I had a Google Ads campaign running for Talk to Sammie, a conversational AI product I built and launched
in 2026.
I believed conversion tracking had been set up.
Google Ads even had a Primary conversion called “Subscribe.”
Then I looked more closely.
By August 29, the campaign had generated 293 clicks and spent $69.74, but it was still reporting zero
campaign conversions. Elsewhere in Google Ads, I could see 17 page views and one sign-up.
Something didn't add up.
What “Subscribe” actually meant
When I dug into the conversion setup, I discovered that the Google Ads conversion named “Subscribe” had
been created months earlier, on May 31.
It wasn't actually looking for a confirmed paid subscription.
It was a website page-load conversion triggered by a URL beginning with:
www.talktosammie.com/billing-success
The conversion had a fixed $1 value, counted one event per interaction, and by August was inactive. It had
recorded zero conversions.
The name “Subscribe” described what I wanted to measure.
It didn't describe what Google Ads was actually measuring.
That distinction seems obvious now. It wasn't obvious while I was building the business.
The strange part: the other pieces were working
Talk to Sammie uses Stripe for subscriptions.
Inside the application, the billing-success process wasn't simply assuming that anyone who reached a page
had paid. The application checked Stripe information including checkout completion, payment status and
the resulting subscription before treating the transaction as a purchase.
The application also had newer events for sign-ups and purchases.
So I had a real subscription system. I had purchase verification. I had application events. I had Google Ads
conversion tracking.
What I did not have was a reliable connection between all of them.
The application's purchase event wasn't sending Google Ads the information it needed to connect that
verified purchase to the Google Ads conversion action.
Meanwhile, the Google Ads conversion called “Subscribe” was still based on an older page-load rule.
From my side of the screen, these things looked like parts of the same system.
They weren't.
Nothing had to be dramatically broken
This was probably the most useful part of the discovery.
Google Ads wasn't mysteriously malfunctioning. Stripe wasn't failing to process subscriptions. The
application wasn't simply declaring every visitor a customer.
The problem existed between the systems.
Each component could appear to be doing its job while the measurement chain I thought I had didn't
actually exist.
That's particularly easy to miss when you're building a business with modern platforms and AI-assisted
development.
You see a conversion named “Subscribe.”
You see a purchase event in the application.
You see Stripe verifying transactions.
You see tracking tags installed.
It's very easy to mentally connect those dots.
The software doesn't connect them just because you do.
What I changed
Once I understood what was happening, I cleaned up the obsolete and duplicate conversion actions and
changed the setup so the campaign focused on the outcomes I actually cared about: sign-ups and paid
purchases.
More importantly, I changed the question I ask.
I no longer consider “Is conversion tracking set up?” a particularly useful question.
Now I want to know:
What exact event is being counted?
What causes that event to fire?
Where does the information come from?
Can I trace it all the way back to the business outcome I think I'm measuring?
If the answer to the last question is no, the label on the conversion doesn't matter very much.
Why I'm writing this down
There are plenty of instructions for setting up Google Ads conversion tracking, Stripe subscriptions and
analytics events.
I followed plenty of instructions while building Talk to Sammie.
What was harder to find was information about what happens between those systems when you're a small
business owner trying to make all of them work together.
That's why I'm documenting this experience.
The surprise wasn't that conversion tracking can be configured incorrectly.
The surprise was that so many individual pieces could look correct while the assumption connecting them
was wrong.
And I didn't discover that by reading another setup guide.
I discovered it because the numbers didn't make sense, so I looked again.
Before you spend another dollar on ads, take the Advertising Readiness Test -https://secondlookworks.com/products/advertising-readiness-test-50-point-scorecard-before-you-run-ads