation never pass through any ledger; they are hand-keyed from PMS and POS at every site where the mechanics were seen. A perfect P&L template still leaves this join unmeasured and unautomated. Any pack engine needs both feeds, not one.
The Addis GM draws the line himself: the back office is where the tedium and the error risk sit, and the commentary is how he comes to own the figures before facing the owner. He is the first to say plainly that he does not want the narrative automated. That is a design constraint, not resistance, and it happens to match where the value is anyway.
Revenue, receivables and cash. Explicitly not the expense lines. The only direct materiality signal any owner has given the programme, and it cuts against a pack organised around GOP. Worth testing with the others before the output shape is fixed.
Every author defends their pack; the one consumer interviewed counts the days it costs. Dar has already proven packs can shrink (50 pages to 29 slides). The future state should cut by consumer testimony and exception logic, not by asking authors what to drop.
At the Dubai cluster the deck reaches the owner and corporate in the same email, so nothing can be caught before an owner sees it. Any pre-issue control must run at the property tier there, or it runs nowhere.
The three corporate seats are one person each, spending about 20 hours a month apiece on reviews described from that seat as "not at all adding a value", with a pre-BRM doubling the load. Any future state that adds corporate review work fails on arrival: quality checks must move to source and to machines, and decisions to the regional tier, which is where the leadership says ownership is heading anyway.
The Excel reporting is dated to be replaced for a 17-hotel first wave around September, with ONOMO signed to follow around year end. It is a reporting and staging platform: no narrative, no analytics judgement, no non-financial data. The programme's lane is exactly the half it does not cover, plus the ONOMO portfolio until it lands.
Where stated, properties self-report the 6th or 7th; the artefacts say day 10 to 22; and the receiving end now says most BRMs run the 15th to the 20th, against an agreed target of the 10th. The middle tier is the target, exactly as drawn since v2.
About 80 slides, six to eight months without a review meeting, no feedback and no questions; the author doubts dummy text would be noticed, and the report tells him nothing he has not already approved. The strongest reader-test result in the programme, volunteered from the author's side. His fix is the programme's: a short template of the crucial information everybody needs.
Evidence classes: ARTEFACT a file date or version string we hold · OBSERVED an event watched happen on a known date · SELF-REPORTED interview testimony, recorded faithfully · INFERRED a programme-wide pattern applied to a site that did not state it.
F1 to F12: the twelve items of the friction register on this page; F1 to F11 are shared with the GM interview question sheet, and F12 was added on 6 August from the corporate-seat session.
Abbreviations: A&G administrative and general (an undistributed expense department) · ADR average daily rate · AHD African Hotel Development (best-reading owner name, unconfirmed) · BRM Business Review Meeting · COA chart of accounts · DH Deep Hospitality · DMAIC Define, Measure, Analyse, Improve, Control (the improvement method this programme follows) · DOF Director of Finance · F&B food and beverage · FC Financial Controller · FM finance manager · GM General Manager · GOP gross operating profit · HO / HQ head office / headquarters · HOD head of department · HR human resources · JOC Jewel of the Creek (the Dubai cluster) · KPI key performance indicator · P&L profit and loss statement · PMS property management system · POS point of sale · Q3/Q4 third / fourth calendar quarter · UAE United Arab Emirates · USALI Uniform System of Accounts for the Lodging Industry · VP vice president.