Blindtest (niveau débutant) https://t.co/RCbfBce9dQ

Avec le festival Octogônes qui approche, c’est l’occasion de vous parler d’un roman de littérature arborescente que j’ai récemment jouquiné et adoré.

Une newsletter avec des trucs que j'aime (faire) dedans : des mots, des jeux et de la voix (off). Click to read Jeu me livre, by Lilie Bagage, a Substack publication. Launched a year ago.

Je n’en reviens pas.

Professional Voices! Free online quote. 24 hour turnaround. Affordable Rates. We have the voice talent for every voice over - explainer videos, IVR, commercial TV and radio, e-learning, promos, Facebo...

C’est la rentrée !

# I Ran a Real SEO Audit On My Own Live Site — Here's a Free SEO Audit Report Sample Four days ago, my site was invisible to Google. Not slow. Not penalized. Invisible — `robots.txt` was telling every crawler `Disallow: /`, and nothing about the site _looked_ broken. No 500s, no console errors, no drop-everything alert. Just a CDN quietly serving a stale, wrong file to every bot that asked for it. I found it while running a full technical SEO audit on my own portfolio site, voixoff-fr.surge.sh — a one-page static portfolio, four pages total, zero JavaScript, 4-13KB per page. I do SEO audits as a service, and I wanted a real, embarrassing, unedited example to show what one actually looks like before I ask anyone to pay for it. So instead of a mocked-up PDF, I published the whole thing: **seo-audit-sample.surge.sh** , a free SEO audit report sample built from a live site with real bugs in it. This post walks through how I ran the audit, what it found, and the one mistake that almost made me miss the worst bug on the site. ## The method: no paid tools, just curl and patience I didn't run this through a $99/month SaaS dashboard. The whole audit stack was: * `curl -I` and `curl -v` for raw response headers * Manual HTML parsing for title/meta/heading structure * Direct inspection of CDN cache headers (`surge-cache`, `age`, `etag`) * The PageSpeed Insights API for performance (when it wasn't rate-limited — more on that below) Nothing exotic. The point of the exercise was to show that most of the value in a technical audit comes from actually reading the raw response the way Googlebot sees it, not from a dashboard with colored gauges. ## The bug: robots.txt lied, and the cache made sure nobody noticed Here's the sequence. I requested `https://voixoff-fr.surge.sh/robots.txt` and got: Disallow: / Full site, blocked. But the file in my deploy folder said `Allow: /`. So either I'd deployed the wrong file, or something was serving a stale copy. I checked the response headers: surge-cache: HIT age: 35619 Age of nearly 36,000 seconds — almost 10 hours old, and climbing, because nothing was invalidating it. The root cause: surge.sh serves its _own_ default `robots.txt` with `Disallow: /` on new sites, and its CDN had cached that default on the bare path. I had since deployed my corrected file (`Allow: /` + Sitemap directive) — the deploy succeeded, the origin file was correct — and yet the CDN kept serving the cached default to every bot. Googlebot had been reading `Disallow: /` for four days straight, with no error surfaced anywhere. That's the finding that anchors this whole SEO audit report sample: a bug with zero visible symptoms that can silently deindex a site. ## The trap: cache-busters hide the bug they're supposed to catch My first instinct when I suspected caching was to check `robots.txt?cb=12345` to force a fresh fetch. It worked — I got the correct, up-to-date file. Which meant, for about ten minutes, I thought everything was fine. It wasn't. Googlebot doesn't request `robots.txt?cb=12345`. It requests the bare path, `/robots.txt`, with no query string — and that's exactly the URL the CDN was serving stale. Adding a cache-buster proved the _correct file existed_ , but it completely masked the fact that the _real_ URL Google actually hits was still broken. The lesson, and the reason I'm calling it out explicitly in the report: **always test`robots.txt` with the bare path, no query string, no cache-busting parameter.** Anything else tells you what you want to hear. ## What else the audit found The robots.txt issue was the critical one, but a real audit checks the whole surface. Summary table: Area | Finding | Severity ---|---|--- Indexability | `robots.txt` serving `Disallow: /` via stale CDN cache (age 35619s) | Critical Title tag | Homepage title is 75 characters — over the ~60 char limit, gets truncated in SERPs | Medium Meta description | 209 characters — over 160, also truncated | Medium Headings | Satellite pages have an H1 but no H2 structure | Medium Structured data | Valid JSON-LD `Product` schema on the index page, but missing on satellite pages | Low-Medium Security | No HSTS header (platform limitation on surge.sh's free tier) | Low Performance | PageSpeed Insights API returned 429 (quota exhausted, shared key) | Info Images | 0 images missing `alt` text | Pass Canonicals / OG | Present and correct on all four pages | Pass Sitemap | `sitemap.xml` present, 4 URLs, matches live pages | Pass Mobile | 0 horizontal overflow at 375px viewport | Pass Load times | 50-214ms across all pages | Pass Accessibility | Contrast ratios meet WCAG AA | Pass The strong points are genuinely strong — this is a hand-built, zero-JS static site, so there's very little surface area for bugs. Which is exactly why the `robots.txt` cache issue stood out: it's the one failure mode that has nothing to do with code quality and everything to do with infrastructure you don't fully control. ## The full sample report I didn't want to just describe this in a blog post — the actual report, with headers, timestamps, and the full findings list, is published here: **seo-audit-sample.surge.sh**. It's the same format I'd hand to a paying client, evidence and all, minus the client's name. If you want to see what an "SEO audit report sample" looks like when it's built from a real site instead of a template with placeholder text, that's it. ## Quick wins if you're checking your own site * Fetch `robots.txt` and `sitemap.xml` with a bare `curl`, no browser, no query string. Compare against what's on disk. * Check `age` and `*-cache` headers on anything served through a CDN — a `HIT` with a large `age` on a file you recently changed is a red flag. * Keep titles under ~60 characters and meta descriptions under ~160 — both of mine were over, and both get silently truncated in search results with no warning from anything. * Don't assume "it deployed successfully" means "it's serving correctly." Those are two different claims. ## If you want one done on your site This is genuinely my own site, audited with the same process I use for clients — no fabricated numbers, no invented scores. If you want a similar evidence-backed audit (headers, raw responses, no guessing) run on your own site, I'm doing these as a paid service, launch pricing starting at $0.10: **alexcarter.privacy@proton.me**. Otherwise, go check your `robots.txt` right now, without the `?cb=`.

YouTube video by djyogourt

#8 ~ Où les idées de game design s'épanouissent

Et un nouvel EnVoix, un !