Social Media Advertising in India: Pick the Platform by the Job

Choose by the job you need done. Every platform is good at something and sold as good at everything.

Most brands pick a platform the way they pick a restaurant: by how many people are already inside. Then they run the same square video everywhere and wonder why one channel returns anything at all.

Paid social is not one product. It is six or seven different auction systems that happen to share a word. This page is the routing layer of our performance marketing guide: which platform does which job in India, how each one charges you, what creative it will not run without, and how to test a new one without burning a quarter finding out.

The running example throughout is Kettle & Co., a hypothetical Pune D2C cookware brand selling a ₹2,400 pan, with one in-house content person and no video budget.

Start from the job, not the audience size

Audience size tells you almost nothing about whether a platform will work for you. What matters is whether the platform can optimise towards the outcome you actually want, and whether the buying decision you need happens in that context at all.

Write the job down in one sentence before you open any ads manager. Six cover most of what Indian brands are trying to do:

• Sell a physical product to someone who has never heard of you. Meta, because it is the only platform in this list where catalogue and purchase optimisation is the main event rather than an afterthought.
• Be remembered in three months. YouTube, bought through Google Ads, where the payoff lands later in branded search rather than in the campaign's own conversion column.
• Reach a job title at a company size. LinkedIn, and nowhere else. This is the one job no other platform can do properly.
• Be present while someone is researching. Quora, where the ad sits against a question that is already the consideration stage.
• Reach people whose screens are off. Spotify, during the commute, the workout, the cooking hour.
• Join a conversation happening today. X, for launches, live moments and category chatter.

Kettle & Co.'s job is the first one. That decides the platform before any audience data is opened, and it also decides what gets built: a product-led vertical video, not a brand film.

If you have not settled the prior question of whether to pay at all, that argument lives on our organic vs paid social page and is not re-run here.

What each platform is actually good at in India

Every row below was checked against that platform's own documentation on 24 September 2026. Meta's Business Help Centre blocks automated fetching, so Meta is cited from its Marketing API developer documentation instead.

Snapchat reaches a younger audience and ShareChat a vernacular one, but we have no setup guide for either yet, so we are not going to pretend to route you there.

The setup detail for each of these lives on its own page: Instagram Ads, Facebook Ads and the Meta Ads sub-hub for the Meta side, YouTube Ads, the LinkedIn Ads guide, X Ads, Quora Ads and Spotify Ads. This page does not repeat any of it.

What you are billed for is a different setting from what you optimise for

Meta's own developer documentation separates the two explicitly: the optimisation goal decides who gets shown the ad, the billing event decides what you are charged for, and Meta's bidding overview states that neither of them directly changes your bid amount. Switching what you are billed on does not repair a bad cost per result. It changes the denominator on your report.

The second trap is assuming a metric name means the same thing everywhere. It does not.

• A YouTube view is 30 seconds, or the whole video if it is shorter, or an interaction. On Shorts it is 10 seconds.
• A LinkedIn CPV is only available if you chose the Video Views objective in the first place.
• An X pre-roll view and an X video view are billed under two different objectives.

So a spreadsheet column headed "cost per view" comparing three platforms is comparing three different things. Compare cost per outcome you defined, or do not compare.

Creative is the real cost line, not the media

The budget most brands underestimate is not the spend. It is the production.

Kettle & Co. wants Meta, YouTube and Spotify. That is not one asset in three sizes. It is a vertical product video with a two-second hook, a 16:9 cut with a different opening, and a 30-second audio script that works with no picture at all. Three briefs, three edits, three approval rounds, every month, because Meta burns through hooks.

This is the honest reason a brand should not launch on four platforms at once. Not budget. Bandwidth. One platform done with fresh creative every fortnight beats four platforms running the same tired asset, every time we have seen it tried.

When a client asks us to run paid social across platforms, the first thing we scope in our performance marketing engagements is creative volume, because that is the constraint that decides how many platforms are realistic. Actual pricing sits on our social media marketing cost in India page, deliberately not here.

Why your platform totals never add up

Add up the conversions reported by Meta, Google and LinkedIn, and you will get a number larger than your actual orders. Nobody is lying. Every platform reports the conversions it can claim inside its own window, and no platform can see the others.

Google Ads makes this explicit: data-driven attribution is the default model for most conversion actions, distributing credit from that account's own past data. It cannot see what Meta showed the same person yesterday, and Meta's reporting has the same blind spot in reverse.

The working rule we use:

• One source of truth for the total. Your order system or GA4 decides how many sales happened. Not the sum of the ad platforms.
• Platform numbers for decisions inside that platform. Meta's data is the right tool for choosing between two Meta creatives, and the wrong tool for choosing between Meta and LinkedIn.
• Blended cost per acquisition for the channel decision. Total spend divided by total orders from the source of truth. Ugly, and the only number that cannot be double counted.

The budget floor below which paid social does not work

No platform publishes one, so here is our arithmetic. A rule of thumb, not a benchmark.

Take your target cost per result. Multiply by 30, because roughly one result a day is the least that makes a weekly read mean anything. Multiply by the number of ad sets you want running, which should be two at most when you start. That is your monthly floor for one platform.

For Kettle & Co., targeting a ₹400 cost per purchase: ₹400 × 30 × 2 ad sets = ₹24,000 a month for Meta alone, before a rupee of production. Add a second platform and the floor roughly doubles. Below that, you are not running a channel, you are buying a sample size too small to read.

One published minimum is worth knowing: Spotify sets a floor of 250 in your local currency per campaign, which is the only hard number any of these six platforms publishes.

Our view: the platform most Indian brands should stop advertising on

This next part is opinion, from running paid social for Indian brands, not a finding from data.

We think most Indian D2C and lead-generation brands should stop buying ads on X.

The reasoning sits in X's own objective list, quoted in the table above: Reach, Video Views, Pre-Roll Views, App Installs, Engagement, Website Traffic. No lead generation objective. No catalogue sales objective. If your outcome is a form fill or a purchase, the platform cannot optimise towards it, so you buy engagements or clicks and hope the correlation holds. It usually does not.

Most Indian brands we see on X are there for founder visibility rather than demand, and that is a communications budget wearing a performance budget's clothes. There are real exceptions: B2B software launches, event moments, and categories where the conversation genuinely lives on the platform. Our X Ads guide covers those cases properly. For a cookware brand in Pune, that money belongs on Meta.

How to test a new platform without wasting a quarter

Seven steps, in this order. Most failed platform tests skip step two and lose the budget at step six.

1. Write the job in one sentence. "Sell the ₹2,400 pan to people who have not heard of us." If you cannot write it, you are not ready to spend.
2. Open the objective list before the audience builder. If no objective matches your job, stop here. This is the step that would have saved most X budgets.
3. Fund the floor for one platform only. Target cost per result × 30 × two ad sets. If you cannot fund it, test a cheaper outcome or wait.
4. Build native creative for that platform. Three variants minimum on Meta, a real script on Spotify, a headline test on Quora. A resized asset is not a test of the platform.
5. Fix the measurement rule before launch. Name the source of truth, tag every URL, write down the blended CAC you need. Deciding this after the numbers arrive is how teams argue for a month.
6. Run 21 to 28 days without touching budgets. Every edit restarts the optimisation. Change creative if a variant is clearly dead; leave the structure alone.
7. Judge on blended cost, then pick one of three. Keep and scale, keep and rebuild the creative, or close it and put the money back on the platform that already works. "Run it a bit longer" is not one of the three.