Menu Engineering: How to Find and Remove the Low-Margin Items That Are Quietly Eating Your Cloud Kitchen Profits

Menu engineering for cloud kitchens is not about making your menu look better. It's about making it perform better — financially, operationally, and algorithmically. And the ugly truth is that most Indian cloud kitchens on Swiggy and Zomato are running delivery menus with 8 to 12 loss-making items buried inside a 30 to 40-item list, generating genuine per-order losses every time those items sell. The owners know their bestsellers. Nobody knows their worst performers.

Most experts get this wrong because they apply dine-in menu engineering logic — which is primarily about customer psychology, item placement, and upsell sequencing — to a delivery context where none of those mechanics apply. On Swiggy or Zomato, the customer sees a flat list sorted by platform algorithm. There is no physical menu design. There is no waiter upsell. The only lever you have is which items are on the list, at what price, and whether the economics of each one work in your favour.

So let's build the correct menu engineering for cloud kitchens framework — with an item-level profitability calculation, the four-quadrant audit matrix, and specific decisions for each category.

Why Your Delivery Menu Is Probably Too Large and Too Unprofitable

Here's something the platforms don't advertise: a larger menu does not mean more orders. Platform data from multiple sources consistently shows that cloud kitchen menus with 15–25 well-selected items achieve higher conversion rates than menus with 40–60 items. More items mean more decision paralysis for the customer. More items also mean more ingredient complexity in your kitchen, more preparation variables, more things that can go wrong during a rush, and — most importantly — more items that are likely to be loss-making.

But the most damaging thing about a bloated delivery menu isn't the customer experience problem. It's the financial one. Every time a customer orders a loss-making item, you have generated a transaction that costs you more to fulfil than you receive for it. And on a delivery platform where Swiggy or Zomato is taking 30%+ of your revenue before you've bought a single ingredient, the margin window for individual items is narrow enough that even a 5% error in food cost pricing makes the difference between a profitable item and a loss-making one.

The Per-Order Reality: What Platform Costs Leave for Your Food Budget

Before you can identify which items are loss-making, you need to know what the platform is actually leaving for you. Here is the complete breakdown on a ₹550 order — the baseline from which all per-item margin calculations start.

The table above shows a critical number that most cloud kitchen operators never calculate: at a 25% commission rate, your absolute maximum food cost budget on a ₹550 order — before any profit — is ₹360.75, or 65.6% of menu price. Everything above that is a loss. And once you add packaging (₹22) and a modest overhead allocation (₹50), the sustainable food cost window is approximately 35% — exactly where the margin math starts working.

The Item-Level Audit: The Calculation Nobody Does

The menu engineering for cloud kitchens process starts with one non-negotiable step: calculating the net margin for every single item on your delivery menu individually. Not a portfolio average. Not an estimate. Every item. With actual food cost data.

The formula is the same one we've covered throughout this series, applied per item:

Do this for every item. Put the results in a spreadsheet — item name, menu price, COGS, net margin (₹), net margin (%). Sort by net margin % ascending. The items at the top of that list when sorted lowest-first are your problem items. And you will be surprised. The "popular" butter chicken that drives half your orders may have a COGS of ₹185 on a ₹380 menu price — barely 18% net margin at the item level. The "boring" dal makhani that sells 30 times a day has a COGS of ₹65 on a ₹280 price — 28% net margin. The boring item is your profit engine. The popular one is your margin leak.

The Four-Quadrant Matrix: Stars, Plowhorses, Puzzles, and Dogs

Once you have per-item margin data, plot every item on two axes: profitability (net margin %) on the Y axis and popularity (weekly order volume) on the X axis. Every item lands in one of four quadrants, and each quadrant has a specific action.

What to Actually Do With Each Quadrant

Stars: Protect and Amplify

Stars are the items where both your platform economics and your customer demand align. Don't touch the price. Don't discount them during promotions — if you're participating in a "20% off" campaign, explicitly exclude your Stars from the discount scope if the platform allows it. Invest your ad spend almost entirely on Stars. On Swiggy, a well-targeted restaurant ad driving traffic to a Star item converts better and costs less per acquisition than one driving traffic to a mixed menu. And include Stars in every combo you build — the combo's overall margin will carry the weight of any lower-margin supporting items.

Plowhorses: The Hardest Decision

Plowhorses are where most cloud kitchen operators get stuck — emotionally and commercially. The item is popular. Customers like it. It drives order volume. Removing it feels risky. But the ugly truth is that running a high-volume low-margin item at scale means you're working harder for less. For every 100 orders of a Plowhorse item at 15% margin, you're generating ₹8,250 in net margin (at ₹550 price). The same 100 orders of a Star item at 30% margin generates ₹16,500. The Plowhorse is costing you ₹8,250 in opportunity cost per 100 orders.

The answer for Plowhorses is not removal — it's restructuring. Two options: raise the menu price until the margin improves (if competitive pricing allows), or reduce the COGS by modifying the recipe, portion size, or ingredient specification. A Plowhorse with a ₹180 COGS on a ₹380 item becomes viable if you can bring COGS to ₹130 through portion adjustment without destroying the customer perception of value.

Puzzles: Rename and Reposition

Low-volume items with strong margins are underperforming for a reason that usually isn't the food. They're buried in the menu. They have uninspiring names. They don't appear in any combo. They get zero ad spend. A Puzzle item called "Rajma Masala" at ₹240 with 28% margin that sells 12 times a week should be "Home-Style Rajma Masala — Slow Cooked" at ₹260, featured in the "Bestseller Combos" section, and included in every meal box. The food hasn't changed. The presentation has. And — based on multiple operator case studies — this kind of repositioning typically increases a Puzzle item's volume by 40–80% within a month.

Dogs: Remove Without Sentiment

Dogs need to go. Full stop. The argument "but some customers specifically order this" is not a business argument — it's a sentiment argument. If the item has low margin and low volume, its loyal customer base is so small that removing it will not measurably impact your total order count. What it will do is simplify your kitchen operations, reduce ingredient waste, and eliminate a source of per-order losses. Remove the item from your platform menu. If specific customers enquire, tell them it's not available for delivery. That's it.

The 2026 Hidden Fee Update: Why Menu Engineering Is More Urgent This Year

The financial case for rigorous menu engineering for cloud kitchens was compelling in 2024. In 2026, four structural changes have made it urgent.

Update 1 — Platform Commission Tiers Have Tightened. Both Swiggy and Zomato have reduced the availability of the lowest commission tiers (18–20%) in 2025–2026, with more restaurants now on standard 23–25% rates. For an item that was marginally profitable at 20% commission, the shift to 25% can push it into negative territory. If you haven't recalculated your per-item margins since 2023 or 2024, you may be running items that were viable then and are loss-making now.

Update 2 — Algorithm Reward for Smaller, High-Converting Menus. Both platforms' search algorithms in 2026 weight "menu conversion rate" — the percentage of users who view your menu and place an order — as a ranking signal. A focused 18-item menu with 22% conversion outranks a 45-item menu with 12% conversion, all else being equal. This means the algorithmic benefit of a leaner menu is now quantifiable in terms of search visibility — not just operational efficiency.

Update 3 — Ingredient Inflation Has Repriced COGS Across Categories. Indian food commodity prices in 2026 have pushed COGS upward across edible oils, proteins, and several spice categories. An item whose COGS was 32% in January 2024 may be 39–42% by mid-2026 without any recipe change. This commodity inflation has silently moved multiple items from the Plowhorse quadrant into the Dog quadrant — and from marginal profitability into loss-making — without any visible operational change. A menu engineering audit based on current COGS figures is not optional at this price environment.

Update 4 — Promotional Co-Funding Now Requires Per-Item Margin Headroom. Platform promotional campaigns in 2026 require the restaurant to absorb 40–60% of the advertised discount. For an item with 15% net margin and a campaign discount of 20% off, the restaurant is absorbing a ₹11 loss per order on a ₹550 item just from the co-funding obligation. Without per-item margin data, restaurant operators are accepting promotional participation for items that cannot mathematically support the discount — and generating losses on every promotional order of those items.

Your Menu Is Either Working For You or Against You. There Is No Middle.

A 35-item delivery menu where 10 items are Dogs, 8 are Plowhorses running thin margins, and only 12 are Stars or Puzzles is not a balanced offering. It's a margin liability. Every time a customer orders one of those 10 Dogs, you are generating a transaction that costs you more to fulfil than you receive for it — and the platform takes its commission regardless.

The menu engineering for cloud kitchens process described here — per-item margin calculation, four-quadrant classification, and specific action per quadrant — is not complicated. It requires your COGS data, your menu prices, and a spreadsheet. It takes 2–3 hours for a 30-item menu. The output is a list of items to remove, items to reprice, and items to promote. And it should be repeated every quarter, because ingredient costs change, platform commission tiers change, and the promotional co-funding landscape changes.

So audit your menu. This week. With the actual numbers. And then make the decisions the numbers demand — not the ones that feel comfortable.