Compare a restaurant menu from fifteen years ago with one today and the differences aren't only about fashion. A substantial share of the change is structural, driven by the fact that a large proportion of orders now travel in a bag on a motorcycle.
What travels and what doesn't
Delivery imposes constraints that dine-in doesn't, and they're unforgiving.
A dish has to survive twenty to forty minutes in a sealed container, during which it continues cooking in its own heat, releases steam, and settles.
Some things handle this well. Curries and gravies actually improve slightly. Biryani travels excellently. Anything braised or slow-cooked is largely unaffected.
Other things are destroyed. Anything crisp becomes soft — the steam has nowhere to go. Delicate textures collapse. Dishes assembled to be eaten immediately arrive as a single mass. Anything requiring a temperature contrast between components is impossible.
The consequence is a systematic shift in menus towards food that's robust in transit, and away from food that isn't, regardless of whether the second category is better.
The packaging arms race
Enormous effort has gone into engineering around this, and it's largely invisible to customers.
Vented containers to release steam. Compartmentalised boxes keeping wet and dry separate. Absorbent liners. Separate sachets for anything that would go soggy. Insulated bags.
All of it costs money per order, which comes out of a margin that's already compressed by platform commission. A restaurant with a fifteen percent margin paying a substantial commission plus packaging is operating in a very narrow band.
Which explains menu pricing that differs between the platform and the counter. That's not opportunism; it's the commission being priced in.
The cloud kitchen model
The structural change that followed. If a large share of orders never involves a customer entering premises, the premises don't need to be presentable, accessible or in an expensive location.
A kitchen in a cheap industrial area, serving several brands from the same equipment, with no front of house, no waiting staff and no seating, has a dramatically lower cost base.
The multi-brand aspect is the part customers rarely realise. A single kitchen may operate as five or six distinct brands on delivery platforms, each with its own menu and identity, sharing staff, equipment and ingredients.
This is efficient and it produces a strange consumer experience: three apparently different restaurants that turn out to be the same kitchen with the same base preparations.
Menu design for search
An effect nobody anticipated. Discovery on delivery platforms happens through search and category browsing, so dish names and categorisation determine whether anybody finds you.
That pushes towards descriptive, searchable names over creative ones. A dish called something evocative doesn't surface for anybody typing an ingredient.
It also pushes towards covering popular search categories even where they're not the kitchen's strength, because absence from a category means absence from those results entirely.
The result is menu sprawl — kitchens producing a wide range of dishes to maximise search coverage, at the cost of focus. It's the opposite of the culinary advice about doing a few things well, and it's a rational response to how the demand arrives.
The ratings dynamic
Platform ratings determine visibility, and the mechanics shape behaviour in specific ways.
A rating reflects the whole experience — food, packaging, delivery time, driver behaviour — much of which the kitchen doesn't control. Restaurants are assessed on things they can't influence.
Because ratings determine placement, and placement determines volume, a small decline can have disproportionate effects. That produces defensive behaviour: generous free items, aggressive complaint resolution, and reluctance to attempt anything that might travel badly.
It also pushes towards consistency over ambition. A dish that's excellent when it works and poor when it doesn't scores worse than one that's reliably adequate.
What's been gained
Worth being balanced, because the changes aren't all losses.
Enormously expanded access. Regional cuisines available far outside their home areas, at prices that don't require a restaurant's overheads.
Lower barriers to entry. Starting a food business without a dining room, a lease in a prime location, or front-of-house staff has allowed a great many people to start who couldn't have before.
And genuine specialisation. A kitchen doing exactly one thing for a delivery radius has an addressable market that a physical restaurant in the same location wouldn't.
Where it's going
The commission economics remain the unresolved tension. Platform fees are high enough that many restaurants are barely profitable on delivery orders, and the pressure to move customers to direct ordering is constant.
Several chains have built their own ordering channels and offer meaningful discounts for using them, which is a straightforward attempt to reclaim the commission.
Whether that works depends on whether customers will maintain multiple ordering relationships rather than defaulting to an aggregator. The historical pattern in most aggregated markets suggests convenience wins, which means the tension is likely to persist rather than resolve.