First-Party vs Aggregator Delivery: A Hybrid Architecture to Reclaim Margin and Data

Chapter 1 — The mistake of treating commission as a tax on all volume

Aggregator commission is defensible as a cost of acquisition and indefensible as a perpetual logistics toll, and that is the entire thesis. In the groups I advise, delivery weighs between 22% and 40% of sales, but every euro through an aggregator enters the P&L amputated by a commission of 25% to 32%. On a dish with 30% food cost, that commission does not trim the margin: it evaporates it. I have watched boards treat that 28% as an inevitable tax on ALL revenue, when half of that volume is repurchase from customers the platform kidnapped from you. The first order was captured by the aggregator and deserves its commission; the tenth order from the same customer does not. Paying 28% once to acquire is marketing. Paying it ten times on the same customer is a structural leak that no P&L can withstand. Neither 100% owned delivery nor 100% aggregator: the winning architecture in 2026 is hybrid by function.

Chapter 2 — What is the winning architecture in 2026?

Use aggregators as a paid acquisition channel —the 28% commission is booked as marketing cost, not logistics— and build an owned channel for repurchase, where the customer already knows you and the marginal CAC approaches zero.

The aggregator charges per transaction; the owned channel charges once, the day you acquire the customer, and repurchase after that is nearly free. The arithmetic is brutal: if a customer orders ten times a year at 25 euros, the aggregator takes 70 euros in annual commission; migrated to the owned channel after the first order, that figure drops to 7. Function rules: discovery and incremental demand go through the platform; loyalty and frequency, through the owned channel. Segmenting by function, not by platform, is what recovers margin. In the aggregator model the customer is not yours: you have no email, no history, no permission to reactivate them, and that is the silent loss that appears on no invoice.

Chapter 3 — The customer is yours or belongs to the platform

The platform hands you an order and keeps the data, which is worth more than the commission. In the hybrid model, every order through the owned channel generates an RFM record —recency, frequency, value— that feeds reactivation by WhatsApp or email with a marginal CAC near zero. A customer inactive for 45 days receives a direct message; WhatsApp reactivation converts between 8% and 15% of that base, without paying commission or bidding again. On a base of 2,000 owned customers, that is hundreds of recovered orders per month that in the aggregator model simply go cold. The customer data is the asset; the commission is the price of not owning it. The status quo of depending on the aggregator alone is pure OpEx with no CapEx: dirt cheap to start and brutally expensive to sustain at scale, because the commission keeps running on every order. The owned channel demands the opposite: a modest CapEx of between 2,000 and 9,000 USD in setup —transactional website, payment gateway, kitchen integration and last-mile logistics— that pays back in weeks, not years.

Chapter 4 — Pure OpEx today, modest CapEx that pays back in weeks

The break-even math is clear: if you migrate orders that previously left 28% commission with the aggregator, every euro migrated above the owned channel's break-even threshold falls straight to margin. In a venue billing 15,000 USD/month in delivery, migrating just 30% of the volume recovers around 1,260 USD monthly in commission; the setup pays for itself in under two months. The aggregator is perpetual rent; the owned channel is an amortizable purchase. The hybrid channel wins in all three input-inflation scenarios I model for the board, because it protects the only margin you control: repurchase. Base scenario, input inflation at 6%: the 30% food cost rises to 31.8% and the aggregator commission amplifies that pressure on all volume. Stressed scenario, 12%: food cost brushes 33.6%, the operating ceiling, and every aggregator order enters near a loss. Severe scenario, 18%: without an owned channel, delivery stops being profitable and survives only to avoid ceding share.

Chapter 5 — Three input-inflation scenarios and why the hybrid wins in all of them

In all three, the order migrated to the owned channel saves the full commission and delivers 20 to 30 points of margin oxygen versus the same order through the platform. Inflation is not fought by raising the price on the aggregator menu; it is fought by taking back from the aggregator the share of volume that is already yours. The 90-day roadmap I deliver to boards has three phases with one hard KPI each, measurable in the board's reporting. Days 1 to 30, foundations: stand up the transactional website and gateway, targeting 100 validated owned orders and an average ticket equal to or above the aggregator's. Days 31 to 60, migration: activate the RFM record and WhatsApp reactivation, with the KPI of moving 15% to 25% of total aggregator volume to the owned channel with no drop in absolute orders. Days 61 to 90, consolidation: optimize blended CAC and repurchase frequency, aiming for an owned channel that sustains 30% of delivery and a weighted average commission below 20%.

Chapter 6 — A 90-day roadmap with KPIs for the board

As Diego F. Parra sums it up at Masterestaurant: the board does not measure clicks, it measures margin recovered per euro migrated, and that number has to climb every month. Commission is defensible CAC on the first order and indefensible leakage from the second onward, and separating those two euros is what changes the conversation in the boardroom. A 28% CAC on a 25-euro ticket is 7 euros to capture a new customer: compared to what acquisition by digital advertising costs —between 5 and 15 euros per conversion in food— it is competitive and even cheap. The problem appears when that same 28% is charged on order 2, 3 and 10, turning a reasonable CAC into a toll that multiplies acquisition cost by purchase frequency. The operating rule is simple: measure real blended CAC —total commission divided by NEW customers, not by total orders— and you will see the honest number is three or four times worse than the one the platform reports.

Chapter 7 — Commission as CAC: the number you actually should defend

That is the figure that justifies every dollar of the owned channel. The aggregator charges commission on EVERY transaction; the owned channel charges it once —the acquisition CAC— and repurchase then costs almost nothing. The gap between paying 28% once and paying it ten times on the same customer is the whole thesis of this document. In the aggregator model the customer is the platform's: you have no email, no history, no permission to reactivate them. In the hybrid, every owned order generates an RFM record that feeds WhatsApp or email reactivation at a marginal CAC near zero. The status quo is pure OpEx with no CapEx: cheap to start, brutally expensive to sustain at scale. The hybrid demands modest CapEx (2,000-9,000 USD of setup) that amortizes in weeks once shifted volume clears the owned channel's break-even. Structural dependency is the silent risk: whoever bills 100% of delivery through an aggregator is exposed to a unilateral take-rate change wiping out a quarter's margin overnight.

Chapter 8 — The differences that decide profitability

The owned channel is the buffer that turns that systemic risk into a manageable one.