Quick commerce advertising works when your ads align with four things: availability, search rank, what competitors are doing, and how much demand actually exists in that category, city and hour. When those four factors are out of sync, you spend money on ads that cannot convert — either because the shelf is empty, your product is too far down the search results to be seen, a competitor has the Buy Box, or demand has shifted. This playbook covers how to diagnose which factor is hurting you and how to fix it on each platform.
Why quick commerce ads are different from Amazon or D2C ads
On Amazon or Meta, an ad that runs while stock is low can still build momentum — there is usually more inventory on its way. On Blinkit, Zepto, Instamart or Flipkart Minutes, there is one dark store serving your neighbourhood. If that store’s shelf is empty, every click you pay for is a lost order and a wasted rupee. The 10-minute delivery promise also means buying decisions happen fast: a shopper who sees your ad and finds you out of stock will not wait for restocking — they buy the competitor product that is available.
The platforms also run on a different attribution model than what brands are used to. Quick commerce platforms typically credit sales to ads on a D+1 basis — but as our case studies show, attribution revisions from D+1 to D+3 on a Blinkit Product-Booster campaign increased attributed sales by +127%, +310% and +3,050% over three days. That means a campaign that looks unprofitable on Monday could look highly profitable by Wednesday. Any playbook that ignores attribution timing will cut campaigns that were working.
The 4-factor diagnostic
Before touching a bid or a budget, run this check for every SKU in every market you sell in. It takes 15 minutes and tells you exactly why performance is off.
Factor 1: Availability
Is the SKU available in the dark stores where your ads are running? A dark store in Bandra can be fully stocked while Koramangala is empty. Ads on a city level will keep spending in Koramangala and converting in Bandra — your average looks fine but you are losing money on half the spend. Check availability at the dark-store level, not just the platform level. Our stockout guide explains what to do when stock runs low.
Factor 2: Search rank
Search rank on quick commerce works differently than on e-commerce. Rank is partly determined by recent sales velocity — which means ads and organic rank reinforce each other. If your rank has fallen, ads alone will not fix it: you need to identify whether the drop came from lower sales velocity, a listing issue (incorrect category, weak search tags) or a competitor gaining share. Our ranking guide covers the levers.
Factor 3: Competitor activity
A competitor running a deep discount on your category, or running aggressive ads on your brand search terms, will inflate your CPC, reduce your conversion rate and make your ROAS look worse than it is. Before cutting a campaign that is “not performing,” check whether a competitor event — a new SKU, a price drop, a sale — coincides with the performance dip. Brand defence (covered below) handles the most expensive version of this.
Factor 4: Category demand
Quick commerce demand is hyperlocal and temporal. The demand for cold drinks spikes on a hot afternoon in Delhi; the demand for dal and rice is different in a south-Indian neighbourhood versus a north-Indian one. If demand in your category is simply low in a given hour, city or season, increasing bids will not help — you are bidding for clicks that do not exist. Our micromarkets guide covers how to read demand at a neighbourhood level.
Platform tactics
Each platform runs its own ads system, and the controls differ. Blinkit, Zepto, Swiggy Instamart and Flipkart Minutes each have different ad formats, targeting options and attribution windows.
Blinkit
Blinkit runs Product-Booster as its primary ad format. Ads are bought on a CPC model with a daily budget and show at the top and middle of search results. Brand Central lets you target by geography, from pan-India down to a specific city, and offers time-of-day targeting to reach shoppers during peak consumption hours. Attribution is typically D+1, with revisions following over the next 2–3 days.
One of ShelfRadar’s case studies ran a 22-day Product-Booster test for a leading plant-protein brand on a ₹2,000/day cap. Hourly ROAS by slot ranged from 0.00× (a dead slot with no conversions) to 3.08× (a high-performing slot that was underfunded). By pulling spend from dead slots and bidding up the winning ones, the same ₹2,000/day cap delivered +32% revenue. The key was D+3 attribution — every decision waited for the full count before acting.
Zepto
Zepto’s sponsored listing system allows brands to bid on keyword-triggered placements. One important caveat noted by industry practitioners: Zepto’s platform ROAS is sometimes calculated on MRP rather than the actual selling price after platform discounts. If you are running deep discounts on Zepto, verify whether your ROAS figure uses your actual realised price or MRP — the gap can be significant. Our budget split guide covers how to allocate across platforms.
Swiggy Instamart
Instamart offers sponsored search and banner placements. Demand on Instamart skews toward food and groceries at meal times, so dayparting — adjusting bids and budgets by hour — is especially impactful. Our dayparting guide covers the method.
Flipkart Minutes
Flipkart Minutes is the newest of the four platforms at scale, and its advertising controls are still maturing. The overlap with Flipkart’s main marketplace creates a cross-platform signal that brands with large Flipkart presence can read across both. Watch for city-level rollouts as the platform expands beyond its initial metro footprint.
Budget split framework
If you are selling on multiple platforms, do not split your budget based on order volume alone. The right split depends on:
- Which platform your category converts best on. A personal-care brand may convert 2× better on Zepto than on Instamart for the same CPM, while a staples brand has the opposite pattern.
- Availability by platform. If supply chain constraints mean your Instamart availability is inconsistent, reduce spend there until it stabilises — you are paying for clicks that cannot convert.
- Competitive density. A platform where three competitors are spending aggressively will have higher CPCs and lower conversion rates for a challenger brand.
Start with a 60/40 split toward the platform where you have the strongest distribution, then rebalance quarterly as you read attribution data. Our budget split guide has the full framework.
Bidding and dayparting
Quick commerce demand moves through the day: morning grocery runs, afternoon snack peaks, pre-dinner shopping. Bidding the same amount at 2 AM and 8 PM wastes budget on hours where shoppers are not buying. The right approach:
- Pull 4–8 weeks of hourly sales data (not just ad data — total sales) by platform.
- Find the hours where ROAS is consistently above your target, and the hours where it is consistently below.
- Set higher bids or higher budget multipliers for the winning hours; pull spend from the losing ones.
- Wait for D+3 before judging any change — a bid cut that looks right at D+1 may be cutting a profitable hour at D+3.
Our dayparting guide covers this in detail, including how to handle attribution lag when making dayparting decisions.
Attribution and measurement
The core measurement mistake in quick commerce is reading D+1 results and making D+1 decisions. Quick commerce platforms attribute orders to ads over multiple days after the click. A Blinkit Product-Booster campaign ShelfRadar ran saw D+1 attribution revisions of +127%, +310% and +3,050% by D+3 — meaning the campaign appeared to be generating almost no revenue on Monday but was actually generating strong revenue that showed up over the following days.
The correct measurement discipline:
- Never cut a campaign within 48 hours of a change. Wait for attribution to settle.
- Compare full weeks, not individual days. Day-of-week demand patterns are strong on quick commerce.
- Use ROAS as a direction, not a target. A 2× ROAS at a ₹2,000/day cap may be worse than a 1.8× ROAS at ₹5,000/day, depending on margin and market share goals.
Brand defence
When a competitor buys ads on your brand search terms — your product name, your brand name, or your hero SKU — they intercept the highest-intent shoppers on quick commerce: the people who already want to buy you. Brand defence means bidding on your own brand terms so your competitor cannot take that traffic. On quick commerce this is especially important because brand search traffic converts at much higher rates than generic category terms.
Our share of shelf guide explains how to monitor whether a competitor is gaining impression share on your brand terms.
Seasonal strategy
Quick commerce ad rates increase significantly during festive season and other demand peaks. This creates a decision point: spend more in peak season at higher CPCs, or hold budget for the weeks before and after the peak when CPCs are lower and you can build rank heading into it. Our seasonal surge guide covers the pre-peak, peak and post-peak budget tactics.
The revenue leak audit
Before adding more spend, check the three biggest revenue leaks we see repeatedly across brands:
Stockout ads: Ads running on SKUs where stock has run out in the most important dark stores. Every click here is a wasted rupee and a negative first impression. The fix: connect availability data to your ad controls so campaigns pause or reduce spend automatically as stock falls below a threshold.
Dead slots: Time slots where your ROAS is consistently 0 or near 0 across multiple weeks. These are dark-store-level demand deserts for your category. Pull spend entirely and redirect to your winning hours.
D+1 misjudgements: Campaigns that looked underperforming and were cut before D+3 attribution settled. These show up as erratic performance — campaigns that work great for a few weeks, then get cut, then restart and work great again. The fix: establish a minimum attribution-wait policy before any campaign change.
ShelfRadar’s case study on a ₹2,000/day Blinkit cap found ₹4,688 of waste inside that cap in 22 days — from dead slots and stockout ads — that was redirected to the winning hours without increasing total spend.
Getting the full picture: solutions and diagnostics
The 4-factor diagnostic, attribution discipline and platform tactics above will improve most campaigns. But the biggest revenue leaks are usually multi-lever: availability is hurting ads, which are hurting rank, which is making the whole picture look like a category problem when it is actually a supply-chain problem. Our solutions page covers how ShelfRadar reads all four factors together and shows the fix for each.
FAQ
How much should I spend on quick commerce advertising as a percentage of revenue?
There is no universal rule. The right level depends on your category’s competitive density, your margins, and your phase (new brand on the platform versus established brand defending share). Start with a small test budget, read ROAS at D+3, and scale the winning hours before increasing total spend.
Should I pause ads on Blinkit if my stock is low?
Yes — if stock in the key dark stores for your category is below 2–3 days on hand, pulling or reducing ad spend prevents you paying for clicks that convert to stockouts. Restart ads once stock lands. Our stockout guide has the thresholds.
Is it worth advertising on all four platforms simultaneously?
Only if you have reliable distribution on all four. An ad that runs where your product is unavailable wastes money and trains shoppers to see you as out of stock. Start with the one or two platforms where your availability is strongest, then expand as you stabilise supply.
Why does my ROAS look different on each platform’s dashboard?
Attribution windows, ROAS calculation methodology (MRP vs actual selling price), and the lag between click and order credit all differ by platform. Build a weekly reconciliation that reads D+3 data from each platform into one view before making budget decisions.
How long should I run a test before judging a campaign?
A minimum of 14 days at stable spend, read at D+3. Shorter windows are distorted by attribution lag, day-of-week patterns and random variation. Use the first 7 days to build data, the second 7 to read results.