Independent business-case illustration · researched 10 November 2026. This is a planning scenario, not commissioned client work, a quotation or a record of measured results. Named businesses supply public context only; their contracts and finances are not assumed.
It is 8:40 on a weekday. A regular wants a coffee and something to eat before work. They could open a delivery marketplace, follow your café’s ordering link or walk to the counter. What would make ordering from you directly easier—and would the business actually earn more?
A branded app can help a regular reorder, collect a reward or choose a pickup time. It does not automatically bring new customers, replace a rider network or make marketing free. The commercial question is whether a direct channel creates enough extra contribution to cover the work of running it.
Keep the discovery. Earn the return visit.
Marketplaces can introduce people who have never heard of your café. Familiar checkout, reviews and delivery tracking can make a first purchase feel easier. A hybrid approach keeps that discovery route while offering willing regulars a convenient direct option. Any migration must respect the merchant agreement; do not harvest marketplace customer identities or assume you can solicit customers through its order data.
| Approach | What the customer gets | What the café must own |
|---|---|---|
| Marketplace-only | Discovery, familiar payment and delivery experience. | Food preparation, availability, packaging and agreed service standards; commission and other contracted charges. |
| Branded web ordering / SaaS | An order link on your own website, without installing an app. | Acquisition, payments setup, order acceptance, refunds, delivery or pickup, permissions and campaign work. |
| Standalone custom app; exit marketplaces | A tailored repeat-order journey if people choose to install it. | All direct-channel duties plus releases, backend, security and the risk of losing marketplace discovery. |
| Hybrid: marketplace + direct web/app | Choice: discover through a marketplace or return directly. | Reconcile menus, stock and orders; avoid counting the same customer or order twice. |
Talabat’s UAE merchant page (accessed 10 November 2026) describes the café preparing orders and its riders delivering them. It also explains that commissions support payments, customer support and other platform costs. Those are services a direct channel must retain or replace.
Start with the alternatives you can actually buy
Neither the Talabat UAE partner page nor Deliveroo’s UAE public terms (accessed 10 November 2026) establishes one numerical delivery commission for every café. Deliveroo refers to fees in the merchant’s Agreement. Ask for the signed rate schedule, charge base, VAT invoice, refund rules, advertising commitments, exclusivity, price conditions and delivery terms. The 25% used below is an assumption, not either platform’s advertised UAE tariff.
GloriaFood’s advertised pricing (accessed 10 November 2026) offers web/mobile-browser ordering without a per-order commission or core monthly fee. Its optional online-card connector is US$29/month and advanced promotions US$19/month, before tax; payment processing and delivery remain separate. That is a serious low-cost alternative to building custom software.
Its ordering overview (accessed 10 November 2026) also advertises a US$9/month website and US$59/month branded app. The customer-app page (accessed 10 November 2026) specifies Android. Confirm current availability, UAE onboarding, supported gateway and exact platform scope before buying. An advertised branded shell and a bespoke two-platform product are different purchases; neither is inherently the right answer.
Integration is another buying question. GloriaFood’s API description (accessed 10 November 2026) says its order feed follows acceptance in the merchant order-taking app. Do not assume every order, refund and stock change will flow both ways into your till. Demonstrate your exact POS and courier combination before committing.
A real menu, an invented sales mix
For a price anchor, Le Café LPM’s public Dubai sample menu (accessed 10 November 2026) lists a latte at AED 22, a butter croissant at AED 25 and a jambon-beurre sandwich at AED 70. We use these three prices only. We have no information about that café’s sales mix, marketplace participation, takeaway availability, costs or plans.
| Illustrative basket | Menu-price total | Assumed share |
|---|---|---|
| Latte + croissant | AED 47 | 60% |
| Sandwich + latte | AED 92 | 40% |
| Weighted average | 0.60 × 47 + 0.40 × 92 = AED 65 | 100% |
The mix is invented for this illustration. We treat these totals as VAT-inclusive, with no extra service charge, and keep the same food prices across channels. Validate the actual receipt, takeaway pricing, tax treatment and product availability before applying this to a business.
The AED 65 order does not create AED 65 of profit
At the assumed 5% VAT rate, the café retains AED 61.90 of food sales before costs. We then allow 31% of net sales for ingredients, AED 6 variable kitchen labour, AED 2 packaging and 1% of net sales for waste/refunds. These are public planning assumptions, not anyone’s actual margins.
| Per average order | Marketplace | Direct |
|---|---|---|
| Food sales excluding VAT | 61.90 | 61.90 |
| Ingredients + variable labour + packaging + waste/refund reserve | 27.81 | 27.81 |
| Assumed marketplace commission: 25% × VAT-inclusive AED 65 | 16.25 | — |
| Card processing: 2.9% × 65 + AED 1 | Included in assumed marketplace charge | 2.89 |
| Net merchant fulfilment cost after customer delivery charges | Included in assumed marketplace charge | 4.00 |
| Rewards: 2% of net sales; extra handling | — | 1.24 + 1.00 |
| Contribution remaining (AED) | 17.85 | 24.97 |
Moving an existing order improves contribution by AED 7.13, not AED 24.97 and not AED 65. A genuinely additional direct order contributes AED 24.97 at these costs. If a direct order simply replaces a walk-in, count only the difference from that walk-in’s contribution. This model assigns no extra in-store uplift.
The AED 4 fulfilment assumption represents 50% pickup at no extra courier cost and 50% delivery with an AED 8 net merchant cost. This is not a courier quote. It is the remaining cost after any delivery fee collected from customers; that fee is not also counted as extra food revenue. Direct customer-service handling is budgeted separately.
Stripe UAE’s standard domestic-card price (accessed 10 November 2026) is the payment benchmark. International cards add 1%, and currency conversion adds another 1% where required; disputes and some other services add charges. A particular SaaS gateway combination still needs verification.
What it costs to own the channel
For a custom app, the illustration budgets AED 40,000 for the customer-facing build, 15,000 backend, 10,000 integrations, 5,000 QA/security/release preparation and 5,000 launch/training/creative. Google’s one-time US$25 account and Apple’s first US$99 year add about AED 455. Three prelaunch months at AED 500 add AED 1,500. Total initial funding: AED 76,955. These are editable planning allowances, not Select Creatives prices.
| Budget, AED | Custom app + Care | SaaS web pilot |
|---|---|---|
| Build / setup + backend + integrations + QA | 70,000 | 10,000 |
| Launch / training / creative | 5,000 | 3,000 |
| First store-account charges | 455 | 0 |
| Prelaunch operations | 1,500 over 3 months | 500 over 1 month |
| Initial funding, rounded | 76,955 | 13,500 |
| Maintenance per month | 1,000 Care | 300 web support |
| Hosting/domain + backend/monitoring per month | 300 + 200 | 50 + 0 |
| Ordering/promotions subscriptions per month | 0; covered by build assumptions | 176 = US$48 × assumed FX |
| Email/CRM allowance per month | 100 | 100 |
| WhatsApp marketing allowance per month | 460 | 460 |
| Campaign/support labour + paid acquisition per month | 1,000 + 500 | 1,000 + 500 |
| Normal monthly cash outlay before annual account renewal | 3,560 | 2,586 |
Care/AMC is a maintenance service, not free future development. Our comparison also allows AED 9,000/year of ad hoc maintenance instead of AED 12,000/year Care, with the same hosting, tools and annual accounts. The AED 3,000 reduction changes base payback from month 36 to month 33 after launch; it does not remove releases, security or bug work. Stress-test AED 18,000/year of ad hoc work and agree response times, backups, security patching, OS support and what counts as a feature change. None of these budgets is a supplier quotation.
The SaaS pilot keeps the existing site and till; the custom option also assumes no POS replacement. If the till cannot accept the proposed integration, scope and price the change. The web pilot has no app-store accounts. Apple publishes US$99/year membership (accessed 10 November 2026); Google publishes a US$25 registration fee (accessed 10 November 2026). We assume AED 3.6725 per US dollar for planning, excluding bank FX fees.
Three adoption scenarios—and two different clocks
Assume the café already receives 3,000 marketplace orders a month. The custom hybrid leaves the marketplace open. The monthly targets below are reached gradually: 15%, 25%, 40%, 55%, 70%, 85%, then 100% from month seven. That means 8.9 months of full-target contribution in the first 12 operating months, not 12. Running costs start at full budget on launch.
| Custom + Care | Mature moved / new orders monthly | Mature monthly net benefit* (AED) | First 12 operating months net cash after startup (AED) | First 12 months from kickoff net cash (AED) | Payback: launch / kickoff |
|---|---|---|---|---|---|
| Low | 150 / 20 | -2,022 | -105,713 | -99,740 | No recovery in 60 / 63 months |
| Base | 600 / 80 | 2,684 | -63,835 | -71,977 | Month 36 / 39 |
| High | 1400 / 220 | 11,881 | 18,024 | -17,711 | Month 11 / 14 |
*Mature monthly benefit averages annual Apple renewals; the cash ledger charges the first year in startup and renewals in operating months 13, 25, 37 and 49. Initial funding is committed at kickoff; launch follows a three-month build. A one-month delay moves kickoff payback back a month and adds its prelaunch cost. There is no assumed growth after maturity. Figures are incremental to keeping the existing marketplace operation, before finance costs and corporate tax.
In the base case, 5,340 orders move channel during year one. The 712 genuinely additional orders create AED 44,076 additional food revenue excluding VAT. Together, moved-order savings and new-order contribution add AED 55,838 before channel running costs. Deduct AED 42,718 running costs and AED 76,955 startup funding: first-year net cash benefit is −AED 63,835. Revenue, contribution and cash benefit are different numbers.
The high case recovers investment in operating month 11, but month 14 from kickoff. It is still cash-negative at the first anniversary of kickoff. The low case loses money even after adoption matures; waiting longer does not repair a negative operating contribution.
Web first is a commercial option, not a compromise
| Base adoption: 600 moved + 80 new / month | First operating-year cash benefit after startup | Payback after launch / kickoff |
|---|---|---|
| Marketplace-only baseline | AED 0 incremental; existing trading continues | No new project to recover |
| Hybrid SaaS web pilot | AED 11,305 | Month 9 / 10 |
| Hybrid custom app + Care | AED -63,835 | Month 36 / 39 |
| Hybrid custom app + ad hoc budget | AED -60,835 | Month 33 / 36 |
All three direct options use the same adoption assumptions. The model gives the custom app no invented sales premium. A lower-cost web pilot wins this comparison unless a tested app-specific benefit or necessary integration justifies the extra cost. The SaaS result is conditional on product availability, payment support and integration fit; free software does not mean free implementation or delivery.
The assumptions that can change the answer
| Change to the custom base case | Mature monthly benefit, annual fees averaged | Practical implication |
|---|---|---|
| Commission assumption falls from 25% to 20% | AED 734 | No payback within 60 operating months at this adoption. |
| Commission assumption rises to 30% | AED 4,634 | Payback in month 23 after launch; still an assumption, not a tariff. |
| Net fulfilment cost rises from AED 4 to AED 8 | AED -36 | Slight operating loss; no recovery while unchanged. |
| 25% commission applies to net food sales, not VAT-inclusive basket | AED 2,220 | Payback moves to month 43 after launch. |
| Exit marketplaces: 2,400 moved, 600 lost, 80 new monthly | AED 4,805 | Assumed 20% discovery loss; payback month 22 after launch. |
| Exit marketplaces: 1,950 moved, 1,050 lost, 80 new monthly | AED -6,432 | Assumed 35% discovery loss; operating loss, no recovery. |
The exit examples phase the marketplace down using the same ramp. Every lost order is separate from the moved orders and loses its marketplace contribution. At the base unit economics, losing roughly 793 of the original 3,000 orders a month wipes out mature operating benefit for the exit strategy, even with 80 new direct orders. This is a sensitivity calculation, not a forecast of customer behaviour.
The custom base case needs 224 moved orders a month plus 80 genuinely additional orders to cover running costs at maturity. That is operating break-even, not repayment of the project. To recover startup as well, the target is much higher:
| Recovery deadline | Moved orders/month from launch | Moved orders/month from kickoff |
|---|---|---|
| 12 months | 1,607 | 2,312 |
| 18 months | 1,052 | 1,261 |
| 24 months | 813 | 912 |
These are mature monthly targets with the stated ramp, 80 new orders monthly and no marketplace loss. A 12-month recovery from launch needs 1,607 existing orders moved each month—more than half the 3,000-order baseline. If the available repeat audience cannot support that, reduce scope or decline the investment.
A customer relationship is useful only when customers want it
Offer a clear reason to return: a saved favourite, a reliable collection slot or a reward with an affordable cost. Ask separately for permission to send email, WhatsApp or relevant app notifications. Keep consent records, preferences and easy opt-outs. Segment by useful behaviour—such as breakfast pickup or occasional gifting—rather than sending every offer to everyone.
WhatsApp’s Business Messaging Policy, updated 23 September 2026 (accessed 10 November 2026) requires opt-in and respect for opt-outs; business-initiated platform messages use approved templates. A customer’s phone number on a receipt is not unlimited marketing permission. A new direct channel also does not unlock historical marketplace customer identities.
The model budgets 2,000 outbound UAE marketing messages each month at an illustrative US$0.0626 each: US$0.0576 Meta fixed marketing reference plus US$0.005 Twilio charge. That is about AED 460. Twilio’s current rate reference (accessed 10 November 2026) describes auction pricing too, so check the account’s actual pricing model and campaign estimate. Twilio’s pricing page (accessed 10 November 2026) also charges inbound messages; replies and utility messages must be added when relevant.
Email/CRM gets an AED 100 monthly allowance; Brevo advertises entry pricing from US$9/month (accessed 10 November 2026), but contact volume, features and sending levels affect the bill. Campaign and customer-support labour has its own AED 1,000 allowance. Add SMS, push-vendor, inbox handling or higher email tiers if used; the backend/tools budget is not an unlimited allowance.
We assume zero avoidable old marketing or data fees. Enter savings only when a real bill disappears, and keep replacement tools and campaign work in the budget. The value of permissioned records is the measured extra contribution they help create. We assign no cash value to a database itself. The 80 new orders are already the retention/acquisition scenario; no extra “loyalty uplift” is added on top.
A small pilot before a large commitment
Start with one location, a short pickup-friendly menu and an eight-week branded web-ordering pilot. A customer opens the café’s link, sees current availability, chooses a collection slot, pays, receives confirmation and collects from an identified handoff point. An optional, separate opt-in allows a relevant return invitation later.
Retain the current marketplace listings, POS, kitchen process and working reservation tools. Assign a staff owner for acceptance, sold-out items, refunds and late collections. Prove the till handoff, receipts, payment reconciliation, allergy information and courier fallback. Do not assume a published API removes all manual work or grants unrestricted data exports.
Measure contribution per order, checkout completion, pickup punctuality, refunds, staff minutes, permission rates and 30/60-day repeat contribution. Separate moved orders from genuinely incremental purchases through first-party order history and a consented comparison cohort or holdout. A last-click conversion count alone cannot establish incrementality. Decide a maximum affordable acquisition cost and a stop/go threshold before spending.
A native app becomes worth testing when regular customers have a recurring reason to install and keep it—such as frequent reorder, membership or useful notifications—and the web journey cannot meet that need economically. For occasional customers, a mobile-friendly link may be enough. A custom app is unsuitable when volume is unproven, fulfilment consumes the margin, staff cannot maintain the channel or a supported off-the-shelf service already covers the job.
Run your own business case
Would direct ordering earn its keep?
Illustrative planning model, not a quote or forecast. Change the assumptions. All calculations run in your browser; inputs are not sent to Select Creatives.
Base illustration: AED 7.13 gain per moved order; custom app first operating year net cash benefit −AED 63,835. Interactive inputs require JavaScript.
Ramp: 15%, 25%, 40%, 55%, 70%, 85%, then 100% from month 7. No growth after maturity. First annual app-store fee is in startup; renewals occur in months 13, 25, 37 and 49. All startup funding is committed at kickoff; lead time adds months before trading. Model excludes financing, corporation tax, VAT timing, fixed rent and unchanged staff overhead. Use only incremental or avoidable costs. Fee estimates require vendor and tax checks.
Before you commit: the accounting and contract checks
Open the formulas, VAT treatment and integration checks
The model assumes a VAT-registered café making standard-rated food sales. FTA guidance states a 5% standard rate (accessed 10 November 2026). Revenue excludes output VAT. Commission is assumed to apply to the gross food basket before fee VAT; it excludes customer delivery fees, tips and other levies. Your agreement may use a different base or add minimums, per-order fees, discounts, refunds, subscriptions or merchant-funded promotions. Add those before deciding.
All cost budgets are shown before recoverable input VAT. FTA conditions for input-tax recovery (accessed 10 November 2026) must be met. If fee or project VAT is not recoverable, add it to the relevant cost. This is a management contribution model, not a VAT return or VAT cash-timing forecast; a non-registered business needs a different treatment. Unchanged rent and fixed salaries are excluded, not erased. Extra shifts, capacity, chargebacks, FX, refunds beyond the reserve and incremental discounts must be added if incurred.
Order payments are for physical food, not digital in-app content. Apple’s physical-goods rule, section 3.1.3(e) (accessed 10 November 2026) directs these purchases to other payment methods; this model does not add an App Store digital-content commission. Gateway charges and developer-account renewal still apply. Confirm platform and payment rules for your actual product.
For data and integrations, request written confirmation of fields available, retention, export/deletion rights, marketing restrictions, permitted promotion, API approval, POS compatibility, fees and exit process. Oracle’s restaurant privacy policy, dated 22 October 2025 (accessed 10 November 2026) distinguishes restaurant-controlled direct interactions and requires appropriate notices and consents. We did not verify a universal UAE marketplace customer-data export entitlement. Deliveroo’s UAE page contains a Menu Manager/API restriction in UK-specific wording; confirm territory applicability rather than importing that rule into a UAE contract.
| Calculator-ready formula | Definition |
|---|---|
| Net food sales S | B ÷ (1 + VAT rate) |
| Common variable cost V | S × (food% + refund/waste%) + kitchen labour + packaging |
| Marketplace contribution CMm | S − V − commission% × agreed base − other marketplace variable charges (assumed 0 here) |
| Direct contribution CMd | S − V − card% × B − fixed card fee − net fulfilment − rewards% × S − direct handling |
| Moved-order lift L | CMd − CMm |
| Monthly contribution gain | ramp × (moved × L + genuinely new × CMd − lost × CMm) |
| Monthly fixed replacement cost F | Care/ad hoc + hosting + backend + SaaS + email + messaging + campaign labour + ads − genuinely avoidable old fees |
| Startup K | Build + backend build + integrations + QA + launch + first account fees + prelaunch months × prelaunch cost |
| Monthly cash change | Contribution gain − F − account renewals due that month |
| Cumulative cash after n operating months | −K + sum(monthly cash change, months 1…n) |
| Payback | First operating month cumulative cash ≥ 0; kickoff month = launch payback + prelaunch months |
| Recovery moved-order target at n operating months | ceil[max(0, (K + nF + renewals − new × CMd × sum(ramp) + lost × CMm × sum(ramp)) ÷ (L × sum(ramp)))] |
| Operating moved-order break-even | ceil[max(0, (F + annual account renewal/12 + lost × CMm − new × CMd) ÷ L)] |
Use decimal rates in formulas. When L ≤ 0, moved-order recovery is not supported by this formula; additional demand needs separate proof. Revenue is not added to contribution. Refund reserves are included once; the default marketplace fee has no separately added payment or rider cost, and no saved fixed marketing fee is assumed. All three adoption scenarios use the same costs and basket. Investment payback is nominal, undiscounted and unfinanced, over a 60-operating-month horizon.
Source register and refresh notes
Published café menu (accessed 10 November 2026) — Public prices; sample menu, not sales mix or delivery availability.
Talabat UAE partner information (accessed 10 November 2026) — Platform role, riders and merchant preparation; no public numerical delivery commission verified.
Deliveroo UAE terms (accessed 10 November 2026) — Fees refer to merchant Agreement. Menu Manager section dated 27 April 2026 contains UK-specific wording; UAE applicability must be confirmed.
Deliveroo UAE partner policies (accessed 10 November 2026) — Policy page updated 14 July 2026; operational duties and customer compensation.
GloriaFood pricing (accessed 10 November 2026) — Public advertised subscription prices exclude taxes; gateway, delivery and operational costs additional.
GloriaFood ordering overview (accessed 10 November 2026) — Advertises optional website $9/month and branded app $59/month; reconcile current availability with account quote.
GloriaFood branded-app scope (accessed 10 November 2026) — Specifically describes Android branded customer app. Do not confuse with iOS/Android merchant order-taking app.
GloriaFood API (accessed 10 November 2026) — Order feed follows acceptance in order-taking app; menu retrieval; POS and fleet integration use cases.
Oracle restaurant privacy policy (accessed 10 November 2026) — Dated 22 October 2025; restaurants control direct interactions and are responsible for notices/consents.
Stripe UAE standard pricing (accessed 10 November 2026) — 2.9% + AED 1 domestic-card benchmark; international/FX and dispute charges extra.
Twilio WhatsApp pricing (accessed 10 November 2026) — BSP fee of $0.005 per inbound/outbound message plus applicable Meta charges.
Twilio WhatsApp rate reference (accessed 10 November 2026) — UAE fixed marketing reference $0.0576; auction/bid model and account eligibility can change campaign cost.
WhatsApp Business Messaging Policy (accessed 10 November 2026) — Updated 23 September 2026; opt-in, opt-outs, templates and service-window rules.
Brevo plan documentation (accessed 10 November 2026) — Starter advertised from $9/month; usage, contacts, features and add-ons affect bill.
Apple developer enrollment (accessed 10 November 2026) — US$99/year where applicable local currency; organization requirements.
Google Play registration (accessed 10 November 2026) — US$25 one-time registration; verification/release requirements still apply.
Apple payment rules (accessed 10 November 2026) — Section 3.1.3(e): physical goods/services consumed outside app use other payment methods.
FTA standard VAT rate (accessed 10 November 2026) — 5% standard rate; individual supplies and registration status require checking.
FTA input VAT conditions (accessed 10 November 2026) — Recovery is conditional, not automatic.
Select Creatives services (accessed 10 November 2026) — Public websites, apps, platforms and automation offering; no private rate card or client result used.
Version 1 · 10 November 2026. Refresh price/terms pages before a proposal and at least every 90 days; merchant-specific quotes, actual margins and consented pilot evidence override the assumptions. No confidential client information, private rate card, testimonial or claimed client result is used.
Make the channel earn its place
Select Creatives is a Dubai studio building websites, apps, platforms and integrations. We can help assess a direct-channel business case, compare existing tools and shape a small test before recommending a build. Ask us for a business-case assessment—bring your order volumes, public menu and an anonymised merchant fee schedule. We will explore whether web ordering, a supported platform, a custom app or keeping the current setup makes commercial sense.
