What actually broke my trust in analytics (after shipping real products)

After I published my first story about moving from client work to my own SaaS - https://www.indiehackers.com/post/from-building-client-websites-to-launching-my-own-saas-and-why-i-stopped-trusting-ga4-cdffa602c4 , a few people asked the same question:
“What exactly made you stop trusting GA4?”

The honest answer:
Not one big failure.
A long list of small ones.

Before SaaS, I built products where behavior mattered more than “growth curves”.
One of them was Around MD — a platform helping people find restaurants, parks, and interesting places in their city.

People didn’t “browse”.
They searched with intent.

And that’s where things started to feel wrong.

When numbers stop matching reality

I’d see:

• pages ranking in Google
• users messaging me about places they found
• content getting shared

But analytics told a different story:

• lower traffic than expected
• weird drops after “improvements”
• conversions that didn’t correlate with anything

So I did what most founders do: I assumed the tool was right and my intuition was wrong.

That was my first mistake.

The cookie banner illusion
At some point I noticed a pattern:
Every time I tried to be more “compliant”, the data got worse.

Consent banners went up.
Opt-out rates went through the roof.
Sessions disappeared.

Nothing changed in the product.
Only the measurement did.

And suddenly: SEO looked broken, content decisions felt random, experiments became impossible to evaluate

The tool was “working as intended”.
But the result was unusable.

That’s when it hit me:

Analytics that requires users to say “yes” before being accurate is fundamentally flawed.

My second mistake

I kept trying to fix the tool instead of questioning the premise.

More events.
More configurations.
More time in dashboards.

Zero clarity.

As a solo founder, that’s deadly.
You don’t have time for uncertainty.

Every unclear metric slows you down more than having no metric at all.

The shift in mindset

The real turning point wasn’t technical.

It was this decision: I’d rather have less data I trust than more data I doubt.

Once I optimized for trust instead of completeness: decisions got faster, experiments became obvious, anxiety dropped

That mindset directly shaped what I’m building today — even if I rarely talk about it publicly. - checkanalytic.com

Looking forward

I don’t think analytics will be “solved” by new features.

It will be solved by: removing assumptions, respecting users by default and designing for people who actually build products

Not for enterprise decks.
Not for legal checklists.

For founders shipping at 2am.

What was the moment you realized your analytics was lying to you?