Marketing Attribution in India: Measuring What Actually Drives Revenue

Why platform numbers can’t be trusted alone

Every ad platform is graded on the sales it can claim, so every ad platform claims generously. Meta counts a sale it touched, Google counts the same sale, and your email tool counts it too, add them up and you’ve credited one order three times. Layer on privacy changes that made cross-platform tracking patchier, and the individual numbers drift further from reality. This isn’t dishonesty so much as everyone marking their own homework.

The main models, and what each is good for

There’s no perfect model, only trade-offs:

• Last-click — credits the final touch. Simple, but ignores everything that warmed the buyer.
• First-click — credits discovery. Good for understanding demand creation, bad for the closing channels.
• Multi-touch — spreads credit across touches. More realistic, more complex, and still assumption-heavy.
• Blended / top-down — ignore platform claims and look at total spend vs total revenue, plus incrementality tests. Coarser, but honest.

How to build a view you can actually trust

1. Anchor on your own revenue — your store or CRM is the source of truth, not any platform.
2. Use a blended baseline — total marketing spend against total revenue tells you the honest overall ratio.
3. Add a model for direction, not gospel — multi-touch to see the assist channels, held loosely.
4. Run incrementality tests — turn a channel down and see what actually happens to revenue.
5. Report to decisions — the point is where the next rupee goes, not a dashboard nobody acts on.

What most teams get wrong

The biggest mistake is summing platform-reported conversions and treating the total as real. It double-counts and flatters every channel at once. The second is worshipping last-click, which quietly defunds the top-of-funnel channels that create the demand the closing channels harvest; starve those and last-click ‘wins’ right up until sales dry up. The third is building an elaborate attribution model and then never using it to move a budget, measurement that doesn’t change a decision is a hobby.

The contrarian take: a rough honest number beats a precise fake one

Teams chase attribution precision as if the right model will reveal the truth to the decimal. It won’t. Every model rests on assumptions, and the data underneath is getting noisier, not cleaner. A blended view that’s directionally honest, ‘we spent this, we made that, this channel looks incremental’. Will make better budget decisions than a multi-touch model presented with false confidence. Accept the fog, triangulate, and test. Certainty is the thing being sold, not the thing being measured.

What tends to improve — a realistic picture

• Business type: a multi-channel advertiser trusting each platform’s self-reported numbers.
• Common problem: reported conversions that far exceed real revenue, and budget decisions made on them.
• Typical approach: anchor on owned revenue, build a blended baseline, add a directional model, and run simple incrementality tests.
• What tends to improve: more confident budget decisions and less spend chasing double-counted credit. Outcomes vary with channel mix and data quality.

Sanity-check your reporting in 15 minutes

1. Do your platforms’ total claimed conversions exceed your actual orders? (They usually do.)
2. What’s your blended number, total spend divided into total revenue?
3. Which model are your budget decisions actually based on, and does everyone know?
4. When did you last turn a channel down to test whether it was incremental?
5. Does your reporting ever change where the money goes, or just describe the past?

If the answers are uncomfortable, that’s the work. It’s the core of our AI & Data Analytics practice, and it pairs with the first-party data shift and what a good ROAS really means.