A/B testing Google Ads in India: a practical guide that actually works

A/B testing in Google Ads gets talked about a lot, but in practice most Indian advertisers either never do it or do it wrong. They change the bid strategy and the ad copy and the landing page at the same time and then wonder why performance shifted. Or they run a test for three days on a ₹500 daily budget and call it conclusive.

Real A/B testing is slower and more boring than it sounds. It is also the only reliable way to compound improvements over time.

What to test and in what order

Start with what has the most impact per change. In order, that is usually: bidding strategy, then landing page, then ad copy, then targeting settings, then smaller elements like ad extensions.

Most guides tell you to start with ad copy because it is easy to change. But if your landing page converts at 1 percent, the best ad copy in the world will not fix that. Fix the biggest bottleneck first.

For a new campaign with limited data, I recommend starting with a landing page test. Build two versions of your landing page, split traffic 50/50 using Google Optimize or VWO, and run for at least two weeks before drawing conclusions.

How to run ad copy tests properly

Google Ads has a built-in Experiments feature under the Campaigns section. Use it instead of just duplicating ad groups and hoping. The Experiments tool splits traffic cleanly between your control and variant.

When testing ad copy, change one meaningful element per test. Testing a new headline and a new description simultaneously tells you something performed better but not what caused it.

A typical ad copy test I run looks like this: control has a benefit-led first headline ("Save 30% on your tax filing"), variant has a proof-led first headline ("500+ Ahmedabad businesses file taxes with us"). Everything else stays identical. Run for two weeks minimum, three weeks if daily spend is under ₹2,000.

Statistical significance on small Indian budgets

This is the hard part. Many Indian small business campaigns run on ₹1,000 to ₹3,000 per day. Getting statistically significant results at 95 percent confidence takes longer than most people expect.

A rough guide: for a test to be meaningful, each variation should see at least 100 conversions. If your campaign generates 10 conversions per week, you need 10 weeks per test variant. That is painfully slow.

Practical adjustment for small budgets: lower your confidence threshold to 80 to 85 percent for low-stakes decisions like which headline to keep. For high-stakes decisions like switching bid strategies, wait for more data or increase budget temporarily to speed up the test.

Landing page A/B testing

For landing pages, use Google Optimize (or now, a third-party tool like VWO or Optimizely since Google Optimize sunset in 2023). Set up a goal that matches your actual conversion event, not just a page view.

Common high-impact tests for Indian landing pages: placing the phone number and WhatsApp link above the fold versus below, a short form with three fields versus a longer form with five, and a headline that mentions the city versus a generic national claim.

I ran a test for an immigration consultancy in Mumbai where the variant simply changed the headline from "Canada PR Visa Consultants" to "Canada PR Visa Consultants in Mumbai Helping Families Since 2010". The variant improved form submissions by 22 percent. The test took four weeks to reach significance on a ₹1,500 daily budget.

Document everything

Keep a simple spreadsheet logging every test. Include the hypothesis, start date, end date, metric measured, result, and what you changed after. This compounds over time. After 12 months of testing, you have a real understanding of what your specific audience responds to, which no generic marketing guide can give you.

Most agencies and in-house teams skip this documentation step and end up re-testing things they already know. Do not be that business.

Frequently asked questions