On May 1, 2026, I deleted Google Analytics from the Tideward marketing site. Two days later (today), I’m writing this post on the new tideward.app blog, and the post you’re reading was published into a website that has zero cookies, no consent banner, no third-party JavaScript except a YouTube trailer embed and a newsletter form, and no analytics dashboard of any kind.

I want to walk through what that decision actually cost, because the indie-dev internet has a lot of “remove your analytics” advocacy that hand-waves the trade-offs. There are real ones. They were just smaller than I expected.

The starting position

Tideward (then “Manu Idle”) had GA4 on the marketing site for about eighteen months. The dashboard showed me:

  • About 30-50 visitors per day, mostly direct traffic + a small organic trickle from Search Console
  • Top pages: the homepage, /faq, the dev blog index, and a handful of individual posts
  • Average session ~90 seconds (this is high for a brochure site; people read the FAQ)
  • Bounce rate ~70% (this is fine for a brochure site)
  • A funnel of “homepage → TestFlight click” that ran roughly 5-8% conversion
  • The geographic breakdown of visitors (mostly US, then UK, Germany, Australia)

Useful information. Not life-altering information. Mostly the kind of thing I checked once a week and felt vaguely satisfied by.

What pushed me to remove it

Three things, stacked:

1. The privacy-stance claim was getting weaker. Every time I described Tideward’s architecture publicly, I’d say “zero third-party SDKs in the binary.” That claim is true for the app — the App Store Privacy Nutrition Label literally reads “Data Not Collected.” But the website? The website was loading Google Analytics, setting cookies, sending behavioral data to Mountain View. The two stances didn’t match. People were going to notice.

2. The cookie banner question was getting louder. I’d been delaying a EU/UK ePrivacy consent banner because GA4 traffic from those regions was a small fraction of total, and the banner felt user-hostile. But “small fraction” wasn’t going to stay small forever, and “I haven’t bothered with the banner” isn’t actually a defensible position under ePrivacy. The trade-off was: add a clunky consent banner OR remove the cookies entirely. The second option was simpler and more honest.

3. The data wasn’t driving decisions. This was the quiet realization. I’d look at GA4 weekly, nod at the numbers, and continue doing exactly what I would have done without the data. The decision-relevant signals (TestFlight install rate, Discord daily active count, App Store Connect search-impression-to-install-ratio) all live elsewhere. GA4 was a comfort blanket, not a steering wheel.

What I actually did

The deletion took about thirty minutes. Concretely:

  • Removed the <script async src="https://www.googletagmanager.com/gtag/js?id=G-..."></script> block from the BaseLayout
  • Removed the corresponding gtag('config', 'G-...') script
  • Removed a small cookie-consent shim I’d added defensively but never finished
  • Cleared the Search Console association with GA4 (Search Console itself stays — it’s privacy-respecting and doesn’t set cookies)
  • Verified via curl + browser network tab that no third-party requests remain except the YouTube trailer embed (homepage only) and the Beehiiv newsletter form (blog only)
  • Updated the privacy policy to reflect “no analytics” instead of “anonymized analytics via GA4”
  • Updated the README’s third-party-services list

That’s it. No archive of historical data; GA4 keeps it for a while if you ever want to look back.

What I lost

Honest accounting:

  • Per-day visitor counts. I don’t know how many people read this post. I will eventually know how many people converted from this post (Search Console shows clicks from indexed pages) but the raw “did anyone open the page” number is gone.
  • Geographic data. I no longer know that today’s visitors are 60% US, 12% UK, etc. Search Console shows this for organic queries but not for total traffic.
  • Funnel attribution. The “homepage → TestFlight click” conversion rate was a real number. I can still get the TestFlight install count from App Store Connect, but the “what fraction of homepage visitors clicked through” metric is gone.
  • Time-on-page / scroll-depth. Whether anyone reads to the bottom of an 800-word post is now unknowable.
  • Real-time visitor counts. The little “1 person on /faq right now” indicator is gone. I wasn’t using it for anything; I’ll still miss the dopamine.

What I gained

The “no third-party SDKs” claim now applies architecturally to both app and website. The privacy policy is genuinely simpler. The cookie banner question is permanently closed. The site’s HTML is roughly 800 bytes lighter on every page load (small, but real). The next time someone audits Tideward for privacy compliance — a journalist, a lawyer, a curious player — there is exactly one third-party request to defend (the YouTube trailer iframe on the homepage), and that one is opt-in by user click.

The data I actually need still exists:

  • Google Search Console (free, cookieless): shows me organic queries, click-through rates, indexing status, sitemap coverage. This is the analytics that drives SEO decisions.
  • Bing Webmaster Tools (free, cookieless): same shape, different search engine.
  • App Store Connect: shows me TestFlight install rate, source attribution (Search vs Browse vs External), conversion percentage. This is the analytics that matters for the actual business.
  • Beehiiv newsletter dashboard: shows me subscriber growth, open rates, click-throughs. Newsletter is the highest-conversion channel for an indie game launch; this is where attention should be.
  • Discord member count + activity: the closest thing to a real-time audience sense. People showing up in #general is the most honest engagement signal Tideward gets.

The aggregate of those four is more useful than GA4 ever was, because each tool is closer to a real decision. App Store Connect tells me whether the App Store screenshot is converting. Beehiiv tells me whether the Friday newsletter outperforms the Tuesday one. Discord tells me whether the alpha is generating organic discussion. None of them require a third-party tag on my marketing site.

When I’d reconsider

Three triggers would push me back toward analytics:

  1. Site traffic crosses ~500 visits/day. Currently I’m at one-tenth that. Once there’s enough volume that funnel-shape decisions become real (e.g., “should the homepage hero be different?” experiments), some kind of measurement is useful. At that point I’d add Cloudflare Web Analytics — free, cookieless, brand-consistent, runs on Cloudflare’s edge so no third-party JS at all. Not Plausible or Fathom; Cloudflare’s free product matches them on quality for this use case.

  2. A specific funnel question App Store Connect can’t answer. Right now App Store Connect tells me everything I need about install conversion. If I ever want to A/B-test landing pages or measure “did the new screenshot help” outside of App Store Connect’s own attribution, that’s an analytics question. Until then, I have nothing to ask.

  3. Search Console stops being enough for SEO. Currently it covers indexing, queries, clicks, impressions, sitemap status. If something specific breaks — say, an unexplained drop in crawl rate that needs deeper diagnosis — adding analytics back is cheap.

The trigger for each is “a specific question I can’t answer with the existing tools.” Not “it would be nice to know.”

What this looks like for the rest of the project

This decision generalizes. The Tideward app already had zero third-party SDKs by design — no analytics, no crash reporters, no ad networks, no attribution libraries. Removing GA4 from the website made the website match the app. The next surface to audit is the Beehiiv newsletter (which uses Beehiiv’s own privacy-respecting form, but worth re-verifying), and eventually the Discord (where I have less control over Discord’s own analytics, but at least it’s not Tideward’s domain).

The bet is that “no third-party SDKs” becomes part of the brand. Not because every player will care, but because the few who do will trust harder, and Apple’s editorial team will notice. The App Store Privacy Nutrition Label literally rewards this with the “Data Not Collected” tag — that’s the visible signal Apple’s curators look for.

If you’re building an indie game and reading this thinking “should I remove my analytics?” — the answer is “audit which decisions you’d actually change.” If the answer is “none,” you should remove it. If the answer is one or two specific things, find the cookieless tool that surfaces only those things. Cloudflare Web Analytics is what I’d reach for first.

But if the comfort blanket is its own justification, that’s also fine. Just be honest with yourself that’s what it is.

The full writeup of Tideward’s privacy stance — what’s removed, what’s kept, the architectural proofs — lives at /no-ads-no-tracking. The privacy policy is at /privacy. The TestFlight install guide is at /install — if you want to test a game built this way, on every Apple device, free.