Improve Swiggy and Zomato Ratings

Your Zomato rating slid from 4.2 to 3.9 over six weeks. The instinctive response is always the same: run a discount, reshoot the menu photos, ask an agency about ads. None of it touches the cause. A rating is not a marketing score. It is the aggregate of hundreds of per-order verdicts, and every one of those verdicts was decided inside your kitchen, between the moment the tablet rang and the moment a rider walked out with the bag.

Zomato and Swiggy together handle over 90% of India's food delivery, per analyst reports from 2024–25. That duopoly means your rating on two apps effectively is your reputation. The good news: the inputs that move it are boring, checkable, and entirely under your control.

What a delivery rating actually measures

Ratings are collected per order and rolled up into the number on your listing. Neither platform publishes the exact formula, so ignore anyone selling you "the algorithm." What you can act on is the anatomy of a bad order. The complaint categories in Zomato's partner systems are explicit: quality, quantity, wrong items, and packaging or spillage.

Read that list again. Not one of those four is fixed by better photography. Wrong items is a packing failure. Quantity is a portioning or forgotten-add-on failure. Spillage is a packaging failure. Quality is usually holding temperature or stale prep. Each one maps to a checklist your shift either ran or didn't.

KPT: the operational delivery clock

KPT, kitchen preparation time, affects the delivery estimate a customer sees and the timing of rider allocation and handover. Zomato does not publicly describe KPT as a general listing-rank factor, so treat claims that slow KPT automatically sinks visibility with caution.

It is worth understanding how seriously this is engineered. Zomato's engineering blog describes how the platform once measured KPT crudely, from order-accepted timestamp to rider-pickup timestamp, then rebuilt the system to predict true food preparation time per item, per quantity, per outlet, adjusted for time-of-day rush. The same dish gets a different predicted prep time at 9 PM on Friday than at 3 PM on Tuesday, because the model knows your kitchen is loaded.

Two operational consequences follow. First, you cannot game the clock: marking "order ready" while the biryani is still being packed looks clever for one order, but repeated late handover after an early "ready" degrades your metrics, because the rider's actual wait exposes the gap. Second, your prep-time setting is a promise. Set 20 minutes and consistently take 30, and every customer watches an estimate you were never going to keep.

Refunds get charged back to your payout

Zomato's restaurant partner terms require handover in spill-proof packaging, and they allow customer refunds to be charged back to the restaurant for acts or omissions attributable to it, including poor quality packaging. A dal makhani that leaks into the carry bag is not just a one-star review; it can be a debit against your weekly settlement.

Regulators are circling this space too. In May 2025 the CCPA opened a probe into cancellation and refund practices after 10,590 complaints against Swiggy and 7,938 against Zomato reached the National Consumer Helpline. Whatever comes of it, the operational lesson stands: every refund has a root cause on a specific shift, at a specific station.

The five fixes a shift can actually run

Each of these is a checklist with an owner, a time, and evidence. Not a memo, not a pep talk.

1. Order-accuracy check at packing

• Owner: Packing lead / dispatch executive | Timing: Every order, before sealing
• Evidence/Proof: Order slip ticked item by item against the bag contents, add-ons and cutlery included.
• Escalation: Any mismatch means the bag does not seal. Call the station, fix the order, then mark ready.

1. Honest prep-time settings per daypart

• Owner: Outlet manager | Timing: Reviewed every Monday
• Evidence/Proof: Prep-time settings compared against actual KPT for lunch rush, dinner rush and off-peak from last week's numbers.
• Escalation: If actuals miss the setting by five minutes or more in any daypart, either change the setting or fix the station causing the delay. Do not leave the gap standing.

1. 86 items before you run out mid-order

• Owner: Shift lead | Timing: Continuous, with a hard stock check at 6 PM before dinner
• Evidence/Proof: Stock-out-risk items flagged at the evening line check; anything below par for the night is switched off on both apps.
• Escalation: An accepted order you cannot fulfil correctly becomes a cancellation or a wrong-item complaint. Turn the dish off before the apps sell what you don't have.

1. Packaging QC on every gravy and beverage

• Owner: Packer | Timing: Every wet item
• Evidence/Proof: Lid pressed until it clicks, container tilted for two seconds over the counter, no seepage, tape seal on the bag.
• Escalation: Any leak means re-potting into a fresh container. Never send it and hope. The full station routine is in our cloud kitchen packing checklist.

1. Weekly complaint-category review

• Owner: Owner / ops manager | Timing: A fixed 30 minutes, same day every week
• Evidence/Proof: Complaints from both apps sorted into the four buckets (quality, quantity, wrong items, packaging/spillage), with each recurring cause logged as a corrective action carrying an owner and a deadline.
• Escalation: A category rising two weeks in a row gets a named owner and a station-level fix, not a "team, please be careful" message on the group.

Ratings recover slower than they fall

Set expectations honestly with yourself. One horror order sits in your aggregate for a long time, and recovery is a volume game of consistently clean orders stacked on top of it. That is exactly why discounting is the wrong lever: it buys more orders through an unchanged kitchen, which at best dilutes the damage slowly and at worst mass-produces the same failure at higher volume. Fix the line first. If quality complaints are your biggest bucket, start upstream with a proper line check before service so dishes leave the pass at standard.

Paper checklists survive until the rush

A laminated packing card genuinely works at one quiet outlet with a strong shift lead. It breaks in the two situations that generate most bad ratings: the Friday-night rush, when nobody ticks anything, and the manager's day off, when nobody checks whether anyone ticked anything. Paper has no timestamps, so a 7 PM check done honestly looks identical to ten boxes filled in at 11 PM. Digital checklists with photo evidence make the packing check auditable per order window, and exceptions reach the manager while the shift is still on, not at month-end.

Your rating is the average of your worst habits at your busiest hour. If you want to see how a per-shift checklist system handles that hour, open the Food Ops demo, build the packing check from this article, and run it on tomorrow's dinner service.