Business plan and operating model · September 2026 · proposal, not a forecast
A subscription and on-demand concierge for the Iranian diaspora. The child abroad pays in hard currency; the parent, the apartment and the paperwork in Iran get looked after. Revenue never touches an Iranian bank.
Ten lines, then the honest caveats.
A Canadian company that looks after what a migrant left behind in Iran: an ageing parent, an empty apartment, a pension nobody is chasing, a title deed, a grave. Trained care managers in Tehran do the work. The customer abroad sees photographs, receipts and reports in a dashboard, in Persian and English.
Somewhere between six and ten million Iranians live abroad, most of them in high-income countries, and the generation that emigrated in the 1980s and 1990s now has parents in their seventies. The alternative to this service is a cousin doing favours — unschedulable, unauditable, and socially expensive to ask twice.
The payer and the recipient sit in different countries. Billing happens entirely in Canada on standard card rails, so the business needs no Iranian payment gateway, no rial pricing and no insurance integration — the three things that stall most Iran-facing consumer ideas.
Six sentences a diaspora child actually says. Only the first four are elder care — which is the point.
Every one of these is a job someone would pay to have taken off them. The naming system in section 18 is built directly from this list — categories and services are named the way the customer already talks, not the way an operations chart would.
The strongest argument in favour of this business is not the market size. It is the geography of the money.
Order-of-magnitude sizing, an honest note on how wide the estimates are, and a reframe that changes the denominator.
Source: Iranian official reporting as cited in the source feasibility draft. The 15–20% share assumption is unvalidated and carries the most weight of any figure in this plan. A field check against it is the cheapest high-value research available.
30–50, parent 70+ in Tehran or a provincial city, visits once every year or two. Buys a subscription. Highest emotional urgency, hardest trust barrier.
Owns one to three units or inherited land. Wants evidence, not care. Buys per door, churns least, and never calls at 3am.
Needs a document, a certificate, a legalised translation, a pension chased. Buys once, urgently, and can be acquired on search intent alone.
Flies back every year or two. Seasonal, high margin, zero liability, and the cheapest route to a first transaction with a stranger.
A parent has died; an estate is unresolved across borders. Predictable follow-on from end-of-life service, and the highest value per matter.
Born abroad, reads English more comfortably than Persian, inherits the problem and the relationship. The reason every report ships bilingual.
There is no direct competitor in this corridor. That is not as good news as it sounds.
| Who they are | What they prove | Where they leave a gap |
|---|---|---|
| Nila Elder Care UK to India | That a diaspora child will pay a monthly subscription for a stranger to visit their parent, and that tiered pricing works in this category. | India has domestic payment rails, insurance products and a large formal home-care sector. None of that exists in the Iran corridor, so the operating model has to be rebuilt, not copied. |
| Propdial Indian NRI property management | Roughly 70% of its customers are non-resident owners managing property remotely. The closest thing to proven demand anywhere in this plan. | Property only. No care relationship, so no emotional switching cost and no route into the higher-value family jobs. |
| The cousin the actual incumbent | That the job is already being done, for free, by family. Demand is not in question; the willingness to pay for it is. | A favour cannot be scheduled, audited, escalated or repeated without social cost. Sell relief for the cousin, never replacement. |
| Informal Tehran fixers WhatsApp, word of mouth | That people already pay individuals in Iran to run errands and watch property. | No contract, no receipts, no recourse, no continuity if the person disappears. Every structural advantage here is a direct answer to that. |
| Persian-speaking professionals abroad lawyers, accountants, realtors | That the paperwork jobs have a real willingness to pay. | They handle one matter and stop. They are better treated as the first acquisition channel than as rivals. |
One hundred and twenty-nine services. Nine jobs a customer would recognise. Colours are how each one gets delivered.
Bar length is the number of services in that job. Teal is a general errand any vetted agent can run, amber is scheduled work by the named care manager, red needs a qualification or carries real liability. Tap a job for its description and its full service list.
Nine capabilities that are not services so much as the reason the services are worth paying for. Without them you are selling promises; with them you are selling evidence.
Two ways through it. The tree groups every service by domain, category and delivery class so you can explore it the way it is actually organised; the table flattens it for searching and comparing. Customer-facing name first, operational description second.
| Name the customer sees | What it actually is | Class | Build step | Caveat |
|---|
The separate catalogue of service ideas has been folded in. Eighteen of its entries were genuine gaps and are now named services; the rest were already covered, in several cases because one service in this catalogue contains a dozen of that list's line items.
| Grouping in that list | Where it lands here | What was added |
|---|---|---|
| General services 21 items | Split across calls I can't make, queues I can't stand in, gifts I can't hand over and homecomings I can't prepare. Companion items map to the accompaniment services; pickup and delivery to mail, documents and errands. | Multi-task visit, open hourly assistance, repair drop-off and collection, a driver for the elder's own trips, accompanied appearance where she must attend in person, banking accompaniment, terminal and gate assistance for a travelling parent. |
| Technology support 14 items | Four services, not fourteen: one-off setup, recurring coaching, home internet and network, and digital errands. Smart TV, app installation, file transfer and printing are tasks inside those, not separate products with separate prices. | Home internet and network, recurring digital literacy coaching, digital errands (transfer, backup, printing, scanning). |
| Online and administrative support 9 items | Portal work becomes two services, deliberately separated by risk: using an account and creating one. | Portal and online form assistance; account and credential setup, with a standing rule that credentials are never retained. |
| Service coordination 8 items | Already covered, and worth being explicit about why: finding a technician, getting quotes, scheduling, supervising, following up and reporting back are the six steps inside repair coordination. Pricing them separately would let a customer buy half a service. | Nothing new. Urgent same-day trade dispatch was added separately, because urgency is a standby commitment rather than a coordination step. |
| Healthcare and caregiving 9 items | The largest real gap in the original catalogue. Medication management, home nursing, physiotherapy and occupational therapy were already present; the rest were not. | Doctor at home, at-home lab sample collection, post-discharge care package, professional home attendant with personal care. |
| Professional technical services 8 items | One vetted trades panel rather than eight products: electrician, plumber, appliance, HVAC and boiler, computer and network, locksmith. The customer buys supervision and two quotes, not a trade. | The trades are now named explicitly in the repair service, and urgent dispatch covers the out-of-hours case. |
| Legal and financial 5 items | Power of attorney, tax work, court monitoring and accountancy were already in papers I can't sign, all via licensed partners. | A brokered consultation with a Persian-speaking lawyer, accountant or tax specialist — the product for a family that needs an answer before it needs a matter opened. |
| Family care services 14 items | Mostly packaging rather than services: weekly visits, monthly plans, a dedicated agent, visit reports and photo updates are tier features, and emergency assistance is a whole category. Hourly assistance and multi-task visits were the two real omissions. | Both added above, plus a new pricing rule letting any quota be drawn down as plain agent hours. |
Three delivery classes, one protocol, and a boundary that is written into the customer contract on day one.
Errands ending in a receipt or a photograph. Any vetted local agent. 20–60 minutes, batchable — several in one neighbourhood cost barely more than one.
Meters as 1 unit.
Scheduled work by the named care manager who knows the family, ending in a written or photographic report that takes real time to produce properly. 2–4 hours.
Meters as 3 units.
Needs a nurse, notary or accountant, or carries genuine liability. The cost is the standby capacity and the insurance behind it, not the hour spent.
Meters as 8 units, or quoted per matter.
Every family is assigned to a single named care manager who owns the relationship, writes the reports and is the person the customer telephones. That ownership is the product. But a named person is also a single point of failure: he is in a hospital corridor across the city, the task is in a district he does not cover, it falls on a day he is unavailable, or it needs a class he is not cleared for. Refusing the task is not an option, and reassigning the family is worse than the problem.
So the tool carries an internal task pool. The owning manager posts the task, another manager claims it, and the commission is split between them. The family never sees a change of manager.
| Step | What happens | Rule that makes it work |
|---|---|---|
| 1 · Post | The owner posts the task with its class, the time window, the district, the spend authority and a short brief drawn from the family's standing notes. | Posting is logged against the owner, not hidden. A task cannot enter the pool without a brief — an unbriefed task is how a handed-off visit turns into a bad report. |
| 2 · Offer | It opens to managers cleared for that class, ranked by district first for twenty minutes, then city-wide. | Proximity first protects route batching, which is where the margin on general tasks lives. Special-class tasks never open to the general pool — only to cleared managers or a contracted partner. |
| 3 · Claim | The first eligible manager to claim takes it, and receives the brief, the address and the spend cap — not the medical file. | Time-limited, task-scoped access to the family's record. Data minimisation is a privacy obligation here, not a preference: the recipient is not the customer and never consented to a city-wide audience. |
| 4 · Split | The taker earns 70% of the task credit. The owner keeps 30% for the brief, the supervision and writing it into the monthly report, and keeps the family retainer in full. | The owner is paid for the relationship and the taker for the work. Because the retainer is untouched, offloading does not pay — but taking a task always does. |
| 5 · Escalate | A scheduled task unclaimed after 24 hours, or a same-day task after twenty minutes, escalates automatically to the operations lead, who assigns it by phone. | An unclaimed task must become someone's problem on a clock, or the pool quietly becomes a place where awkward jobs go to die. |
| 6 · Report | Evidence flows into the same activity log. The customer sees “delivered by Reza, supervised by Maryam” — one owner's name on the report as always. | Transparent without being confusing. Complaints, quality score and the report deadline all stay with the owner. |
Two components: a family retainer per assigned family, which pays for the relationship, the reporting and being reachable; and task credits per completed task, weighted exactly like customer metering — 1 for general, 3 for semi-special, 8 for special. A hand-off moves 70% of the task credit and none of the retainer.
Same-day, evening and out-of-district tasks carry a 1.5× credit multiplier. Without it the pool fills with easy morning errands and the 6pm pharmacy run sits unclaimed until the operations lead has to chase it personally.
Above 30% of a manager's monthly tasks handed off, the account goes to capacity review. This is treated as a staffing signal, not misconduct — a manager consistently over the line is either overloaded or wrongly matched to that family.
Key-person risk is the single largest operational exposure: the relationship lives with one care manager and the company lives with one founder. Named backups and documented client files are the only mitigation that works.
Timestamped arrival, photographs against a fixed checklist, itemised receipts visible in the dashboard. Nothing is marked done without artefacts.
Official fees and contractor invoices pass through at cost with the receipt attached. Two-agent rule on any high-value transaction. No valuables held, ever.
The platform coordinates care and never provides or interprets it. A care manager reports observations and never says "it's probably nothing" — the moment one does, the liability profile of the whole business changes.
"Your manager is on leave" must never mean "nobody knows your mother". Every file is written so a second manager can take over the same week.
A response-time target per tier, a route past the local agent to the Canadian entity, and a written post-incident report within 24 hours of anything going wrong.
The parent is not the payer, and dignity is not a feature request. Cameras, monitoring and data retention require the elder's documented, informed consent.
Four mechanisms, combined: an always-on base, metered quotas by class, published overage, and modules priced per asset rather than per family.
A flat tier with an implied open scope invites the heaviest users to consume the most, which is exactly backwards once the catalogue is 129 services wide. The fix is not a higher price — it is making consumption explicit and welcome.
What the tier buys unconditionally: the emergency line, the dashboard, the named manager, the reporting rhythm. This is standby capacity, and it costs money whether anyone calls or not.
A monthly allowance of general, semi-special and special requests, on three separate counters. One pooled bucket lets a customer spend the whole allowance on the most expensive class.
Exceeding a quota is normal, and priced. Overage is a feature: it is how a customer discovers for themselves that they belong on the tier above.
Property per door, legal per matter, travel per trip. Bundling those into a family tier destroys the tier's margin and undercharges the heavy asset owner.
Every tier is a starting point, not a box. Eight levers let a customer reshape one, grouped by the question they are actually asking. Each reads the same way: what they say, what happens, and why the rule is written that way.
+40% of the tier, not double. The visit, the travel and the manager relationship are shared; only the reporting and the medical coordination genuinely double.
Charging full price for a second parent at the same address is the commonest way this category loses a sale.
+$25 a month per additional household or city.
Covers the travel and the lost route batching when the second address is nowhere near the first.
Up to four payers on one account, each billed their agreed share separately, each with dashboard access.
Turns one $95 decision into three $32 decisions — a materially easier conversation in a family that discusses everything.
Rollover: unused quota carries one month on Watch, two on Vigil, none on the lower tiers.
Quiet months are normal in elder care. Losing the whole allowance every month reads as a con and drives cancellation.
Overage at a published rate that falls as the tier rises — $8 to $14 for a general request, $18 to $28 for an extra visit. Nothing is refused for being over quota.
Overage is how a customer discovers for themselves that they belong on the tier above, without being sold to.
Swap down: one semi-special may be traded for three general requests, once a month. Never upward, and special events are never swappable.
Real flexibility without letting a $45 customer convert an allowance into partner-delivered work that costs $70 to fulfil.
Time instead of tasks: any quota can be drawn as plain agent hours — one hour per general unit, three per semi-special. Clinical and partner work excluded.
Some families don't know what they want done; they know they want someone present. Selling time absorbs every request nobody anticipated.
Ten months for twelve.
Front-loads hard currency, removes eleven chances for a card to fail, and cuts the number of cross-border transfers you have to make.
Pause at 50% for up to two months instead of cancelling. The file, the vault and the manager stay put.
A temporary absence is the commonest reason accounts close. A pause keeps the relationship that took four months to build.
Change tier any month, up or down. Cancel monthly, no notice, no clause.
In a trust-deficit category a lock-in clause is read as a trap. Retention comes from the document vault and the relationship, never the contract.
Reading the quoted cells. Quoted means available but not drawn from quota — priced per matter because partner costs dominate and vary too widely to average. That is the honest structure for anything involving a notary, a lawyer, an accountant or a licensed nurse.
A subscription is one of three ways to buy, and it is not the first one most people choose. The other two are not extras or upsells — they are the acquisition engine, and together they are expected to carry a fifth of revenue.
Recurring, priced per asset. A flat, a parcel of land, a car, a pension, a grave. Bought monthly like a subscription but attached to a thing rather than a person, so an owner with three flats pays three times.
Single purchases, priced individually. No commitment, no account required. Roughly 1.6× the in-plan equivalent — the gap is deliberate, because this menu exists to acquire subscribers rather than to compete with them.
A balance of credits, six-month validity. For the buyer who wants the service occasionally and refuses any recurring charge. The bridge between one job and a tier.
Never folded into a family tier. An owner with three apartments costs three times as much to serve and values the service three times as highly; bundling property into a care tier undercharges exactly the customers most willing to pay.
The complete published menu, grouped by the category each job sits in. Official fees, contractor invoices and gift costs are always passed through at cost with the receipt visible — the platform earns only the stated fee.
5 general credits. Saves $15 against one-off pricing.
3 semi-special credits — a welfare visit with a report, or an appointment escorted and summarised.
7 general and 3 semi-special. The natural purchase after one good experience.
Credits spend on any general or semi-special service in the catalogue at the same 1 and 3 unit weights used inside the tiers, or as plain agent hours. Special events are never sold as credits — they are quoted. The conversion mechanic: a customer who spends a strip within 60 days is offered the matching subscription with the unused balance carried across, which is a warm upgrade conversation with evidence already delivered.
Five questions, asked the way a customer would answer them. The quote assembles itself as you go, and the internal economics sit behind one click.
Two people have to say yes, and the second one is not paying.
The same seven steps, once as one person can run them and once with paid channels and a team. The shape barely changes; only the top of it does. Both are planning placeholders, and the step to measure honestly is the parent gate.
| Step | Why it is shaped that way |
|---|---|
| Content before offer | This category does not convert on a direct advertisement. The buyer is not searching for a product; they are carrying a worry they have not named yet. Naming the worry — "the Friday call that tells you nothing" — outperforms feature advertising in comparable trust-sensitive services. |
| A real sample report on the landing page | The single highest-leverage asset in the business. An anonymised monthly report with photographs, checklist, doctor summary and receipts answers every objection at once without a conversation. Build it before the website copy. |
| An assessment call, not a demo | Twenty structured minutes about the parent's situation produces a tier recommendation the customer arrives at themselves — and doubles as field research from people who are already qualified buyers. |
| Parent acceptance in Persian | Introduce the named manager to the parent directly, before money changes hands. Costs some sales and saves the churn plus the bad word of mouth. In a referral-driven community a failed account is more expensive than a lost one. |
| Referral ask at day 90 | Asking earlier converts badly because the customer has nothing to vouch for. Three reports in, they have evidence, and evidence is what gets forwarded to a sibling. |
Ranked on intent — how close the person already is to wanting this — because in a trust-heavy category intent beats reach every time. The meters are judgement calls to be replaced with measured numbers as soon as anything is running.
Twelve things to make, in three waves. Each one exists to kill a specific objection or to remove a specific piece of work from a sales conversation. The cost column is founder hours plus cash, because that is the real constraint here — not money.
One assumption drives every number on this page.
The source draft's hours column was internally inconsistent — its upper-tier figures implied a semi-special request taking about ninety minutes while the class definition says two to four hours. Everything here is rebuilt from one explicit rate card instead, so the arithmetic can be checked and corrected against real timesheets.
20–60 minutes including travel, assuming it rides an existing neighbourhood route. An urgent, unbatched general request costs closer to 1.5 hours, which is what the overage price covers.
Mid-point of the 2–4 hour class definition, including the time to write the report properly. Report-writing is the part that gets underestimated.
The standard protocol visit plus travel across a large city. Counted separately from the semi-special quota because every tier includes visits outright.
Direct time only. It excludes the standby capacity and the partner invoice, which is exactly why special events are capped or quoted rather than bundled.
Check-in calls, being reachable, keeping the file current and producing the report rhythm. Rises with the tier: half an hour on Voice, three on Vigil.
Mid-point of the 40–70% target band. Nobody consumes a full quota every month, and that gap is most of the real margin — which is why it is stated as an assumption here rather than hidden inside a margin figure.
Two views of the same tier. The left pair is what happens at the expected 55% utilisation; the right pair is the stress case where a customer consumes every unit of their quota every month.
| Product | Price | Hours at 55% use | Margin at 55% | Hours at full quota | Margin at full quota | Accounts per manager | Read |
|---|---|---|---|---|---|---|---|
| Voice | $19 | ~1.1 | 83% | ~2.0 | 68% | ~109 | Three errands, no visit, a shared manager. Healthy on paper because it delivers little — its job is conversion, and it should be capped at three months. |
| Visit | $45 | ~4.7 | 69% | ~8.5 | 43% | ~25 | The volume tier. Note the gap between the two margin columns: a Visit customer who uses everything every month is barely profitable, which is what the overage rates exist to correct. |
| Watch | $95 | ~12.1 | 62% | ~22 | 31% | ~10 | Slightly lower margin percentage than Visit, roughly four times the absolute contribution. This is the tier the business needs, and the one to aim every sales conversation at. |
| Vigil | $175 | ~25.9 | 56% | ~47 | 23% | ~5 | Thinnest margin and highest concentration risk. Five of these occupy one manager entirely, and at full quota use it loses money against a $360 salary. Cap at 15% of the book and require two managers before selling one. |
| Run a door | $65 | ~3.0 | 86% | ~3.5 | 84% | ~40 doors | Almost no gap between the two columns, because property work is scheduled rather than requested. The best margin in the catalogue, and the most predictable. |
| Arrival pack | $120 | ~5.0 | 87% | ~5.0 | 87% | seasonal | Fixed scope, no ongoing obligation, bought by the payer for themselves. The ideal first transaction with a stranger. |
Property earns a better margin, scales better per agent hour and carries none of the reputational tail risk of a bad medical outcome. Sequencing elder care first is right for trust-building. It is not a judgement about where the profit is.
Four Vigil clients consume one full manager. Sell it before the Tehran team has depth and a single resignation takes out your highest-paying accounts simultaneously. Cap it at 15% of the book.
The table assumes full quota consumption. Real utilisation in subscription concierge sits well below 100%, and that gap is most of the actual margin — which means utilisation has to be measured before final prices are set.
Three years, built line by line from the assumptions rather than from a target. The arithmetic is shown so you can disagree with it in specific places.
| Year 1 exit | Year 2 exit | Year 3 exit | |
|---|---|---|---|
| Subscribing families | 40 | 120 | 300 |
| Property doors | 15 | 80 | 300 |
| Tier mix | 15 / 45 / 30 / 10 | 10 / 40 / 38 / 12 | 8 / 35 / 42 / 15 |
| Monthly recurring revenue | ~$4.1k | ~$15.3k | ~$47k |
| Annualised | ~$49k | ~$184k | ~$565k |
| Care managers | 3 | 12 + ops lead | 33 + 2 leads |
| Monthly cost base | ~$2.4k | ~$9.4k | ~$28k |
| Monthly contribution | ~$1.7k | ~$6.4k | ~$21k |
| Founder compensation | none | part-time from month 18 | first full Canadian salary |
| Cities | Tehran | + Mashhad, Isfahan, Shiraz | + Tabriz, Karaj, provincial coverage by contract |
Three lines, added together, every month:
Families × blended price. The blend comes from the tier mix, not from an average guess: at the Year 1 mix of 15 / 45 / 30 / 10 across Voice, Visit, Watch and Vigil, the blended price is $69. It rises to $77 in Year 2 and $83 in Year 3 as the book shifts upward.
Doors × rate. Assumed at a blend of $52–58 per door, reflecting a mix of vacant watch at $39 and tenanted management at $65. Doors grow faster than families from Year 2 because they are batchable and carry no parent gate.
Assumed at 18% of subscription revenue — arrival packs, Nowruz packages, extra visits, credit strips and the à-la-carte menu. Low for a concierge business, deliberately, because it is the least predictable line.
| Year 1 exit, worked through | |
|---|---|
| 40 families × $69 blended | $2,760 |
| 15 doors × $52 blended | $780 |
| One-off and overage, 18% of subscriptions | $500 |
| Monthly recurring revenue | $4,040 |
| Less Tehran: 3 managers at $360, plus agents and transport | −$1,380 |
| Less Canada: entity, bookkeeping, tools, insurance | −$600 |
| Less card fees at about 3%, and marketing | −$430 |
| Monthly contribution, founder unpaid | ~$1,630 |
For the market, yes — obviously. Three hundred families is about 0.02% of the households with a parent still in Iran. Nothing in this plan is limited by how many people want the service. Four things limit it instead, and all four are worth stating plainly because they are also the list of things to attack.
A manager carries about ten Watch-equivalent families. Three hundred families needs roughly thirty trained, supervised, named managers in a city with no HR function and one operations lead. Recruiting, training to protocol and quality-sampling that many people is the actual growth rate of this business.
Every care sale involves a twenty-minute assessment call and a Persian conversation with a parent. Sixty paid starts a month needs three people doing nothing else. One part-time founder produces the Phase 2 funnel, not the Phase 4 one.
Referral-led growth in a category where the incumbent is a cousin runs at roughly 0.3 referrals per retained account per year. That is a good number for this category and a slow one in absolute terms.
Growth is funded from contribution, which caps hiring ahead of demand. This is a choice, not a constraint — but it is the choice this projection is built on.
Three levers, each of which changes the shape rather than the effort. They are listed with the honest caveat attached: each needs its own funnel, its own landing page and its own language, and attempting all three at once is how a two-person company fails.
| Lever | What changes | Year 3 instead | What it costs |
|---|---|---|---|
| Lead with property | A manager carries about forty doors against ten families, doors need no parent gate, and they churn far less. A property-led book of 300 families plus 1,200 doors rather than 300. | ~$99k/mo ≈ $1.2M a year | Roughly double the managers, a separate funnel, and the remittance question answered by counsel rather than avoided. |
| Sell the paperwork on search intent | Certified translation, powers of attorney, pension chasing and document retrieval have real search volume and no trust barrier. 500 one-off jobs a month at an average $45. | +$22k/mo with no subscribers at all | A paid-search budget, faster partner capacity for notarial work, and the discipline to treat it as a funnel into care rather than a separate business. |
| A second country, not a second city | The Tehran operation is the asset; the Canadian entity is a template. Germany, the UK and the US each hold a diaspora of comparable size and buy from the same Tehran team. | 2–3× the same book | An entity and a payment stack per country, local counsel on sanctions each time, and marketing in a third and fourth language. |
The base case above is one book. The same catalogue, the same Tehran team and the same rate card produce very different businesses depending on what you sell and how many of it. Three scales down the side, three mixes across the top. Every figure is computed from the rate card in section 11 — change an assumption there and all nine move together.
Federal or Ontario corporation, registered address, bank account, Stripe.
One on services to family members under the sanctions regime, one on handling third-party money. The only line here that cannot be done cheaply, and the one that must not be skipped.
One landing page, one anonymised sample report, the price sheet. Built by the founder where possible.
Two to three care managers for two to three months, transport, phones, five to ten pilot families. Capped, as in the source plan.
Deliberately unglamorous: a spreadsheet back end, a shared drive vault, WhatsApp for field comms, Stripe for billing. Custom software only after the protocol stops changing weekly.
Self-fundable. No external capital is needed before the property layer, and the first sensible use of outside money is hiring Tehran capacity ahead of demand, not marketing.
Everything else is reporting. These change what you do next.
Ranked by severity, with the mitigation that actually addresses it rather than the one that sounds reassuring.
| Risk | Severity | Mitigation |
|---|---|---|
| Trust, with no known brand | High | Referral-led entry, a published sample report, transparent pass-through, and a Canadian entity the customer can actually pursue. |
| Heterogeneous agent quality | High | Written protocol, photo evidence per visit, continuous family feedback, sampled quality audits, named backup manager. |
| A bad outcome in an emergency | High | Coordination-only boundary in the contract, a rehearsed emergency protocol, liability cover where obtainable, and a written post-incident report within 24 hours. |
| The parent refuses the service | High | Parent-acceptance call as a pre-sale gate; occasions and gifts as a softer entry frame; measured as a funnel metric, not an anecdote. |
| Agent theft or fraud | High | No cash custody, receipts for everything, two-agent rule above a value threshold, bonding where available. |
| Key-person dependency | High | Two managers before any top-tier account is sold; client files written so a second manager can take over the same week. Applies to the founder equally. |
| Outbound customer money via informal rails | High | Rent remittance excluded from the price list until counsel clears it. Do not let this arrive as a customer request you improvise an answer to. |
| Founder time alongside a full-time role | High | Keep the pilot small and pull in only the service families that add revenue without adding founder hours. |
| Quality drift through the task pool | Medium | Separate quality scores for owned and handed-off tasks, a 30% hand-off cap triggering capacity review, task-scoped file access, and the owning manager remaining accountable for the report and the complaint. |
| Canada–Iran fund transfer complexity | Med-high | Identify a stable exchange channel in month one, document the relationship, and never commingle customer money with operating transfers. |
| Cross-border handling of elder health data | Medium | Canadian privacy obligations apply to the entity. Data minimisation, explicit elder consent, defined retention. Cheap to get right now and expensive to retrofit. |
| Currency volatility on the outbound leg | Medium | Price in hard currency, review local pay quarterly, hold a small buffer. A rial collapse improves margins; a sudden appreciation compresses them. |
| Internet and communication disruption in Iran | Medium | Offline-capable protocol, SMS and voice fallback, agents who can report by phone to Canada if platforms are unreachable. |
Four questions that need answers in writing before the first paid customer, not after.
The customer is in Canada, pays a Canadian company in hard currency, and receives a service. That is a different position from prohibited trade, but the answer must come from counsel in plain language, be published as a customer-facing FAQ, and never be improvised in a sales call.
Collecting rent, holding a deposit or moving funds outbound for a customer is where a service coordinator becomes something that needs a licence. The standing policy: no custody of cash, gold or valuables; no discretionary investment; no currency transfer as a standalone product; read-only financial follow-up only.
The elder's health observations, photographs and documents belong to them. Canadian privacy law applies to the entity holding them. Consent is obtained from the parent directly, in Persian, and recorded — including for cameras and for reporting to siblings.
Customer terms with the coordination-only boundary and a published SLA; care-manager agreements with confidentiality and conduct rules; partner agreements with the nursing agency, clinics, notary and accountant; and a data-processing appendix for the dashboard.
Sequenced by what blocks each step, not by what is most exciting.
Validation costs almost nothing and decides whether anything else in this document is worth building.
Diaspora adults with a parent in Iran. One question above all: what did you do the last time something went wrong, and what did it cost you in money, hours and favours?
A van Westendorp survey inside the same interviews: too cheap, cheap, expensive, too expensive. Four tiers means four passes, and it will probably move at least two of the prices in this document.
Publish the just this once menu with prices and a waiting-list button. Measure which of the 129 services people actually click. Expect the ranking to contradict the emotional narrative.
Two real conversations in Tehran about what $360 a month buys, and how many hours survive travel and admin. Every margin figure here rests on this one number.
Before selling anything, ask five elderly parents in Iran whether they would accept a weekly visitor arranged by their child. If more than a third refuse, the product needs redesigning and no amount of marketing fixes it.
One fictional but realistic monthly report, bilingual, with photographs, a checklist, a doctor summary and receipts. It is the landing page, the sales deck and the answer to the theft objection, all at once.
One pattern, used everywhere: the sentence the customer was already saying before they found you.
Every category and every service is named in the customer's first person, as a thing they cannot do from where they are. The pattern is The [concrete noun] I can't [physical verb]. It does three jobs at once: it describes the service, it states the problem, and it puts the customer's own words on your price list.
| Layer | Name | Why |
|---|---|---|
| Tiers | Voice · Visit · Watch · Vigil | One word each, escalating presence, and they alliterate without being twee. A customer understands the ladder before reading a single feature. The I can't line sits underneath as the subtitle — "for the weeks I can't call", "for the month I can't get there". |
| One-off menu | Just this once | Names the buyer's actual state of mind: not ready to commit, needs one thing done. It also makes the subscription the obvious second purchase without a sales pitch. |
| Prepaid credits | Once in a while | The honest description of how most people will use the service, and a natural step between a single job and a subscription. |
| Property module | Watch a door · Run a door · Fill a door · Rebuild a door | "Door" is the unit of pricing, so it should be the unit of language. Four verbs, one noun, instantly comparable. |
| Platform layer | Proof I can't get | The dashboard, vault, receipts and reports are not features; they are the evidence a person abroad has no way of gathering. Naming them as proof is also the answer to the theft objection. |
| The report | The monthly look | What the customer is buying is a look at something they cannot see. Naming the artefact plainly makes a late one feel like the broken promise it is. |