Marketing10 min read
ROAS drop during Diwali: 7 checks to run before you cut budget
admin
It is the Monday of Diwali week. Spend is up, the dashboard says ROAS fell from last week, and the founder wants to know whether to pull budget before the weekend.
A ROAS drop during Diwali rarely has one cause. Auction costs, a reset learning phase, attribution that has not caught up, a sold-out hero SKU, deeper discounts and COD orders that never get delivered can all pull the same number down. Each needs a different fix. Here are seven checks, in order.
Some dip in Diwali week is normal, but it is not automatic, so treat it as a question rather than an excuse. ROAS here means revenue from ads divided by ad spend. More brands bid for the same shoppers, so costs can rise. Conversion rates often rise too, which can cancel out the higher cost.
The evidence points both ways:
The maths matters more than either number. The Clevrr D2C Festive Report 2026 shows that a 40% CPM increase with a 50% conversion rate lift can still improve CPA by roughly 7%. If your ROAS fell, something in that balance broke. The checks below find what.
The big marketplace sales ran a month before Diwali this year, so your demand curve probably looks different from 2025. Diwali 2026 falls on Sunday 8 November, with the festival running from 6 to 10 November, per Zoutons' 2026 sale calendar. In 2025 Diwali fell in October.
The marketplace sales, though, stayed in early October:
| Sale | 2026 start |
|---|---|
| Flipkart Big Billion Days | 8 Oct (early access), 9 Oct (everyone) |
| Amazon Great Indian Festival | 8 Oct |
| Myntra Big Fashion Festival | 7 Oct (early access), 8 Oct |
| Meesho Mega Blockbuster Sale | 9 Oct |
| AJIO All Star Sale | 7 to 18 Oct |
Source: Indian Retailer, 7 October 2026.
That gap matters. Many shoppers bought big-ticket items in October, and Diwali week now sits about four weeks after the marketplace peak. Festive demand also runs far longer than one week: the Clevrr report describes 5 to 6 months of festive demand, from July to New Year. Judge Diwali week against that whole season, not against a single good week in October.
Diagnose a Diwali ROAS drop with seven checks, in order: last year's festive week, cost versus conversion rate, learning resets, attribution lag, stock, discount depth and RTO. The first four tell you whether the drop is real. The last three find leaks that sit outside Ads Manager.
Compare Diwali week with Diwali week last year, not with last week. Niblin's seasonal ROAS guide makes the same point: always compare with the same period last year, not last month.
Because Diwali moved from October in 2025 to November in 2026, line up the weeks by their distance from Diwali, not by calendar date. Week minus 1 in 2026 is 1 to 7 November. Week minus 1 in 2025 was in October.
ROAS moves when cost per click, conversion rate or order value moves. Pull CPM, CTR, CPC, conversion rate and AOV for the drop period and the comparison period, side by side.
Look at your change log for the 7 days before the drop. Festive weeks invite constant edits, and some of them send ad sets back into learning.
Meta's learning phase generally needs about 50 optimised events in 7 days to exit. Publishing new ads, changing ads, updating targeting or significantly increasing budget can send an ad set back into learning, per Jon Loomer's breakdown. The rules have loosened over time: in one test, adding one new ad to an ad set of 22 ads did not restart learning.
Large budget jumps carry the most risk. Linkrunner's Diwali guide notes that large budget increases can push a campaign back into learning. If three ad sets doubled budget on the Friday before Diwali, their ROAS on Monday tells you very little yet.
Recent days always look worse than they will once delayed conversions arrive. Festive shoppers browse longer and buy later, so the gap widens in Diwali week.
Linkrunner points out that Big Billion Days runs roughly ten days, longer than a typical seven-day click window. A shopper who clicked your ad on day one of a long sale may convert outside the window you report on. Before you call a drop, wait a few days and re-pull the same date range. Our guide to Meta's attribution changes covers how the windows work, and why Meta purchases do not match Shopify covers the reporting gap.
Festive demand sells out hero SKUs and popular sizes faster than usual. When that happens, Meta keeps spending on product pages where shoppers cannot buy, and ROAS falls even though the campaign did nothing wrong.
Pull a list of every live ad and its landing product, then check stock for each variant. A sold-out size on a best-selling ad can drag down the whole campaign. We wrote a full guide on how to stop ad spend on out-of-stock products.
A festive offer can lift orders and still lower ROAS, because each order brings in less revenue. If you moved from 20% off to 40% off for Diwali, revenue per order fell even if conversion rate rose.
Say a brand spends ₹5L in Diwali week and gets 1,000 orders. At an AOV of ₹1,500, that is ₹15L revenue and a ROAS of 3.0. If a deeper discount drops AOV to ₹1,200 on the same 1,000 orders, revenue falls to ₹12L and ROAS to 2.4. (Illustrative numbers.) Nothing broke in the ad account. The offer changed the maths.
Platform ROAS counts an order when the shopper places it, not when you get paid. During the festive rush, return to origin (RTO), cart abandonment and NDR all rise, as Inc42 reported, and delivery times stretch.
So your Diwali ROAS can look fine in Ads Manager and still lose money once COD orders bounce back. Delivery costs went up too: Delhivery added a ₹4 surcharge per express shipment and ₹2 per surface shipment for D2C brands from 1 September 2026, according to Inc42. Track RTO-adjusted ROAS for festive campaigns separately.
Change only the thing the checks point to. Cutting budget across the board fixes none of these causes and can reset learning on ad sets that were fine.
| Cause | What you see | What to do |
|---|---|---|
| Auction cost | CPM up, CVR flat or up | Hold steady if CPA still works; shift budget to ad sets with the best CVR |
| Learning reset | Recent big edits, "Learning" status | Stop editing; give ad sets time to gather events |
| Attribution lag | Last 2 to 3 days look worst | Wait, then re-pull the same date range |
| Stock-out | Ads live on sold-out variants | Pause those ads or point them to in-stock products |
| Discount depth | Orders up, AOV down | Judge on contribution margin (revenue left after product, shipping, payment and marketing costs), not ROAS alone |
| RTO on COD | Platform ROAS fine, delivered revenue down | Track RTO-adjusted ROAS; review COD exposure |
Plan for after Diwali too. Razorpay Rize notes that a second buying wave often follows the first, so a slow day right after Diwali is not a reason to switch everything off.
Use them only if you expect a sharp, short change in conversion rate. Google's own guidance says Smart Bidding already manages seasonal events, and seasonality adjustments suit short events of 1 to 7 days.
Google also warns that they may not work well over more than 14 days. That rules them out for the whole festive season. If you run a 3-day Diwali flash sale and expect conversion rates to jump, an adjustment can help bids react faster. For the long festive stretch, let Smart Bidding adjust. If Performance Max is the campaign that dropped, our Performance Max ROAS checklist goes deeper.
Watch ads, orders, stock and deliveries together every morning, not once a week. Most of the seven checks above need data from outside Ads Manager, which is why teams find these leaks days late.
This is the job Clevrr AI does. Klev, Clevrr's AI agent, checks Shopify, Meta Ads, Google Ads, inventory and logistics data every day. It splits each revenue change into four drivers: demand loss, value loss, operational loss and wasted ad spend. Then it puts a rupee value on each, so you know on Monday morning whether Diwali week needs a fix or just patience.
A ROAS drop during Diwali is a symptom, not a verdict. Compare like-for-like weeks, split cost from conversion rate, rule out learning resets and attribution lag, then check stock, discounts and RTO. Most festive drops trace back to one or two of these, and the fix for each is different.
If you want those checks run for you every morning, start free with 500 credits.
A small ROAS drop during Diwali can be normal, but it is not guaranteed. Competition for shoppers rises, which can lift CPMs, yet conversion rates often rise too. Criteo data showed conversion rates up to 22% higher before Diwali 2023. If ROAS fell sharply, check cost, conversion rate, stock and attribution before blaming the season.
No. Meta CPMs do not always rise in festive weeks. Jon Loomer's analysis of Meta's earnings found price per ad fell year on year through 2023 as impressions grew. Indian agency upGrowth estimates festive-week CPMs run 30% to 50% above off-peak, but that is an estimate. Compare your own CPMs with the same festive week last year.
It can. Meta's learning phase generally needs about 50 optimised events in 7 days, and a significant budget increase is one of the edits that can send an ad set back into learning. Smaller, steady increases carry less risk. Avoid stacking budget jumps, new ads and targeting changes on the same ad set in Diwali week.
Not before you know the cause. A blanket cut can reset learning on healthy ad sets and miss the second buying wave after Diwali. Pause ads on sold-out products, wait for attribution to catch up, and judge discounted campaigns on contribution margin. Cut spend only where the checks show the auction no longer pays back.
Join hundreds of D2C brands using Clevrr AI to automate their growth and efficiency.
Join 500+ D2C founders using Clevrr AI to automate their operations and scale faster.
Get Started Free
