Analytics
Public, privacy-respecting traffic stats for donray.dev. No cookies, no raw IP addresses stored. Ever.
Tracked page views
Real page loads, measured from October 2026.
Page views
265
All time · bots filtered
Unique visitors
47
All time · bots filtered
Page views per day
Last 30 days
Oct 3Oct 15
Page views by page
All time, most viewed first
| Page | Page views | Unique visitors |
|---|---|---|
| / | 108 | 43 |
| /analytics | 20 | 9 |
| /blog | 18 | 7 |
| /projects/patternspell | 17 | 9 |
| /principles | 15 | 9 |
| /version | 14 | 8 |
| /projects/pico-domains | 13 | 4 |
| /projects/hide-zero-cards | 12 | 6 |
| /projects/cyclei | 9 | 1 |
| /blog/ai-code-review-less-context | 6 | 2 |
| /privacy | 4 | 2 |
| /terms | 4 | 2 |
| /blog/422-code-reviews | 4 | 0 |
| /blog/ldx3-new-manager | 4 | 1 |
| /projects/donray-dev | 4 | 2 |
| /blog/we-automated-the-wrong-bottleneck | 2 | 1 |
| /es | 2 | 1 |
| /zh | 2 | 1 |
| /tl | 2 | 1 |
| /sitemap.xml | 1 | 0 |
| /llms.txt | 1 | 0 |
| /es/principles | 1 | 0 |
| /blog/review-to-learn | 1 | 0 |
| /robots.txt | 1 | 0 |
Before tracking started
CDN-request estimates from before page-view tracking existed: a different measurement, not page views, and not comparable to the tracked numbers above.
Historical traffic (estimated)
CDN requests per month: includes page loads, images, scripts, stylesheets, and bots. Not page views; approximate, for rough scale only.
Apr 2026
≈88,000
May 2026
≈55,000
Jun 2026
≈75,000
Jul 2026
≈95,000
Aug 2026
≈128,000
Sep 2026
≈118,000
Oct 2026*
≈15,000
* Oct 2026 covers Oct 1–2 only. Source: AWS CloudWatch (Amplify Hosting requests), pulled Oct 2, 2026. Page-view tracking started Oct 2026.
How it's measured
What these numbers are (and aren't)
- Tracked page views are real page loads in a browser, recorded from October 2026. Images, scripts, and stylesheets don't count. Only the page itself. Known bots and crawlers are filtered out before counting.
- Unique visitors are real people, counted once ever: each visit is hashed (IP + browser, salted and non-reversible) and the hash is kept in a permanent set. A visitor who returns a year later still counts once, and a visitor who reads several pages counts once. Shared networks can undercount; changing IPs can overcount. True unique counting started in October 2026; earlier visits used day-bound hashes that cannot be converted.
- No cookies are set and no raw IP addresses are stored. All records are kept permanently.
- Historical estimates are a separate, older measurement: approximate monthly CDN request counts from AWS CloudWatch, from before page-view tracking existed. They include images, scripts, stylesheets, and bots, so they are not page views and can't be compared with the tracked numbers.
Last updated Oct 16, 9:48 AM · Tracking started October 2026
What I've learned
Notes from running my own analytics
- CDN requests are not page views. Before real tracking existed, CloudWatch showed ~560K requests over six months. But that counts every image, script, stylesheet, and bot. Real human page loads are a fraction of that. The dashboard keeps the two measurements visibly separate because conflating them would be dishonest.
- Almost all traffic is the homepage. The per-page table above makes that plain. The site's job is to be found, read, and contacted from a single page.
- Privacy-respecting analytics is a solved problem. One DynamoDB table, salted daily hashes, no cookies, no raw IPs. And the dashboard still answers every question I actually have about my traffic.