Third-party delivery is the channel where the platform owns the order, the data, the search rank, the commission, and the Guest relationship. The platform dispatches the driver. The operator is a listing inside someone else’s marketplace.

The conventional read on third-party delivery is the inverse of the first-party read, and just as one-sided: platforms are bad, commissions are theft, get off the platforms.

This read is wrong. Not because platforms are good — they are not friends of the operator — but because it ignores what third-party delivery is actually buying.

Third-party delivery is buying reach. The platform has demand aggregation the operator does not have and cannot build. The Guest who opens DoorDash or Uber Eats at 7:14 on a Tuesday is not opening it to find this operator. They are opening it to find food. The operator who is not on the platform is not in the consideration set.

What the operator gives up: the order, the data, the Guest relationship, 25-30 percent of the ticket, and authority over the last mile. What the operator gets: access to a demand pool they could not reach otherwise, zero driver management burden, and reach into a Guest segment that was never going to walk through the door.

The tradeoff is real. Neither side of it is optional.

The Duopoly Is Hardening

When two competitors are described as “highly rational, likely to move pricing in tandem,” that is not a reassuring market structure. That is a duopoly with no escape valve. Two-thirds of marketplace customers are already on subscription. The platform-side growth lever is mostly pulled. From here, platform revenue growth comes from charging operators more — not acquiring new consumers. Operator economics get worse from here, not better. The operator who has no first-party channel is renting their entire delivery customer base from two landlords who just announced rent is going up. That is not a delivery problem. That is an ownership problem.

What the data is showing: the gap between DoorDash and Uber Eats is widening. DoorDash’s execution advantage and lower required ad spend are producing stronger ROI. Uber Eats requires significantly higher investment to maintain comparable sales with inconsistent results. The operator who treats both platforms identically is leaving margin on the table.

The deeper issue: paid visibility is no longer optional. Going dark on ad spend — even temporarily — produces measurable sales declines within weeks. One forum with major brand off-premise leaders found that going dark on Uber Eats resulted in an immediate 8 percent sales decline that only recovered when spend was reinstated. That is not a marketing line. That is recurring rent. The platform tax is not 25-30 percent of the ticket. It is 25-30 percent of the ticket plus the ad spend required to appear in the results. The operator who is not accounting for both is not reading the actual cost.

The Vertical Integration Threat

The real threat is not commission rates. It is vertical integration. When a delivery platform owns your POS, your reservations, your catering channel, and your local commerce pipeline, they own your Guest data stack. The delivery relationship becomes the operating relationship. The operator who waits until the full suite is offered before deciding whether to participate has already lost the negotiation. The decision about first-party data ownership is not a future decision. It is the decision that is being made right now, shift by shift, order by order, every time a Guest books through a platform instead of directly.

Promotional Orders Are Subsidized by the Operator

Promotional orders on the major platforms run approximately five dollars higher in average check than baseline. That lift is funded by the operator — through markups, free-item attachments, and promotional budget. The operator running heavy promotional activity on the platforms is not driving incremental revenue. They are subsidizing platform engagement metrics with their own margin, generating check averages that look healthy on a report and destroy profitability on a cost basis. The promotion is not a marketing expense. It is a margin transfer to the platform dressed up as a sales strategy.

There is one more cost that does not appear on the commission statement. The platforms bid on your brand name in Google. A Guest who searches “Portillo’s delivery” does not necessarily land on Portillo’s — they land on whichever marketplace won the auction for that search term. The platform uses your brand equity, built by your operation, to intercept a Guest who was already looking for you specifically, and routes them through the platform’s order flow instead of yours. You pay the commission on an order the Guest intended to place directly. Portillo’s negotiated a clause preventing their marketplace partners from bidding on branded search terms. Most operators have not. If you are on a platform and you have not addressed this in your contract, your brand is working for their search strategy.

The Agentic Web and Discovery Arbitrage

AI agents are already booking restaurant reservations on behalf of users. They are searching availability, checking waitlists, comparing options — before a human ever sees a result. The operator with no direct booking channel, no structured data on their site, no clear positioning is invisible to the agent. The platform that owns the reservation owns the agent relationship. That is Discovery Arbitrage accelerating into a layer most operators don’t know exists yet. The defense is the same defense that has always existed: direct relationships, owned Guest data, clear positioning that both humans and agents can find, understand, and act on. Unclear positioning is no longer just a marketing problem. In the agentic web, it is an invisibility problem.

Third-party delivery is not a channel you opt into casually. It is a structural commitment with a structural cost. Understand both before you sign on — and understand them annually after you do.

Stop Treating Delivery as a Margin Channel

Delivery is not a margin channel. It is a paid-acquisition channel — and the cost of acquisition just doubled. Marketing spend on the major platforms has gone from negligible to 6-8% of sales in two years. Commissions held. The fight is no longer over commission rates. It is over visibility inventory. The operator who is not budgeting marketplace marketing as a line item alongside every other paid channel is being outbid for placement by every operator who is. The delivery P&L that made sense in 2022 does not make sense in 2026. Run the current numbers before you decide whether you are in this channel or not.

The cost that never appears in the tradeoff analysis is the loyalty cost. Every order that goes through a third-party platform is an order where you did not build a direct relationship with that Guest. They didn’t interact with your team. They didn’t experience your environment. They received food in a bag with someone else’s branding on the bag. Whatever loyalty might have developed from a direct experience — the server who remembered their name, the manager who stopped by the table, the moment of surprise and delight — none of that happened. The platform took the Guest relationship and kept it. You got the ticket. That is the actual cost of third-party dependency, and it compounds every week you don’t build an alternative.