CPM is not a cost. It is a price.
You do not set it, you cannot negotiate it, and it is decided by every other advertiser bidding for the same person at the same moment. Which means almost every article about lowering your CPM is aimed at a variable you do not own, and that is why following them feels like effort with no result.
What you own is everything that happens after the impression. Whether the person stops. Whether they click. Whether they buy. Whether they come back. Those four are the entire game now, and the rising auction price is simply the tax you pay to enter it.
So the honest version of this question is not how to make Meta cheaper. It is how to make each impression worth more than it was last year, faster than the auction price rises.
What is actually moving
The benchmark round-ups published this year put cross-industry Meta CPMs up roughly 20% year on year going into 2026, with price per ad growing on top of impression growth. Those are largely global and US figures, so do not import the absolute numbers.
India sits far below them in absolute terms. The Indian agency benchmarks put baseline CPMs somewhere around Rs 85 to Rs 155 across major sectors, with lead objectives higher than awareness ones. Cheap by world standards. Moving in the same direction.
Two India specific things worth planning around.
The October and November festive window runs CPMs 30% to 50% above normal, because every retailer in the country is bidding at once. That is not a bad quarter, it is a different economy, and judging your annual efficiency on Diwali numbers will send you to the wrong conclusion.
And Reels placements run meaningfully cheaper than feed, commonly quoted at 25% to 40% below. If your creative is not built vertical first, you are choosing the expensive inventory.
The lever, and the data is unusually blunt about it
Creative volume. Not creative quality in the awards sense. Volume.
The benchmark studies this year make the case better than any argument I could write. Average new creative output runs about 2.8 per week for the smallest spenders and up to roughly 19 per week for the largest. Fine, bigger budgets make more ads.
Here is the part that matters. Within every single budget tier, the top quartile launches two to three times more creative than its same-budget peers. Same money. Two to three times the output.
And a study across more than 200 direct to consumer accounts found that brands shipping 30 or more new creatives a month scaled roughly three times faster than those shipping fewer than ten.
That is not a budget difference. It is an operating capacity difference, and it is the one thing in this entire discussion that is genuinely within your control.
Your competitor is not beating you because their media buyer is better. They are beating you because they shipped forty things last month and you shipped six.
The reason this works is mechanical rather than mystical. The platform decides who sees your ad largely from the creative itself now, so each new concept is a fresh attempt at finding an audience you have not reached. Six attempts a month is not a testing programme. It is a hope.
What high volume actually looks like in India
Not a production house. That is the assumption that stops most founders before they start.
Phone footage. The founder talking. A customer talking. The product being unboxed on a desk. Somebody in your warehouse showing how an order gets packed. Three second hooks, vertical, subtitled, because most of this is watched with the sound off on a crowded train.
Concepts, not variations. Twenty colour changes of the same ad is one creative. A testimonial, a problem demo, a price comparison, a founder story and a use case are five.
Set a mechanical rule so it survives a busy week. One new concept per fixed slice of spend, reviewed weekly, shipped whether or not anybody is inspired. Inspiration is not a production schedule.
And retire losers on evidence rather than sentiment. Most of what you make will not work, which is the correct outcome, because the alternative is that you are not trying anything unfamiliar.
Where I disagree with the standard advice on measurement
The usual counsel is to stop looking at platform ROAS and start looking at blended CAC. Half right, and the missing half costs people a year.
Platform ROAS overstates. It claims credit for people who were coming anyway, particularly anyone who already knew your brand.
But blended CAC understates, and nobody says this loudly enough. Blended divides all your marketing spend by all your customers, including the ones who arrived through organic, WhatsApp, word of mouth and repeat purchase. Benchmark analyses this year put paid customer acquisition cost at roughly 2.4 to 3.1 times blended CAC. A brand feeling comfortable about its blended number can be paying close to three times that for each genuinely new customer bought through ads.
So report three, not one.
Platform ROAS, treated as a directional signal about creative, not as truth about profit.
Blended CAC, as the business level sanity check.
New customer acquisition cost, meaning paid spend divided by first time buyers only. This is the number that tells you whether the machine actually works, and it is the one most Indian D2C brands I talk to have never calculated.
If you only watch blended, growth from organic and repeat will mask a deteriorating paid engine for as long as the other channels keep growing. The month they stop, the problem appears fully formed and looks sudden. It was not sudden.
The part that actually decides survival
Retention, and it is unglamorous enough that it gets discussed last in every conversation including this one.
If the auction price rises every year and your repeat rate does not move, you are running up an escalator that speeds up annually. There is no creative strategy that outruns that indefinitely.
Lifetime value is the only variable on your side that compounds. A second purchase from an existing customer costs you a WhatsApp message. A first purchase from a stranger costs you the going rate, and the going rate is what we have spent this whole article talking about.
Practically: know your repeat rate, know it by cohort, and treat a one point improvement in it as worth more than a one point improvement in ROAS. Because it is.
Build the second channel before you need it
The last argument, and the one founders resist because it costs money during a good quarter.
A business with one acquisition channel is a business with a single point of failure priced by somebody else. Meta does not have to fail you. It only has to get expensive enough that your margin stops working, which is a slow, quiet process that looks fine right up until it does not.
The second channel does not need to be big. It needs to exist, with somebody responsible for it, before you need it. Search, for the demand that already exists. A WhatsApp list you own outright. Marketplaces, if the margin arithmetic works. Quick commerce, if your category fits.
I went through how to decide between Google and Meta in the first place separately, and the budget floor arithmetic decides whether a second channel is fundable yet or just a distraction.
The short version
You cannot control the price of the impression. You can control how many swings you take, what you count, and whether the customer comes back.
Ship more creative than feels reasonable. Calculate your new customer acquisition cost this week, because it is probably not what you think. Then spend some of the money you were going to put into ads on making people buy twice.
My notes on performance marketing cover the rest. If you’d like to think through your own numbers, drop me a line on email, WhatsApp or LinkedIn and we can have a quick chat. I’m contracted full time so this isn’t a pitch.