This is a private pitch document prepared for the board of Hotel de France. Access requires a password.

For access, contact Grace Parker directly.

Operations & product · capital-efficient bespoke build

Bespoke software, built for a boutique hotel.

This page covers two distinct software stories. The first is a small portfolio of guest-facing and marketing-facing products — the sleep report delivered at programme departure, the FITVO fitness app refurbished as a free top-of-funnel marketing tool, and the companion app that holds each guest's itinerary, results, a private stay-window messaging channel, and ongoing health record across their stay and beyond (with a Phase 03 partnership extension built with Medilab). The second is a bespoke replacement for our operational SaaS stack — property management, restaurant, spa booking, accounting, and the rest — which the hotel currently rents at roughly £100,000 per year. Both stories sit more naturally alongside the renovation work than the programme build, and both depend on the same premise: AI-assisted coding has made small, focused, well-specified software dramatically cheaper to build than the SaaS subscriptions we currently pay for.

Treat the numbers here as working estimates. They are grounded in a realistic remote developer market rate, but software projects routinely slip on scope, and the build cost ranges carry a 40% contingency precisely because small bespoke projects underrun about as often as they overrun.

Story one of two · the product portfolio

Three products, three distinct roles.

Guest- and marketing-facing software. Programme deliverables, guest companion, and acquisition tools, sitting alongside the operational stack story below.

The hotel needs a small product layer alongside the operational stack. Three products, each playing a different role in the funnel and the guest relationship. FITVO sits at the top of the funnel as a free fitness app the hotel uses as a soft marketing surface. The sleep report sits inside the programme as a substantiating deliverable, generated automatically from the Eight Sleep data collected across the stay. The companion app sits in the guest's hand from the moment they book through to long after they leave — itinerary and push reminders during the stay, wayfinding, results delivery, and at Phase 03 the addition of a Medilab-partnered post-stay layer that keeps the bloodwork, the retest schedules, and the ongoing nutrition guidance alive in one place rather than two.

Together they describe a product portfolio rather than a single app: FITVO acquires, the programme converts, and the companion app delivers the experience and retains the relationship. The three products are at different stages of maturity — and the page treats them honestly as such.

Product 01 · in build for Phase 02

The sleep report.

A personalised sleep report delivered to every programme guest at departure, generated directly from the Eight Sleep data collected across their stay — heart-rate variability, sleep stages, breathing rate, snoring, all measured contact-free from sensors woven into the cover. Guests who already wear an Oura ring, an Apple Watch, a Garmin, a WHOOP or a Fitbit can connect it before arrival via the Terra health-data platform; their day-time activity and habitual baseline then merges into the report alongside the on-property nights. Guests who don't track themselves still get a complete report from the Eight Sleep alone.

The product matters because it is auto-generated rather than hand-written. Every programme guest receives a report of the same quality, regardless of Health Coach availability that week. Weekend guests get a two-night report; Pause guests a five-night report; Week and View guests a continuously-updated report across the full stay, reviewed with the Health Coach at each coaching session. Without the auto-generated layer, sleep reporting becomes a Health Coach time-cost that scales linearly with bookings; with it, the marginal report costs nothing to produce.

Build status. Already scoped, costed, and budgeted as a Phase 02 launch capex line at £5,000 for the MVP build (visible on the Forecast page capex breakdown). Built on top of Terra as the data integration layer — Terra handles the wearable connectors, our software produces the reporting layer on top. Iterative refinement after launch sits within the operational software team's retainer rather than carrying its own ongoing cost.

Product 02 · existing app, refurbishment in scope

FITVO — the free fitness app.

FITVO (fitvoapp.com) is a built-but-unlaunched fitness app that the family owns. The product is essentially finished — including a library of around 200 exercise videos already produced — but it never went to market. The proposal is to refurbish it and launch it as a free, branded marketing surface for the Long Hotel, keeping the FITVO name and visual identity intact rather than rebranding it as a Long Hotel app. The brand has further commercial opportunities of its own, and folding it into the hotel identity would foreclose those rather than add to them.

The starting position matters. Because the core build and the content library already exist as sunk costs, the hotel is not funding a fitness app from scratch — it is funding the modernisation needed to ship an asset that's already been paid for. That changes the economics of the decision considerably.

What the refurbishment covers. Three buckets of work. First, package modernisation — removing dependencies that are no longer maintained, upgrading what stays, and re-establishing a current build pipeline. Second, screen-size and form-factor work — adapting layouts for the iPhone 16/17 generation, the larger Pro Max displays, the Dynamic Island, and current Android device variants. Third, design tidy — particularly the icon set, plus a small number of UX changes targeted at reducing friction at the highest-drop-off points (sign-in, first session, recurring engagement). This is refurbishment, not a rebuild.

Marketing role. A free fitness app puts the FITVO brand in front of an audience interested in fitness and wellness — broadly the same demographic the Long Hotel programmes target. The hotel benefits in three ways: brand presence with a relevant audience at no per-impression cost; a soft funnel into the programme proposition through occasional in-app content, links, and seasonal promotion; and a content surface that gives the hotel's content team a place to publish short-form fitness pieces that wouldn't fit elsewhere. None of this depends on FITVO becoming commercially significant in its own right — the marketing effect operates regardless.

Indicative build estimate. Approximately 2–3 developer-months of work at the same effective loaded rate as the operational team below — £2,900 per productive developer-month, totalling £8,000–£12,000 all-in including contingency. Could trend higher if the package debt is severe; the two-week paid trial pattern proposed elsewhere on this page applies cleanly here as a way to validate scope before committing the full budget. Best done by the same remote development team that handles the operational stack, sequenced after the first one or two operational modules ship to confirm the team's productivity. Given that the core product and the 200-video library are already built, the £8,000–£12,000 figure represents the full marginal cost of bringing this asset to market.

Product 03 · Phase 02 in-stay MVP, Phase 03 Medilab extension

The companion app.

One app, downloaded by the guest at booking confirmation, used continuously from then onwards — across the stay, into the post-stay window, and as a long-term home for their health record. The product evolves in two phases. Phase 02 ships the in-stay MVP: itinerary, push notifications, wayfinding, and a results inbox where bloodwork, scans, and the sleep report land as they come through during the stay. Phase 03, built in partnership with Medilab, extends the same app with the post-stay layer — retest schedules, nutrition and supplement guidance grounded in the actual bloodwork, longitudinal tracking of results across visits, HealthKit and Android-equivalent integration. Same login, same data, same product. The in-stay surface delivers the experience; the post-stay extension keeps the relationship alive.

Phase 02 — in-stay MVP

What ships at Phase 02. Three jobs, all of them practical rather than ambitious. First, the app holds the guest's full personalised itinerary — every consultation, treatment, class, coaching session, meal sitting, and free-time block scheduled across their stay, pulled live from the operational booking system. Second, it sends push notifications a sensible interval before each appointment with the time, the location on the property, and brief directions for getting there from wherever they were last. Third, it acts as the place where their results land — bloodwork as Medilab returns it during the stay, DEXA scans, Styku body composition images, the in-progress sleep report, any imaging or clinical write-ups generated as the programme runs. The guest leaves with everything in one place rather than scattered across emails, paper printouts, and reception's filing cabinet.

Why this matters operationally. A Long View guest has fifty to seventy scheduled touchpoints across their fourteen-night stay — clinical consultations, Ayurvedic treatments, fitness sessions, coaching, meals, group activities, free blocks. A Long Week guest has thirty to forty. Without a structured itinerary in their hand, the guest experience is a continuous low-level scramble: paper schedules, reception phone calls to confirm timings, missed sessions, late arrivals, awkward "where am I supposed to be?" interactions with staff. The current Hotel de France experience already runs this way; the Long Hotel programme experience cannot. Push reminders shift the burden of remembering from guest-and-staff to software, and ten or fifteen avoided "I'm sorry, where is the consultation room?" reception phone calls a day across a busy season is a meaningful operational saving.

Why this matters experientially. The companion app is the single most-used piece of software the guest interacts with across their visit. Every appointment they remember on time, every treatment room they find without asking, every result they see arrive in real-time, builds the impression of a programme that is operationally serious — that the property is a place where things are tracked carefully and delivered intentionally rather than improvised. The app is therefore brand-grade rather than utility-grade: the visual identity, the tone of the push notifications, the typography and animation of the itinerary view all matter, because they are the product the guest holds in their hand most often.

Wayfinding, specifically. The Hotel de France site has roughly thirty named spaces a guest might be sent to across a programme — the clinical wing's consultation rooms, the Healthhaus gym and treatment rooms, the spa, the restaurant, the gardens, the Ayurvedic suite, group-class spaces, the lobby and reception. New guests cannot navigate that from memory in the first three days, and the property is large enough that "go through reception, past the bar, second door on the left after the courtyard" doesn't read on a paper schedule. The app carries simple floor plans with each booking — an arrow from the guest's current location to the destination, a brief written direction, and a fallback "tap to call reception" if they get genuinely lost. Not Google Maps; something closer to the way airport apps show you the gate.

The Stay Channel — guest connection within the app

What it is. A private group messaging channel built into the companion app, connecting guests who have overlapping reservations during a shared stay window. Access is provisioned automatically at check-in and revoked at check-out — tied directly to the reservation database, with no separate sign-up or account creation required. Identity within the channel is first name only: no profile photos, no persistent profile pages, no member directory. Guests are not visible in the channel until they choose to post; lurking is the default and the correct default.

How it works in practice. A single channel per stay window, not split by programme type — the intention is cross-pollination between longevity guests, Ayurvedic guests, and general wellness guests rather than siloed groups. Staff post optional activity invitations through lightweight polls: a kitchen herb walk at 10 a.m., three spots on the sunset breathwork session, the morning sea swim group departing at 6:45. The voice is conversational rather than programmatic — "Chef is doing a kitchen herbs walk tomorrow at 10 — three spots open, tap to join" rather than a formal activity listing. Guests can post their own informal invitations with the same ease ("anyone want to do morning yoga?"), which simultaneously reduces social friction for solo guests and gives the team a live read of what the current group wants more of. A pre-arrival post from the hotel lands in the channel the day before the earliest check-in in a given window, so no guest opens the app to an empty channel on day one.

What it deliberately is not. Not a matching algorithm. Not a long-term community platform. Not a profile-based social network. No likes, no reactions, no read receipts, no active-now indicators. No message history persists beyond the stay. No direct messaging between guests — if two people want to continue a conversation privately, they exchange contact details in person. Default notifications are digest-style (one or two summaries per day) to protect signal-to-noise ratio and prevent the channel being muted within 24 hours; real-time alerts are opt-in for activity posts only. The feature exists to reduce the friction of casual in-person encounters during a stay — particularly for solo guests and those on differing schedules — and nothing more. Post-stay community, if it develops, is addressed separately and manually in early stages.

Moderation. A lightweight in-app reporting flow routes to the duty manager. The staff account holds moderator privileges. No elaborate infrastructure is needed given the small, vetted, reservation-authenticated user base — the population is never anonymous.

Build approach. The channel is built natively within the companion app, sharing the same backend as reservation data. Real-time messaging infrastructure is provided via an established third-party service — Stream, Sendbird, or PubNub — embedded into the app. Authentication, identity logic, channel provisioning, and access control are handled by the hotel's own systems and tied directly to the reservation database. This avoids the reconciliation problems of a third-party chat tool while keeping build scope manageable. Guest behaviour and the staff voice are validated during a soft-launch period with early guests before press visits begin. Ongoing messaging-service cost is low — at the volumes this hotel runs, Stream or equivalent sits at approximately £1,000–£2,000 per year — and is absorbed into the operational running cost rather than carrying its own capital line.

Phase 02 build cost and timing. Built as a React Native cross-platform app (iOS and Android from one codebase) with a thin server layer reading from the operational booking system. The MVP scope is itinerary view, push notification engine, simple wayfinding with floor plans, results inbox with PDF and image rendering, basic profile and preferences. At the operational team's effective loaded rate of £2,900 per productive developer-month, the MVP scope — itinerary, push notifications, wayfinding, results inbox, and the Stay Channel with its messaging-service integration — is approximately 4–5 productive developer-months — £15,000–£20,000 all-in including contingency, assuming the operational booking system exposes the schedule and reservation data cleanly (which the Phase 1 SaaS replacement build below would do natively). The Stay Channel integration (Stream or equivalent embedded into the app, provisioning tied to the reservation database) adds roughly one developer-month to the base MVP scope but does not require its own capex line beyond that — the ongoing messaging-service subscription (£1,000–£2,000 per year) is an operational rather than capital cost. Timing is sequenced behind the operational stack's booking module so the app has something to read from; the practical sequence is operational booking module ships first, the companion app builds against it across the following two to three months, both go live at Phase 02 opening. The two-week paid trial pattern proposed elsewhere on this page applies here cleanly to validate developer fit. If the operational stack story is deferred and the bespoke booking module is not built, the cost shifts materially upward — the app then needs to integrate against whatever SaaS booking systems are kept (each with its own API, some without one), and a realistic range becomes £28,000–£36,000 with some recurring integration maintenance whenever a vendor changes their API.

Phase 03 — Medilab post-stay extension

What Phase 03 adds. The Phase 03 extension layers the post-stay clinical surface onto the existing app. Bloodwork results held alongside DEXA and Styku scan data; HealthKit (and the Android equivalents) integration to pull in activity and sleep continuously after the stay; nutrition guidance and supplement recommendations grounded in what the actual bloodwork shows; retest schedules calibrated per marker — ferritin moves over months and warrants different cadence to ApoB or Lp(a). Same app the guest has been using throughout their stay, with new tabs and new functionality enabled when they engage Medilab as their post-stay clinical provider. No second download, no migration of data, no friction at the moment of checkout where retention matters most.

Why combining the two phases in one app matters. Splitting the in-stay and post-stay products into two separate apps would mean two App Store entries, two builds, two sets of push notification certificates, and a moment at checkout where the guest is asked to download a different app to keep using the thing they have been using all week. Every download step loses users, and the post-stay retention case is precisely the moment the proposition can least afford the friction. One app, one login, one data layer, two phases of functionality is the right architecture both for the guest experience and for the partnership conversation with Medilab — one product, one regulatory wrapper, one IP question.

Why the post-stay layer matters strategically. Bloodwork-at-every-tier is the structural credibility anchor of the longevity proposition (see the Market page for context). A guest who returns home with results sitting in a notes app loses most of the value within a few weeks; a guest who returns home with their results inside an app they are already using daily — recommendations attached, retest schedule visible, longitudinal trends accumulating — keeps the proposition's value compounding past the stay. The same logic that drives the Long View's posted three-month follow-up panel applies here at lower clinical complexity: the relationship has to extend past the stay or the proposition is half-delivered.

The Medilab partnership shape. The intended structure for the Phase 03 extension is a partnership with Medilab — already the hotel's bloodwork provider through their four-tier panel offering and an established regulated lab in Jersey. Medilab brings the clinical credentials, the regulatory framework that the post-stay clinical surface needs to sit inside, and the existing patient-data infrastructure that already handles bloodwork results securely. The hotel brings the guest data flow (programme participants are already Medilab patients via the bloodwork tiers), the in-stay product surface that the Phase 03 extension layers onto, and the ongoing relationship that turns episodic test results into a continuous companion. The partnership terms are still to be agreed; the working assumption is that the regulatory burden — what defines the product as informational versus medical advice, and what the appropriate UK MHRA / FDA-equivalent posture looks like — sits on Medilab's side of the partnership rather than the hotel's. That is the single most important practical reason to do the post-stay extension as a partnership rather than a hotel-led build, and it materially de-risks the product compared to the hotel attempting it alone.

Open questions, honestly. Three things remain to be defined for the Phase 03 extension. First — the commercial structure of the partnership. Three plausible shapes: hotel pays Medilab to build with Medilab owning the IP (lowest hotel exposure, lowest upside); hotel and Medilab co-build and co-own with shared capex by skill contribution (medium exposure, shared upside); Medilab builds and owns, hotel licenses access for guests (lowest hotel capex, no IP upside). The right shape depends partly on Medilab's appetite — whether they see this as a Medilab product with the hotel as a launch customer, or as a genuinely shared venture. Second — the build cost. At the same loaded rate (£2,900 per productive developer-month) the Phase 03 extension is roughly 12–20 developer-months for the full clinical-content, retest-scheduling, longitudinal-data and regulatory-wrapper scope — £50,000–£90,000 at full Phase 03 build cost, including contingency. How that is split between the partners depends on the structure agreed; a hotel-pays-Medilab-builds shape puts the full figure on the hotel side, a co-build shape splits it by skill contribution, and a Medilab-led shape puts most of it on Medilab. Third — the data ownership question. Whose system of record holds the guest's bloodwork in five years? The cleanest answer for the guest is "one place, that I trust" — and Medilab's existing patient-data infrastructure is the natural home, with the companion app as the guest-facing surface on top.

Recommended treatment. Phase 02 ships the in-stay MVP as a committed hotel-built product on the timeline above. Phase 02 also runs the partnership-definition workstream alongside the build — six to twelve months of negotiating and contractualising the Phase 03 extension terms with Medilab (commercial structure, build allocation, IP ownership, data architecture, the regulatory wrapper). Once those are agreed, the Phase 03 build itself can proceed with the regulatory and clinical-sign-off costs sitting where they should — on the regulated party. The in-stay MVP and the FITVO refurbishment, alongside the sleep-report build, are the three committed Phase 02 product spends; the Phase 03 Medilab extension is committed once the partnership shape is defined.

What the app deliberately is not. Not a full guest experience platform, not a content surface, not a persistent community platform, not a chat-with-the-clinician channel. The temptation with an app the guest opens fifteen times a day during their stay — and continues opening weekly afterwards — is to load it with adjacent features. The discipline is to keep it focused on the jobs that earn its place: in-stay, schedule and reminders and results; post-stay, bloodwork tracking and retest schedules and grounded recommendations. Resist scope expansion until the core has been live for at least one full season. The app's quality is judged on whether the guest never misses a session, always knows where they're going, and continues to find their bloodwork useful six months after they leave — not on feature breadth.

Combined committed product capex (Phase 02): £28,000–£37,000 — being £5,000 for the sleep-report MVP (already in the Phase 02 forecast), £8,000–£12,000 for the FITVO refurbishment, and £15,000–£20,000 for the in-stay MVP of the companion app (including the Stay Channel messaging integration). The Phase 03 Medilab extension to the same companion app sits outside this envelope, deferred until the partnership is contractualised; the hotel's eventual capex contribution depends on the commercial structure agreed with Medilab and is currently not estimated above zero pending those terms.

Story two of two · the operational stack

Replacing £100,000 a year of rented SaaS.

From here, the page covers the second software story — the bespoke replacement for the hotel's operational SaaS stack. Distinct from the products above, larger in budget, and addressing a different operational problem. The team, the modules, the economics, and the risks are set out section by section below.

Operations · the opportunity

£100,000 a year in feature breadth we don't use.

The hotel currently pays roughly £100,000 per year across property management, restaurant reservations and ordering, guest communications, staff scheduling, accounting integrations and assorted smaller tools. We are evaluating spa booking software, which would add to that total. These systems are built for a broad market — mid-market hotels, resort groups, boutique chains — and price accordingly. A meaningful share of the annual fee pays for feature breadth we will never use: multi-property management, complex loyalty schemes, third-party integrations for vendors we don't work with, reporting sophistication beyond anything a boutique family run hotel actually needs.

The premise for a bespoke build is not to match SaaS feature for feature — it's to build only what this hotel uses, designed around how we actually operate, and save the rest. A lean build of the modules below would cover our full operational requirement in a way that feels genuinely designed for this property rather than adapted from a generic template.

Three things make this more viable in 2026 than it was five years ago. First, AI-assisted development has materially reduced the time cost of shipping small-to-medium operational software — the marginal hour of a competent developer produces more working code than it used to. Second, developers in third-world countries are just as good, if not better, than developers that can be hired more locally and remote working has become very normalised. Third, the integrations most hotels need most urgently — payment processing, accounting, email delivery — are now available as well-documented third-party APIs (Braintree, Xero, Postmark), so we build on top rather than reinventing the plumbing.

Operations · the team

Three people, $5,500 a month.

The team is deliberately small and fully remote. Two developers in parallel, one QA engineer reviewing and testing everything that ships. This is the minimum viable structure that avoids single-developer bus-factor risk while staying affordable at the boutique-hotel scale.

Role Monthly rate Annual Why
Developer × 2
Remote freelance, full-stack
$2,000 each ($4,000 total)
≈ £3,200/mo
≈ £38,400 Two developers working in parallel halves calendar time and creates a minimum viable code-review discipline.
QA Engineer × 1
Remote, testing & release
$1,500/mo
≈ £1,200/mo
≈ £14,400 A dedicated QA is what separates a bespoke build that's robust enough to run a hotel from a bespoke build that isn't. Catches regressions, forces disciplined release practice.
Total team cost $5,500/mo
≈ £4,400/mo
≈ £52,800/yr Running rate during the build phase; reduced to a retainer once launched.

FX conversion at $1 = £0.80 (mid-April 2026). Effective productive throughput ~1.5 developer-months of shipped work per calendar month, accounting for QA integration, code review and rework. Gives an effective loaded rate of roughly £2,900 per productive developer-month, used in the module costing below.

Currency risk note: Developer team costs are denominated in USD at current exchange rates (~$1 = £0.80); a material sterling depreciation would extend the payback period, though the five-year net benefit remains strongly positive in all reasonable FX scenarios.

Operations · scope & costs

Twelve modules, 22 developer-months.

The twelve modules below cover the full operational scope of the hotel. Each is sized in developer-months — one developer working for one calendar month — at the effective loaded rate of ~£2,900 per developer-month. The base total is £63,800; with a 40% contingency (standard for a small bespoke build without dedicated product management), the realistic budget sits at £89,000–£90,000. The build phase runs approximately 14–15 months at two-developer parallelism; in practice, 17–21 months is realistic once dependencies between modules and iterative sign-off on each are factored in.

# Module Dev-months Cost (base) With contingency What it does
01 Core infrastructure & auth 1.5 £4,350 £6,090 User accounts, roles, database schema, hosting, authentication, audit log. The foundation every other module depends on.
02 Property Management System
the spine — rooms, rates, guests
6 £17,400 £24,360 Rooms, rate plans, availability calendar, guest records, check-in/check-out, folio, housekeeping status. The single largest build and the operational backbone.
03 Direct booking engine 2 £5,800 £8,120 Website-embedded booking widget. Rate display, availability check, guest details, Braintree checkout handoff. Cuts OTA commission on direct bookings.
04 Braintree (PayPal) payment integration
Stripe does not operate in Jersey
0.5 £1,450 £2,030 Integration with Braintree's existing API for card and PayPal payments. Not building payment processing — PCI compliance stays with Braintree, we build only the integration layer.
05 Restaurant POS + reservations 3 £8,700 £12,180 Table map, reservation calendar, iPad order-taking, kitchen printer integration, basic menu management. Lean — not trying to match Lightspeed's feature breadth.
06 Spa booking & therapist rota 2 £5,800 £8,120 Treatment menu, therapist calendar, room assignments, programme guest integration (pulls from PMS), Postmark confirmation emails.
07 Housekeeping & maintenance app 1 £2,900 £4,060 iPad/phone interface for housekeepers: room status (dirty/clean/inspected), maintenance tickets, simple checklists.
08 Staff scheduling & rota 1.5 £4,350 £6,090 Employee records, shift scheduling, time-off requests, holiday tracking. Designed for a small-team boutique operation.
09 CRM & Postmark email automation 1.5 £4,350 £6,090 Guest communications engine. Pre-arrival sequences, post-stay surveys, marketing list segmentation. Uses the existing Postmark integration as the delivery layer.
10 Accounting integration (Xero) 1 £2,900 £4,060 Nightly push of folios, revenue allocations, supplier invoices to Xero via its API. Single source of truth for accounting.
11 Wi-Fi captive portal 0.5 £1,450 £2,030 Guest Wi-Fi splash page with terms, PMS room-number validation, auto-expiry at check-out.
12 Reporting & dashboard 1.5 £4,350 £6,090 Occupancy, ADR, RevPAR, F&B covers, spa utilisation, programme bookings. Read-only view for management.
Total — Phase 1 scope 22.0 £63,800 £89,320 14–15 months calendar at two-dev parallel; 17–21 months realistic with dependencies.

Channel manager integration deliberately deferred. Connecting directly to Booking.com, Expedia and the major OTAs is the single most complex integration on the landscape — rate parity, availability sync, and constant API evolution. A first build would instead keep paying an existing channel manager SaaS (SiteMinder or STAAH at roughly £2,000–£5,000 a year) and focus internal development energy on the modules where bespoke gives us real operational leverage. If and when the Phase 1 stack proves itself after 12–18 months in production, Phase 2 could replace the SaaS channel manager with a bespoke integration at an estimated 3 additional developer-months (~£8,700 base / £12,180 with contingency).

Operations · economics

Payback inside a year and a half.

Once built, the system runs on a retainer of one developer part-time (for fixes and iterative improvements), hosting, Postmark, and the retained channel manager SaaS. The saving against the current £100,000 annual SaaS spend is meaningful.

Line item Current After Phase 1 build Saving
SaaS stack (PMS, POS, spa, CRM, comms, etc.) £100,000/yr £100,000/yr
Developer retainer (1 dev × 12mo @ $2k/mo) £19,200/yr (£19,200)
Hosting & infrastructure £5,000/yr (£5,000)
Postmark (email delivery) £1,000/yr (£1,000)
Channel manager SaaS (retained) (in £100k) £3,500/yr (£3,500)
Total £100,000/yr £28,700/yr £71,300/yr
Scenario Build cost Annual saving Payback period 5-year net benefit
Base estimate £63,800 £71,300 ~0.9 years £292,700
With 40% contingency £89,320 £71,300 ~1.3 years £267,180
If build overruns by 100% £127,600 £71,300 ~1.8 years £228,900

Even in the worst-case build overrun scenario, the five-year net benefit remains comfortably above £200,000. That is not the most important number on this page — the most important number is that we stop paying £100k a year for software that is largely built for a different kind of hotel than ours.

Operations · honest risks

The reasons this might not work.

A pitch for bespoke software that doesn't name its risks isn't a pitch, it's a pipe dream. The case above depends on several things being true, and they aren't automatically true. The four risks below are real.

Risk · delivery discipline

Distributed-team management needs an owner

Small bespoke software projects most often fail not because the developers can't code, but because no one is writing clear specifications, prioritising ruthlessly, reviewing pull requests, and saying "no" to scope creep. Those are product-manager and tech-lead jobs. On a two-developer team there is no one playing them by default, and the role has to be explicitly assigned to someone internally. This is material time, probably 10–15 hours per week during the build, on top of whatever else that person is already running. The QA Engineer helps on quality but does not replace the product discipline.

Mitigation: Explicit module-by-module sign-offs, weekly review cadence, written acceptance criteria before work starts on each module.

Operational risk without a vendor to call

If the booking system goes down on a Saturday check-in, there's no support line — just the developer retainer. A good retainer can produce a fix within hours, but "within hours" on a busy Saturday is still hours of zero check-ins. Mature SaaS systems have 24/7 on-call support, SLA guarantees, and redundancy we would not replicate.

Mitigation: Developer-on-retainer SLA with guaranteed response times, clear escalation path, and a documented manual fallback process for every critical system failure (paper check-in forms, manual card payment via Braintree portal, etc.).

The "we don't need feature breadth" premise cuts both ways

We genuinely don't need half of what our current SaaS provides — but we also haven't catalogued which half. Some features that feel superfluous today turn out to matter the first time we try to handle a specific guest scenario. Building lean is the right starting principle; it doesn't remove the need to carefully audit every current SaaS tool before replacing it.

Mitigation: Full feature audit of the current stack before build begins. Document which features are in-scope, which are deferred, and which are deliberately dropped.

Scope creep

"Could we also add…" is the most expensive sentence in software. The module list above holds at 22 developer-months only if the scope holds. Every additional "small" feature request tends to be 2–4 developer-weeks once actually built, tested and integrated.

Mitigation: Scope frozen at start of each module. Change requests batched into an explicit Phase 2 backlog, with separate budget and sign-off.

None of these risks are disqualifying. All of them are manageable — but only if the management is genuine rather than notional. The honest read is that this is a project worth doing only if someone internally can genuinely commit 10–15 hours a week to it through an 18-month build. If that time genuinely is not available from anyone on the team, the project is significantly more likely to drag, overrun, and disappoint. In that case the right answer is to keep paying the £100k SaaS bill.

Operations · optional upside

Other Jersey hotels pay the same £100k.

Once the Phase 1 system has been running the hotel for a year or two — long enough to be genuinely robust, not just working — there is an optional commercial path worth naming lightly. Jersey has dozens of small and mid-sized independent hotels paying similar annual sums for software that wasn't built for them. A Jersey-built, boutique-hotel-scaled system carries a small but genuine commercial story: a Jersey operator who actually runs a hotel selling to Jersey operators who actually run hotels, at a materially lower annual price than the incumbents.

The numbers scale straightforwardly. If the system were licensed to ten other small Jersey hotels at, say, £30,000 per year each — roughly a third of what they currently pay — the resulting £300,000 per year would fund a larger development team, pay for proper enterprise-grade infrastructure, and still deliver a meaningful saving to each customer hotel.

This is genuinely optional. It should not be factored into the Phase 1 investment case. Multi-tenant architecture, customer support, sales, and the commercial management of a SaaS business are all separate workstreams that would need to be sized and scoped on their own terms. But a well-built Phase 1 does not preclude it — and if the internal build proves itself, the door to that business stays open.

Operations · decision gate

What needs to be true to proceed.

Three conditions should be met before Phase 1 begins in earnest.

  1. Internal time commitment confirmed. 10–15 hours per week across 18 months. This is not a nominal commitment; it is the single most important precondition.
  2. Full SaaS feature audit completed. Every current subscription catalogued; every feature classified as in-scope, deferred to Phase 2, or deliberately dropped. Ideally signed off by whoever in the team uses each tool daily.
  3. Two-week paid trial with the developer team on a small, contained first module — most likely the Wi-Fi captive portal or the housekeeping app, both being low-complexity, well-contained deliverables. This validates the team's productivity, communication, and code quality at realistic scale before committing to the full build.

If those three conditions are comfortably met, Phase 1 is a reasonable investment. If any of them are in doubt, the honest answer is to defer — the £100k annual SaaS spend is well understood, predictable, and supported. The case for replacing it is strong, but it depends on execution that has not yet been tested.