On this page
The core difference: intent vs. interruption
Pull channel
Google Ads
Someone types "best magnesium glycinate India", you show up, they click, they buy. Demand already exists. You compete on ad rank, landing page, and price.
Push channel
Meta Ads (Facebook + Instagram)
Nobody woke up looking for your new mead brand or ayurvedic retreat. Creative interrupts their feed and manufactures the want. You compete on hook, offer, and creative velocity.
ROAS benchmarks we actually see
Across the D2C, wellness, hospitality and AlcoBev brands we run, these are the ranges to plan against. Not guarantees, but honest reference points.
| Channel | Cold ROAS | Blended ROAS | Best for |
|---|---|---|---|
| Google Search (non-brand) | 2.5–5x | 4–8x | Category-aware buyers |
| Google Search (brand) | 8–20x | 10–25x | Defending owned demand |
| Google Shopping / PMax | 2–4x | 3–6x | Wide catalogue e-comm |
| Meta Prospecting | 1.2–2.5x | 2.5–5x | New brands, launches, creative-led |
| Meta Retargeting | 4–10x | No figure | Warm audiences, cart abandon |
A note on these numbers: the ranges above come from our own accounts. Actual ROAS varies significantly by category, season, location, product, and price, among other factors. Treat these as a sanity check to plan against, not a target to promise a client or measure your brand against. Create your own benchmarks. India-specific benchmarking on this is still rare; we're publishing ours because somebody should.
Highest-ROAS line item in almost every account
Branded search. If someone is Googling your name, buy that click before a competitor does. Turn it on before anything else.
When Google wins
You're in a searched category.
Supplements, mattresses, travel, insurance, B2B software. If people type your keyword, Google is the cheaper first click.
You have a brand people search by name.
Branded search is the highest-ROAS line item in almost every account we run. Turn it on before anything else.
Long consideration + comparison purchases.
Buyers Google 'X vs Y', read reviews, then buy. Meta creative can't shortcut that.
When Meta wins
Visually strong products.
Food, beauty, fashion, home décor, hospitality. A 6-second reel outperforms a text search ad every time.
You can produce creative weekly.
Meta rewards creative velocity. If you can't ship 8–15 new hooks a month, don't over-invest here.
Category creation
Cosmix had to create demand for superfoods and adaptogens as a category, not just for their brand within it. Zero-waste personal care brands like Bare Necessities face a similar challenge: the sustainability angle is not enough on its own, the product itself has to be introduced.
The clearest example, from the inside
QuristQurist's cannabis-for-sleep launch. Magnesium was an established, well-searched sleep ingredient, but cannabis for this use case had virtually no search demand in India at all. We restructured the account for a full-funnel approach: landing-page-view awareness campaigns (which outperformed the link-click-only setup they had used for other product lines), paired with education content explaining what the product actually was and did. We also shifted their budget structure away from concentrating spend on their one bestseller, toward separate, dedicated budgets for category-specific campaigns. That combination built real demand for cannabis-as-a-sleep-aid from scratch, and drove real revenue for it before most people were even searching the term.
Practically, category creation means generating search intent that does not yet exist, and it has to happen everywhere the category will eventually be searched for, not just on Google. That means seeding intent across Amazon, Blinkit, Flipkart, quick-commerce apps, and e-commerce marketplaces, as well as Google itself, so that by the time someone hears about the category from a friend or an ad, the platforms are ready with an answer when they go looking.
One layer worth flagging honestly: today's buyers increasingly do that "going looking" inside AI search tools rather than typing into Google. That shift, and what it means for how brands get discovered, is a big enough topic that it deserves its own guide.
Our decision framework
Search volume check
Pull monthly search volume for your top 20 non-brand keywords. Under 5,000/month combined? Meta first.
Creative capacity check
Can you produce more than 10 fresh static/video assets per month? If no, Google-heavy. If yes, Meta unlocks.
Margin check
Under 40% contribution margin post-COGS? You need Google's higher intent to hit break-even ROAS. Meta prospecting will bleed.
Attribution honesty
Don't judge Meta on last-click. Use a 7-day-click / 1-day-view window and blended MER (revenue ÷ total ad spend) as the real scorecard.
Judge the account, not the platform
Multiple times, clients have come to us convinced they need to scale one specific platform, usually because a competitor is doing well there, or because a channel feels more modern. But when we dig into the actual business, we often find the real move is to launch on the platform they had written off entirely.
This is why we calculate ROAS at the account level, not the platform level, while still keeping attribution honest per channel. The reasoning is straightforward: platforms do not compete for your business, your customer's actual behaviour does. A customer might discover you on Meta, forget about you for two weeks, then convert through a Google branded search when they are ready to buy. Judge each platform in isolation and you will optimise for the wrong thing every time. Judge the account as a whole and the platforms start looking less like rivals and more like a relay team.
Judge the account as a whole and the platforms start looking less like rivals and more like a relay team.
Assumption reversed
Moonshine Meadery
The founding assumption was that their audience skewed 25-plus and lived on Instagram, so Meta looked like the obvious, almost exclusive, channel. When we tested Facebook properly instead of writing it off, 95% of purchases came through Facebook, from an audience mostly above 35. The assumption was not just slightly off, it was backwards.
₹5L → ₹12L / month
Studio13
Came to us believing Meta was their most profitable channel and wanted to scale it further. When we looked at the actual data, we found real, untapped search volume on Google that nobody was capturing. We tested PMax, YouTube, and Search alongside their existing Meta activity. Had we simply scaled the channel they already believed in, that demand would have stayed uncaptured.
Both stories point at the same lesson: the platform a client believes in is not always the one doing the real work, and the only way to know is to look at the account as a whole, not defend a channel because it is the one everyone already trusts.
The budget-split starting points we use
20% Google branded + Shopping, 80% Meta prospecting + retargeting. Prove the creative can sell.
35% Google (branded, non-brand, PMax), 55% Meta (prospecting + creative testing), 10% experimental (YouTube, influencer whitelisting).
50/50. Google absorbs demand efficiently. Meta funds category growth.
Common mistakes we clean up
- Running Meta without in-house or agency creative production. It always fails.
- Not bidding on your own brand terms. Competitors will.
- Judging PMax too early. My view is simple: let the spend do the work, and give it enough time to learn before judging it. If a business wants to review results seven days after launching PMax, honestly, they shouldn't be doing performance marketing at all. This connects to something I believe in more broadly: failing fast, meaning testing with real volume and real intent across different messaging angles and audiences before drawing conclusions. Let the sales start coming in first; only once they do does a real conversation about CAC and efficiency become useful.
- Chasing platform-reported ROAS instead of MER. Meta and Google both over-claim.
How we'd run your account
We've spent 9+ years running Meta and Google budgets for e-commerce brands, from ₹2L/month experiments to eight-figure yearly spends. Case studies from Cosmix, Moonshine, and Qurist walk through the exact channel splits and ROAS we hit.
