Delivery marketing in Paris means running two platforms well

It doubles every decision. Uber Eats and Deliveroo both operate here, both sell placement, both rank storefronts, and neither knows what you are doing on the other. Where a single-app market poses one auction and one ranking system, Paris poses two of each, so every price, photo, offer and ad budget gets decided twice.

Both also deliver with their own courier networks, so neither storefront is a listings-only fallback for the other: if one app throttles your visibility, those orders do not resurface elsewhere in your arrondissement. Concentrating on a single app is a larger bet here than owners expect. It is also why comparing pays: the same dish, same kitchen, same hour, live in two places at once.

Not by picking a ratio on day one. Most of the waste sits inside each platform, before any split is argued about: a budget spread flatly across opening hours burns much of itself in windows that were never going to convert. Midi and soir are not one market, and a storefront that competes well at lunch can be outranked by 20:00. Concentrating spend where you can realistically win the order, and pausing it in dead hours, has cut roughly 30% of wasted ad spend across the 60+ stores we manage in the Netherlands and taken the strongest to as much as 14x return on ad spend.

Only then does the split stop being guesswork. Same dishes, same hours, two sets of results: we compare net payout after commission, promotion cost and ad spend, and move budget toward whichever app returns more for that store. With only two in play the trade-off is stark: fund both adequately, or fund one properly. The answer moves with the weekday and the season, so the ratio is reviewed weekly. Each move is a recommendation with its expected effect stated up front; you approve, then we execute.

A menu rebuild is the cheapest lever either platform gives you. It covers photography, item names and descriptions written in French because that is what diners read and type into the app's search bar, the order items appear in, and formule-style combos that lift basket size rather than discount it. Across the 60+ stores we manage in the Netherlands, menu conversion has roughly doubled after a rebuild and basket size has risen 38%. Platform-published benchmarks agree: dish photos are reported to drive up to 20% more clicks and 32% more orders.

The two apps do not display a menu identically, and accents are their own trap. Image crops, category layout and description truncation all differ, so we write and sequence per storefront rather than pasting one menu twice; and because diners type dish names with and without diacritics, an item named only in its accented spelling misses the searches typed plain. Conversion is measured separately on each: under 20% needs work, 20% is decent, 30% and above is strong, 40% is best in class. A store converting at 30% on one app and under 20% on the other has a platform problem, not a menu problem.

Yes, because the work happens inside the platform back offices, not in your dining room. We will not pretend otherwise: OrderLift is based in Amsterdam and works remotely, and the 60+ restaurants we manage today are in the Netherlands, not in France. What we bring to Paris is method and platform craft, not a local client list.

The rhythm is weekly. You get a plan per storefront with the reasoning and the expected effect, you approve or reject it line by line, and we execute what you approved. Reporting runs against your own pre-OrderLift baseline on net payout, not platform gross. Across those 60+ Dutch partners, orders have grown 40% on average over six months, with no partner below 30% and some locations above 200%. From you we need back-office access to both apps and a weekly approval slot.