Multi-channel attribution in India: which touchpoint actually drove the sale

Last-click attribution tells you the last ad a customer clicked before buying. It is the most commonly used model in India and it is also the most misleading one. It gives full credit to whatever channel happened to be last, and zero credit to everything that warmed the customer up.

If your Google Search ad is always last because customers come back to search your brand after seeing a Meta ad, last-click will make Meta look useless. You will cut Meta budget, Meta-warmed customers dry up, and then your brand searches drop too. I have watched this exact cycle play out at multiple Indian e-commerce companies.

How the customer journey actually looks in India

Indian customers typically need more touchpoints before buying than the global average suggests. Consider a customer buying a ₹15,000 air purifier. They might see a Swiggy Instamart ad, ignore it. See an Instagram reel from an influencer, save it. Google "best air purifier under 15000" and read three comparison blogs. Click a Google Shopping ad. Land on the website, abandon cart. Get an email with 10% off. Return via direct URL. Buy. Eight touchpoints across six days.

If you only count the email, you think email won the sale. If you also count last non-direct click, you think Google Shopping did. Neither captures the Instagram reel that started the consideration.

Attribution models available in Google Analytics 4

GA4 now uses data-driven attribution by default for conversion events, which is a significant improvement over Universal Analytics. Data-driven attribution uses machine learning to distribute credit across all touchpoints based on which ones actually correlate with conversions. It needs at least 400 conversions per month per conversion event to work well, which rules out smaller businesses.

For businesses under that threshold, the most useful model is position-based (also called U-shaped), which gives 40% credit to the first interaction, 40% to the last, and distributes the remaining 20% across middle touchpoints. This at least acknowledges that both discovery and close matter.

Time decay is useful if you run short promotional campaigns where recency genuinely matters, like a 48-hour Diwali sale.

The WhatsApp attribution problem

This is unique to India. A huge portion of Indian conversions happen after a WhatsApp conversation, but WhatsApp chats are almost invisible to standard web analytics. The customer clicks a "Chat on WhatsApp" button, has a conversation, then either returns to the website or buys through a payment link sent in chat.

Solutions: always UTM-tag every link you send inside WhatsApp conversations. Use a WhatsApp Business API provider like Interakt or WATI that logs conversation outcomes. If you send payment links via Razorpay or Cashfree, tag those links too. Even rough attribution is better than none.

Practical steps for mid-size Indian businesses

Start by pulling the multi-channel funnels report in GA4. Look at the top conversion paths. You will likely see paths with 3-5 steps, and you will probably be surprised which channels appear in the early steps.

Then compare your last-click attribution data with position-based attribution for the same period. Any channel whose attributed conversions go up under position-based is being undervalued under last-click. That is often awareness channels: Meta, YouTube, display.

For Google Ads specifically, switch to data-driven attribution in your campaign settings if you have enough volume. If not, use position-based. Both will give you a more honest picture than last-click.

Import your conversion data back into Meta Ads Manager using the Conversions API, not just the pixel. CAPI reduces the impact of iOS privacy changes and ad blockers, and gives Meta better signal to optimise toward actual buyers rather than clickers.

Frequently asked questions