The learning phase costs the same number of conversions in November as it does in August. Each one just costs several times more.
That single fact is what separates festive campaign structure from ordinary campaign structure. Restructuring in a quiet month costs you a few days of inefficiency. Restructuring during Diwali week costs you the most expensive conversions you will buy all year, spent on teaching an algorithm something you could have taught it in September.
Diwali 2026 falls on Sunday 8 November. Here is how to structure for it, and how to make sure Meta's numbers and yours agree afterwards.
What Meta actually costs in India
Before the structure question, the baseline. Across roughly 2.6 million Meta-attributed installs at 39 India-based apps on Linkrunner between May and July 2026, with around Rs6.5 crore of Meta spend:
-
Median Meta cost per install: around Rs24
-
Middle half of apps: roughly Rs11 to Rs43
-
Median activation rate on Meta traffic: around 70 per cent
The comparison worth having is against all networks. Across the same customer base measured on all connected networks, the median cost per install was just under Rs20 and median activation was around 67 per cent.
So Meta runs roughly a quarter more expensive per install than the blended cross-network median, and delivers slightly better-activating traffic for it. That is a trade most teams would take, but it is a trade, and it is worth knowing which side of it you are on before you concentrate a quarter of your annual budget there.
Methodology: Linkrunner projects with a connected Meta account and INR base currency, three months to 31 July 2026. Blended figures divide connected Meta spend by installs attributed to Meta. Activation is each project's own defined onboarding event, so definitions vary by app. Bands, not targets.
Why structure matters more during festive
Festive Meta campaign structure is the arrangement of campaigns, budgets and creative for a compressed high-spend period, organised so the algorithm's learning phase completes before auction prices peak rather than during them.
Two things change:
-
Learning is priced in conversions, and conversions get dearer. Exiting learning requires roughly the same conversion volume whenever you do it. In late September that volume costs one price. In Diwali week it costs considerably more.
-
You have six weeks, not six months. Every restructure resets learning. In an ordinary quarter you can afford two or three resets and still finish ahead. Across a six-week window with three distinct spend waves, you can afford roughly none.
This is why structure decisions you would make casually in March become budget decisions in October.
Consolidate before the window, not during
The most common festive structure mistake is proliferation: a new campaign for Big Billion Days, another for Navratri, another for Dhanteras, another for each major offer. It feels organised. It is expensive.
Fewer, better-funded campaigns exit learning faster because the conversion volume concentrates rather than splitting. Splitting a fixed festive budget across eight campaigns can mean none of them ever leaves learning, and you spend the entire window paying learning-phase prices.
A separate campaign is genuinely justified when:
-
The optimisation event is different, for example installs against purchases
-
The audience is genuinely non-overlapping, such as acquisition against reactivation
-
The budget owner is different and needs separate reporting
Different creative is not a reason for a different campaign. Different festival dates are not a reason for a different campaign. Both belong inside one structure with rotating creative. Our budget allocation framework covers splitting across networks rather than within them, which is the split that does earn separate treatment.
Advantage+ or manual for festive
The honest answer is that it depends on how much you already know, and festive is a bad time to find out.
Advantage+ handles volatility well, and festive is volatile. Auction prices, audience behaviour and creative performance all move faster than a manual structure can be adjusted by hand. If your team is small or your festive priors are weak, automation is usually the better default.
Manual gives control where you have a strong prior, and by September you should have one. If you know which audiences and placements converted last festive, a manual structure lets you act on that rather than asking the algorithm to rediscover it at peak prices.
The practical answer: test both in September while inventory is affordable, then run the winner from October. Do not switch between them in November. Our comparison of Advantage+ Shopping and manual campaigns for app growth sets out the decision in more detail.
Budget pacing across the three waves
India's festive window has three distinct spend waves, not one peak:
-
Roughly 23 September to 2 October: Flipkart Big Billion Days, with Amazon's Great Indian Festival overlapping
-
Mid to late October: Navratri and Dussehra
-
6 to 10 November: Dhanteras, Diwali on the 8th, Bhai Dooj
Pace against those rather than against the calendar month. Two rules:
Release budget in tranches, tied to waves. Each tranche unlocks against the previous wave's results, not against optimism. Front-loading the whole festive budget in September leaves you nothing to respond with when Diwali week behaves differently from expected.
Raise budgets in small steps. Large budget increases can push a campaign back into learning, which during festive is the most expensive possible self-inflicted wound. Smaller, more frequent increases achieve the same total spend without the reset. This is the single most common way teams lose money to structure during festive.
Hold 15 to 20 per cent of the festive budget unallocated so you can scale the one thing that is working on 7 November rather than watching it hit its cap.
Creative rotation inside a validated set
Frequency climbs fast during festive, so creative fatigues in days rather than weeks.
The instinct is to introduce fresh concepts when performance dips. During festive that is usually wrong, because an untested concept at peak CPM is an expensive experiment with no time to read the result.
Do this instead: test several concepts in September using a five-day creative testing sprint, keep the winners in reserve, and rotate within that validated set as each one fatigues. You are managing a bench, not writing new plays mid-match. Our guide to creative fatigue detection and refresh cycles covers the thresholds that tell you when to rotate.
Making Meta's numbers and yours agree
This is the section that decides whether any of the above is knowable, and it is where most festive Meta guides stop after a sentence.
Meta is a self-attributing network. It applies its own attribution rules and windows and reports conversions it believes it caused. Your measurement platform applies different rules to the same events. The two will not match, and during festive the gap widens for three reasons:
-
View-through attribution volume rises with impression volume, and view-through is where the two systems disagree most.
-
Attribution windows collide with sale lengths. Big Billion Days runs roughly ten days against a typical seven-day click window, so conversions from early-sale clicks fall outside it.
-
Spend concentration amplifies everything. A gap you could ignore at ordinary volumes becomes a material number when a quarter of the annual budget flows through it.
Three things to configure before the window:
-
Verify Meta postback mapping against a live test conversion rather than assuming last year's setup survived. The Meta Ads integration docs cover the configuration.
-
Confirm Meta Install Referrer is enabled on Android. This is the most consequential single setting for Android-heavy Indian apps and the most commonly missed. Without it, Meta-driven Android installs are systematically under-attributed, which during festive means quietly defunding your best channel. The requirements are documented.
-
Record your baseline gap in September. Measure the difference between Meta-reported and independently-measured conversions before spend climbs. A gap is normal. A gap that changes shape mid-window is a problem, and you can only see that against a baseline.
If events are not appearing as expected, our guide to troubleshooting Meta SDK event attribution works through the common causes. When the two systems disagree at the end of the window, which one to trust depends on the decision you are making.
Having an independent view of Meta-attributed installs, activation and revenue is what turns that gap from a mystery into an explainable difference. If you want to see yours before the window opens, request a demo.
Decide the structure in September
None of this is hard. All of it is time-sensitive.
Consolidate, choose Advantage+ or manual, test creative, and verify Meta measurement, all before the last week of September. From October you are executing a plan rather than building one at peak prices.
FAQ
How many Meta campaigns should I run during Diwali?
Fewer than instinct suggests. Splitting a fixed festive budget across many campaigns can mean none accumulates enough conversion volume to exit the learning phase, so you pay learning-phase prices throughout. Separate campaigns are justified by a different optimisation event, a genuinely non-overlapping audience, or a different budget owner. Not by different creative or different festival dates.
Should I use Advantage+ or manual campaigns for festive?
Advantage+ handles volatility better, which suits festive, and is usually the right default for smaller teams or weak festive priors. Manual pays off where you have strong evidence from previous festive windows. Test both in September when inventory is affordable and run the winner from October rather than switching in November.
How much can I increase a Meta budget without resetting learning?
Prefer smaller, more frequent increases to large step changes. Large increases risk pushing a campaign back into learning, which during festive means re-buying the most expensive conversions of the year. The same total spend delivered in smaller increments avoids the reset.
Why do Meta's conversion numbers differ from my MMP during festive?
Meta is a self-attributing network applying its own windows and rules, and the disagreement concentrates in view-through conversions, which rise with impression volume. Sale periods that run longer than the click window widen the gap further. Establish your baseline gap in September so you can tell a normal divergence from a broken one.
Should I launch a separate campaign for each festive offer?
No. Each offer needs its own creative and its own landing destination, but they belong inside an existing campaign structure. A new campaign per offer fragments conversion volume across the window when concentrating it is what gets you out of learning.