Rough rule from running these accounts across India, the UAE and the US: you need one fresh creative asset per ₹20,000–22,000 of weekly ad spend to keep an Advantage+ shopping campaign from decaying. At ₹6 lakh a month, that's six to seven new assets every week. At ₹40 lakh a month it's forty-plus, which is why brands at that level stop briefing individual ads and start briefing shoots. In dollar markets the number is different because CPMs are different: roughly one new asset per $1,200–1,500 of weekly spend in the US or UK. Below that rate, frequency climbs, CTR slides, CPM creeps up as Meta pays more to force a tired ad in front of people, and your CPA drifts north over three to four weeks without any single day looking like the problem.
The rest of this explains where those numbers come from, so you can rebuild them for your own CPM and audience instead of trusting mine.
What Advantage+ shopping campaigns take away, and what that leaves you
Meta has been folding the standalone ASC flow into the regular sales objective with an Advantage+ toggle, but the mechanics haven't changed much. You get one campaign, one budget, broad delivery, an existing-customer budget cap, country and exclusion controls, catalogue integration, and a ceiling of 150 ads in the campaign. What you don't get: ad set level audiences, meaningful placement control, or the ability to protect a specific test from the algorithm's spend concentration.
Strip out audience targeting, lookalike laddering, placement exclusions and manual bids, and there is exactly one lever left that moves performance week to week. Creative. Advantage+ shopping campaigns turn media buying into a production scheduling problem. Most brands still staff it like a media buying problem, which is why performance holds for six weeks and then quietly stops.
The arithmetic behind the asset count
An asset doesn't die of old age. It dies of frequency against the slice of people Meta has decided are responsive to it. That slice is much smaller than your reachable audience. For a niche D2C brand in India selling, say, a ₹2,400 skincare set, the pool the algorithm actually works is usually in the low hundreds of thousands, not the 250 million Ads Manager quotes at you.
So define an asset's useful life as: effective audience × the frequency at which CTR starts falling. In our accounts that second number sits around 3 for statics and 4 for video. Take a 150,000-person effective pool and three exposures, and one static image is finished at about 450,000 impressions. Call it 400,000 to be safe.
Now put money against it. At a ₹220 CPM, 400,000 impressions costs 400 × ₹220 = ₹88,000. That's the spend it takes to burn one winning asset.
But you don't only ship winners. Our hit rate on new concepts for an established account runs about one in four. For a brand under a year old with no creative library, it's closer to one in six. At one in four, burning one keeper means shipping four assets. ₹88,000 ÷ 4 = ₹22,000 of weekly spend per asset shipped.
Change any input and the answer moves. A ₹450 CPM in a premium category doubles the rupees per asset and halves your production load. A one-in-eight hit rate doubles the load. Video with a genuine four-exposure tolerance buys you a third more life. Run your own numbers; the method matters more than my ₹22,000.
Weekly asset targets by budget
- ₹1 lakh/month (about ₹23k/week): one new asset a week. You can do this with an iPhone and a good editor.
- ₹3 lakh/month: three a week. Still manageable in-house if someone owns it.
- ₹6 lakh/month: six to seven a week. This is the level where most brands break, because it's more than a designer can do alongside other work and less than justifies a dedicated creative team.
- ₹15 lakh/month: fifteen a week. Needs a creator roster and a standing edit pipeline.
- ₹40 lakh/month: forty-plus a week. Nobody hand-makes forty ads. You shoot once a fortnight and cut modularly: eight hooks × three body sections × two end cards produces far more than forty, and the marginal cost of each is an editor's hour.
The uncomfortable part of that ladder is the ₹6 lakh rung. Brands scaling from ₹3 lakh to ₹6 lakh usually double the budget and keep the same creative cadence, then blame the algorithm five weeks later. Double the budget, double the assets, or don't double the budget.
Concepts, variants, and why the ratio is about 1:7
A variant is the same idea in a new wrapper: different hook line, new first three seconds, a fresh thumbnail, 4:5 recut to 9:16, price shown instead of discount. A concept is a different argument for buying the product: proof instead of promise, problem instead of product, founder instead of customer.
Variants extend the life of an idea the audience hasn't rejected. Concepts reset the clock when they have. Our working ratio is one new concept for every six to eight variants, which at seven assets a week means roughly one genuinely new idea a week and six executions of ideas that already work.
Inverting that ratio is the common mistake. Teams brief seven new concepts a week, get one winner, and throw away the winner's learnings by moving on to seven more new concepts next week. If a hook works, milk it. We've seen a single angle carry an account for four months through fourteen variants.
What 28 assets a month actually costs
Take the ₹6 lakh/month brand needing 28 assets. A realistic mix and Indian market rates:
- 6 UGC videos from creators at ₹10,000 each: ₹60,000
- 4 in-house shot videos, editing only at ₹3,000 each: ₹12,000
- 18 statics and carousels at ₹1,200 each: ₹21,600
Total: ₹93,600, or about 15% of media spend. Our planning rule is 12–18% of the media budget on creative production in India, higher in month one when there's no library. Under 10% and you're recycling, which shows up as rising CPM before it shows up in ROAS.
That budget line is the one finance pushes back on hardest, because it doesn't look like it buys anything. It buys the right to keep spending the other 85%.
Don't fill all 150 ad slots
The cap tempts people into dumping everything into the campaign. Delivery doesn't work that way. In a healthy Advantage+ shopping campaign, five to eight ads take the overwhelming majority of spend and the rest collect a few hundred impressions each and tell you nothing.
Keep 20–30 ads live. Retire anything that's had a real shot and failed, so the learning phase isn't diluted every time you upload. And name your ads properly: format, hook type, offer, aspect ratio, shoot date. Meta will not group these for you, and without a naming convention you'll have nine months of data and no way to answer "do problem-first hooks beat product-first hooks for us". We rebuild someone's naming convention on roughly every second account we take over.
Judging an asset without lying to yourself
Two rules. Kill an asset at 2× your target CPA with zero purchases. Don't declare a winner until it has five to eight purchases, whatever the ROAS says at three. A ₹4,000 spend and one sale at ₹6,000 order value is not a 1.5 ROAS; it's noise wearing a costume.
Then the India-specific correction. Meta optimises on the purchase event fired at order placement, and if you sell COD that event is a forecast, not revenue. Say COD is 55% of your orders and RTO on COD runs 28%. You lose 0.55 × 0.28 = 15.4% of orders. A campaign reporting 3.1 ROAS is delivering 3.1 × 0.846 = 2.62 on goods that actually reach a doorstep. If your break-even is 2.8, the campaign that looks profitable is losing money, and the algorithm is happily optimising toward the pin-codes that fake it best.
Fix it by sending a delivered-order event through the Conversions API and letting Meta see net values, or at minimum by tracking delivered ROAS in your own sheet and setting targets against that. Ad platforms optimise for whatever you feed them. Feed them gross COD orders and they will find you more COD orders that don't stick.
One more delivery setting worth an argument: the existing-customer budget cap. Left loose, Advantage+ shopping campaigns will happily spend 30% of budget re-buying people already on your WhatsApp and email lists at effectively zero marginal cost. During a growth push, cap it at 10–15% and judge the campaign on new-customer CAC, not blended ROAS.
The page the creative lands on
Cold Meta traffic on a 4G connection is the least patient traffic you will ever buy. Seven new assets a week driving to a PDP with a 3.8-second LCP is a subsidy to Jio. Before adding creative headcount, check what your product pages actually do on a mid-range Android over a throttled connection, not on the office wifi on a MacBook.
Usual culprits: an unoptimised hero image, four review apps loading synchronously, and a sticky cart drawer that ships 200KB of JavaScript. We've moved PDPs from the high threes to under 1.6 seconds mostly by deleting things. If you want a running check on where you stand, SwiftStore scans the store, fixes what's safely fixable and tracks the score over time; for the deeper theme-level work there's more detail on our speed optimisation page.
Where to start this week
Open Ads Manager, filter to the last 30 days, and count how many distinct assets took more than ₹10,000 in spend. Divide your monthly budget by that number. If the answer is above ₹1 lakh per asset, you are running on fumes and the decay is already priced into your CPA, you just haven't attributed it yet.
Then book the next shoot before you write the next brief. Production lead time, not idea quality, is what actually caps most accounts. If you'd like a second pair of eyes on the account and the store it points at, our free audit covers both.