India digital ad benchmarks 2026: CPM, CAC and payback by category

Every Indian founder asks the same first question and almost nobody publishes an honest answer: what should this cost? Agencies avoid the question because a published benchmark is a rod for their own back. So here is ours, drawn from the accounts we run and the accounts we audit for prospective clients.

**Methodology and limits.** These are ranges observed across live Meta, Google and marketplace accounts in Indian D2C, SaaS, healthcare, education, real estate and local-services categories during the last four quarters. They are not a national census. Spend levels vary from ₹2 L to ₹3 Cr a month. Where a range is wide, it is wide in reality — category, offer quality and creative velocity move these numbers more than any targeting setting. Treat them as a sanity check on your own numbers, not as a target to chase. If your figures differ materially, we would genuinely like to see them.

**City tier changes CPM more than category does.** Delhi NCR, Mumbai and Bangalore consistently sit at the top of the Meta CPM range; Gurgaon specifically runs well above the national average because advertiser density per capita is the highest in the country. Pune, Hyderabad and Chennai sit in the middle. Kolkata, Jaipur, Bhopal, Patna and Ranchi routinely clear at roughly half of North Indian metro levels. The practical implication: identical budgets buy category leadership in an under-served city and marginal visibility in a saturated one. Most national brands allocate as if the map were flat.

**Google search costs track intent, not geography.** A bottom-funnel commercial term costs broadly the same in Kolkata as in Gurgaon, because the auction is national for most service categories. What changes is competitor count. Terms with fewer than five serious local advertisers — very common outside the top four metros — are where cost per qualified conversation collapses.

**Platform ROAS overstates reality, consistently.** In nearly every account we have audited, Meta-reported ROAS overstates incremental revenue by roughly a third to a half. The mechanism is not fraud; it is attribution windows, view-through credit and modelled conversions all counting demand that would have arrived anyway. Any account that has never run a holdout or geo-split test does not know its true number. That single missing test is the most expensive gap in Indian performance marketing.

**CAC is under-reported by most Indian teams.** The common definition — media spend divided by conversions — excludes agency fees, creative production, tooling, salaries, shipping subsidies and returns. Fully loaded CAC is typically a third to a half higher than the number in the monthly deck. Rebuild it once with your finance team and most 'profitable' channels re-rank immediately.

**Payback period is the number that decides whether you can scale.** Under six months is excellent and means you can reinvest aggressively. Six to twelve is healthy. Twelve to eighteen is workable only with patient capital. Over eighteen months means you are growing on cashflow rather than economics, and a single bad quarter ends the strategy.

**Creative velocity beats targeting sophistication.** Across the accounts we run, the strongest single predictor of falling CPA is not audience structure or bidding strategy — it is how many genuinely distinct hooks ship each month, and how quickly the losers are killed. Twelve to twenty new hooks a month, judged on CPA rather than on taste, outperforms a beautifully engineered account with four creatives a quarter. This is the cheapest lever available to most Indian brands and the one most agencies under-deliver on because production is where their margin leaks.

**Blended MER is the metric that survives attribution decay.** Total revenue over total marketing spend, all channels in. Under 2 usually signals a marketing problem. Over 5 usually signals under-investment rather than excellence. It is the one number that cannot be gamed by a platform's reporting choices, which is precisely why so few monthly decks lead with it.

**What we would do with these numbers if we were the client.** Rebuild fully-loaded CAC with finance. Calculate payback by cohort, not in aggregate. Run one geo holdout to learn your true incrementality multiplier, and apply it to every platform number thereafter. Reallocate budget by payback period rather than by ROAS. Then raise creative volume until the production process, not the media budget, becomes the constraint.

We will re-publish this study every quarter and widen the sample as we go. If you want your own numbers checked against it, that is exactly what our free audit does.