Can a website that displays with consistent colors and a seemingly functional menu be considered finished? For an entrepreneur preparing their online presence, this impression is reassuring, but it leaves several questions open. Do the buttons lead to the right place? Does an interior page remain correct when opened directly? Can search engines access the published content?
- Un site peut paraître fini alors que ses boutons ou ses pages internes ne fonctionnent pas correctement.
- Il faut vérifier le résultat attendu de chaque action, et non seulement l’apparence du bouton.
- Il faut tester les pages internes ouvertes directement, puis vérifier les réponses HTTP et le contenu affiché après exécution des scripts.
- Il faut contrôler séparément la mise en ligne, l’indexation et la visibilité, puis documenter la date, les limites et les pages vérifiées.
A static website project hosted on Netlify, with a domain purchased from Namecheap, provides two concrete examples. Checks on October 14, 2026, revealed subtle on-screen errors, but significant for navigation. Their relevance goes beyond the choice of generation tool: they show what a delivery report should demonstrate.
Start from the expected action, not the button’s appearance
The first flaw concerned a button announcing access to eight modules. It reloaded the top of the page that presented them. The link worked technically and produced no visible error; its result simply did not match its label.
The correction involved targeting an anchor located at the start of the modules, then verifying the arrival in the browser. This check is easily generalized: describe what the visitor should obtain, perform the action, and compare the result. For a form, an on-screen confirmation alone does not prove the request was received. For a video, the presence of a play button does not prove the media starts. A useful recipe therefore describes a journey and its outcome, for example, ‘the link to the program opens the start of the modules, on desktop and in mobile view.’
Open pages other than from the homepage
The second flaw was more deceptive. At the address /formation/index.html, the server provided the expected content, but the script executed afterwards in the browser displayed the homepage. A check limited to the server’s response would have let this contradiction slip through.
The correction focused on the code’s recognition of addresses and the redirection of known variants to their corresponding page. An unknown address should, itself, remain unfound: systematically replacing it with the homepage would have masked the problem. For a small site, a few tests are enough to spot this type of error: open an interior page directly, reload it, test an old variant, and enter a non-existent address. Then compare the main content and title obtained, and check the HTTP responses, i.e., the status returned by the server.
This distinction also matters for SEO. The Google documentation on JavaScript SEO explains that processing these pages involves steps of crawling, rendering, and indexing. Checking the initial HTML therefore does not replace examining the result after script execution.
Specify the scope of checks and separate the steps
On this project, an article was checked at a width of 390 pixels, without horizontal scrolling. This observation is only valid for this page, this width, and this version of the site. The rest of the recipe must include elements that move or overlap, such as the open menu, the fixed banner, the video player, and the close button.
During an initial check, the sitemap was accepted in Search Console, while the homepage was still listed as detected but not indexed. This observation describes a dated state, not the current status of the site. It especially reminds us that going live, indexing, and visibility are checked separately: pages open, indexing is controlled in the search engine’s tools, and impressions and clicks allow observation of visibility. A crawl request does not guarantee indexing, according to Google Search Central. Delivery thus benefits from including the list of pages checked, anomalies corrected, known limitations, and the date of the checks.
Transparency: this article was prepared with AI assistance based on the project’s documented checks. The cited site belongs to its owner and includes affiliate activity. The text does not constitute an independent opinion on the training presented there.
