A customer sees your Instagram ad on Monday, searches your name on Google on Wednesday, clicks a Google ad on Friday and buys. Who gets the credit? Marketing attribution models answer that, and the answer decides where your next rupee goes. This guide is for owners and marketers in India who run ads on Google and Meta. You'll learn what each model does, why Google and Meta never agree on your numbers, and how to pick a model without a data team.
What are marketing attribution models?
Marketing attribution models are rules or algorithms that decide how credit for a sale or lead is shared among the touchpoints a customer met on the way. Google defines attribution as assigning credit for important actions to different ads, clicks and factors along the user's path. A model can give all the credit to one touchpoint or split it across several. Last click, where the final ad takes everything, is the simplest. Change the model, and the same sale can make a different channel look like the winner.
A touchpoint is any contact with your brand, such as an ad view, an ad click or a search visit.
Why do attribution models matter for Indian businesses?
Attribution models matter because small businesses spend a large share of India's digital ad money, and the model you pick decides which channel gets the next rupee.
About ₹363 billion of digital ad revenue in 2025, about 38%, came from small businesses and long-tail advertisers.
Example: a Pune coaching institute runs an Instagram ad on Monday. A student searches its name on Wednesday, clicks a Google ad on Friday and enrols. Last click gives Google every bit of the credit. The owner cuts Instagram, and a few weeks later fewer people search the institute's name. That's the risk of reading one number as the whole story. The channel that starts journeys can look useless on paper.
How do attribution models work?
Attribution tracks touchpoints, links them to a key event such as a sale or a booked call, and then applies a rule to share the credit.
Google Analytics 4 uses key events for this. Google Ads reports a conversion on the date of the click, so last week's results can still grow this week. Google Analytics 4 can reattribute conversions for up to 7 days after the conversion, so a number you read today can change by next week. Read every report with that delay in mind. Meta counts actions within a window, 1 or 7 days after a link click and 1 day after an impression.
Our conversion funnel guide shows the stages these touchpoints sit in.
Which attribution models exist today?
Five models matter for a business running Google and Meta ads, and the table shows who gets the credit in each.
| Model | Who gets the credit | Where you'll see it |
|---|---|---|
| Last click | The final ad that was clicked | Google Ads, Google Analytics 4 |
| Paid and organic last click | The last non-direct channel | Google Analytics 4 |
| Data-driven | Split by machine learning on your data | Google Ads, Google Analytics 4 |
| Meta standard | Ads clicked or viewed inside the window | Meta Ads Manager |
| Meta incremental | Conversions an ad likely caused | Meta Ads Manager |
Older guides still list first click, linear, time decay and position-based. Google Ads says those four are no longer supported, and accounts that used them were upgraded to data-driven.
Why do Google and Meta show different numbers?
Each platform counts credit in its own way, inside its own window, and each one claims the sale for itself.
Meta warns that attribution models use different counting mechanisms, so comparing across them leads to inaccurate conclusions. Google Analytics 4 adds another twist, because all its models leave direct visits out of the credit.
So one customer can appear twice, once in Google and once in Meta. Neither number is wrong. They answer different questions. Here's a quick check: add up the conversions each platform claims, then compare the total with the orders in your sheet. If the platforms add up to more, they're overlapping.
Treat each platform's report as that platform's view. Count real sales in your CRM or order sheet.
How do you choose a model, step by step?
Seven steps will get you to a model you can defend, and none of them needs a data team.
- 1.Name the action that counts: a sale, a booked call or a form.
- 2.Write down a typical journey, from the first ad to the purchase.
- 3.Start with each platform's default. Google Ads says data-driven is its default model for most conversion actions.
- 4.Compare it with last click for a month.
- 5.Ask customers "How did you hear about us?" when they book.
- 6.Keep one master sheet that merges platform numbers with real sales.
- 7.Change one setting at a time, and note the date.
If your journey is short, a simple model is fine.
How do you test a model without a data team?
Test a model by running it next to your current one, holding the budget steady and comparing the results against real sales.
Google Ads recommends testing a new model first and checking how it affects your return on investment. The model also changes how automated bids are optimised, so a switch can move your spend.
Meta has a compare attribution settings feature for the same job. Don't compare ad sets that use different models in the overview table.
Example: a Surat saree seller reviews both models for four weeks. Orders in the sheet decide who's right. Give any test at least four weeks, and avoid festival weeks, because a sale season distorts every comparison.
What mistakes and challenges should you expect?
Attribution mistakes feel harmless, because the reports still look neat, but they send budget to the wrong place.
- Trusting one platform's number as the truth.
- Cutting a channel that starts journeys because it doesn't close them.
- Switching models mid-campaign and reading the jump as growth.
- Comparing Meta ad sets that use different models.
- No "how did you hear about us" question.
Common Mistake: expecting a model to find the "true" credit. Every model is a rule. None sees what the customer felt.
Pro Tip: keep a simple log of every setting change with its date. Our ROI in digital marketing guide shows how to tie the results to profit.
What attribution trends matter in 2026?
Three shifts matter in 2026: rule-based models are fading, modelled measurement is spreading and consent rules are tightening.
Google Ads now supports last click and data-driven, with the other four retired. Google also released Meridian, its open-source marketing mix model, to everyone on 29 January 2025. A marketing mix model estimates each channel's contribution from total spend and sales, without following individuals.
India notified the DPDP Rules on 14 November 2025, with an 18-month phased timeline and clear consent notices. Expect less user-level data, and lean on your own sales records. Review your attribution settings every quarter, because platforms retire options, as Google just did. Write down what you changed.
What Should You Do Next?
Marketing attribution models are rules, and you choose the rule that fits how your customers buy. This week, write a typical customer journey, compare data-driven with last click in Google Ads for a month, and check both against real sales. Keep the model that agrees best with your sales sheet, and revisit it every quarter. If you want help setting up tracking and a fair test, our PPC services team can build a simple measurement plan with you.
Frequently Asked Questions
Common models include first click, last click, linear, time decay, position-based and data-driven. Each shares credit differently. Google Ads still supports last click and data-driven, and Meta supports standard and incremental models, so your options depend on the platform you use.
No model is best for every business. Data-driven models use your own data, so they suit accounts with enough conversions. Last click is simple and easy to check. Google recommends testing a new model first, so run two side by side and judge them by real sales.
Textbooks often list first touch, last touch, linear and position-based or time decay. Google Ads no longer supports first click, linear, time decay or position-based. Today you'll mostly meet last click, paid and organic last click, and data-driven in Google, plus standard and incremental in Meta.
Each counts credit its own way. Google Ads credits a conversion to the click date, Google Analytics 4 leaves out direct visits, and Meta counts clicks within 1 or 7 days and views within 1 day. One sale can appear in both reports. Use your CRM as the final count.
Not to start. Google Analytics 4, your Google Ads and Meta reports and a shared sheet of real sales cover the basics. Add a tool only when you can name a question these can't answer, and check that you can export your data and handle consent properly under the DPDP Rules.
An attribution window is the number of days between a person seeing or clicking your ad and taking an action that still counts for that ad. Meta's click-through window is 1 or 7 days, and its view-through window is 1 day. Shorter windows count fewer conversions.