Thread by @blvckledge

If you can test Google Shopping campaigns properly, you can scale any e-commerce business.

Here’s how we do it (most media buyers don’t know how):

A while ago, we did a test on 2 different product images for a client.

Same product, same price, similar titles.

Over 90 days, here’s the result:

- Variation A: $189,645
- Variation B: $8,970

That’s a 21x difference. Just by switching the image style.

The example above goes to show how important testing is.

A simple tweak could double your ROAS.

Let it compound over time, and the effect is even more dramatic.

With that being said, here’s our testing system:

In Search Ads, you have built-in features to A/B test your copy.

But Shopping Ads are trickier.

Google pulls everything from your product feed.

Testing requires a special process so you don’t disrupt current performance.

Let me show you how:

But before we get into the how, let’s talk about what you should test.

There are 3 core levers every eCom brand should experiment with:

- Product titles
- Product images
- Landing pages

I recommend starting with titles, they’re the biggest driver of your results.

1. Setup

Now, the overview of our method:

First, duplicate your product feed.

Create a new feed version with the variations you want to test.

Then launch this modified feed in a separate campaign.

You have 3 options to duplicate your feed:

- Manual
- Feed Management Tools
- Supplemental Feeds via GMC

We recommend using management tools like DataFeedWatch.

These platforms let you apply conditional rules to titles, descriptions, images, and scale this process.

In your duplicated test feed, try things like:

- Using lifestyle images
- Writing product titles with the benefits upfront
- Sending traffic to a new landing page variant (e.g, sales page)

Test 1 variable at a time.

Otherwise, you won’t know which one made the difference.

When you set up your tests, follow these guidelines:

- Run one campaign for each feed
- Use clear naming (e.g., “Title Test A”)
- Set identical targeting & budget across campaigns

This ensures your tests are fair, and the data truly shows what performs better.

2. Accelerate Testing with AI

AI tools like ChatGPT-4o help you rapidly:

- Generate different image styles
- Mock up landing page sections
- Get various title variations

This made the whole process scalable and efficient.

3. Testing Duration

Your testing time should depend on your traffic volume:

1,000+ clicks/day → run for 2 weeks

300–999 clicks/day → run for 3–4 weeks

Under 300 clicks/day → run for 4–6 weeks.

4. Key Metrics to Monitor

When running tests, track these 3 metrics closely:

- Impression Share: stabilizes within the first week
- CTR: meaningful after 500 impressions
- ROAS: becomes reliable near the end of your testing period

Pick your winner based on this data.

5. Testing Cycle

Every month, we allocate 10-20% of our Shopping budget to testing.

Here’s what that process looks like:

- Run test for 2–4 weeks
- Analyze and identify winners
- Set the winning variant as your new baseline
- Start the next test with new variations and repeat

That's our full playbook for testing Google Shopping Ads.

Apply this approach to your account to uncover new winning variations.

And if you need our direct help to set it up, let’s chat:

echelonn.io/services/googl…