Pages that only exist in JavaScript
Pricing, clinic types, customer stories and guides arrived as an empty HTML shell. Google queues such pages for rendering; most of these never came out of the queue.
A fictional business. The figures are a modelled scenario, not a client result.
14 months · Latin America · Signal plan
+207% organic trial sign-ups
Server-side rendering · Internal linking · Help-centre migration · XML sitemaps
Each station is a section of the site; each line is a path a crawler can follow from the homepage.
01The problem
Marketing pages rendered only in the browser, the help centre sat on a noindexed subdomain and feature pages were orphaned, so only the homepage ranked.
The marketing site was built as the same single-page app as the product, so every page was drawn in the browser. Googlebot reached the homepage, which had a static fallback, and stopped there. Vets searching for scheduling software, digital medical records or e-invoicing found competitors, although Nodo Austral had a page for each.
Pricing, clinic types, customer stories and guides arrived as an empty HTML shell. Google queues such pages for rendering; most of these never came out of the queue.
The 34 feature pages were reachable only from a menu that navigated with click handlers. No links, so nothing for a crawler to follow.
900 help articles sat on a subdomain, tagged noindex during a beta two years earlier and never switched back. No sitemap listed them.
Before · raw HTML of /precios
<!doctype html><html lang="es"> <head><title>Nodo Austral</title></head> <body> <div id="root"></div> <script src="/static/js/main.4f9c1a.js"></script> </body></html>After · the same URL, rendered on the server
<body> <div id="root"> <nav><a href="/funciones/agenda">Agenda</a> …</nav> <h1>Precios para clínicas veterinarias</h1> <p>De 1 a 40 profesionales, sin permanencia.</p> <a href="/ayuda/primeros-pasos">Primeros pasos</a> </div>Same URL, same design. The difference is whether the words and links are in the first response.
02The survey
Rendering
94%
of marketing and guide URLs returned an empty HTML shell; the text appeared only after the whole JavaScript bundle ran.
Finding 1 of 4
Help centre
900
help articles on a separate subdomain, all tagged noindex and missing from every sitemap.
Finding 2 of 4
Internal links
34
feature pages with no crawlable link pointing to them: the menu used click handlers instead of links.
Finding 3 of 4
Crawl logs
63%
of Googlebot’s requests went to JavaScript bundles and API calls rather than to pages.
Finding 4 of 4
03The route
The work ran on our Signal plan: technical SEO alongside Nodo Austral’s own developers, who kept shipping the product the whole time.
01
Marketing pages and guides are rendered on the server and hydrated in the browser, so the first HTML response carries the headings, text and links. A render test in CI fails the build if a template ever ships empty again.
ReopensMarketingGuides
02
A features hub with real links, breadcrumbs, related-feature blocks and a footer that names every section. Every page now sits within three clicks of the homepage.
ReopensFeaturesGuides
03
The 900 help articles moved from the subdomain to /ayuda/, each with a one-to-one 301 redirect. Noindex came off and canonical tags point at the new URLs.
ReopensHelp centre
04
A sitemap index split by section with honest lastmod dates, and API routes and bundles kept out of the crawl, so Googlebot spends its visits on pages.
Reopensevery line
04Route the crawler
Switch on a repair and send Googlebot’s train from the homepage. It stops at every station it can reach and halts at the first closed one on each line.
Swipe sideways to see the whole map
Marketing0 of 4
Features0 of 4
Help centre0 of 5
Guides0 of 4
05The timetable
Months 1–2
A crawl of the site as Googlebot sees it, raw versus rendered HTML for every template, 30 days of server logs and Search Console coverage.
Months 3–5
Server-side rendering for pricing, clinic types, customer stories and guides, shipped template by template behind a feature flag.
Months 5–6
The features hub, crawlable menu links, breadcrumbs and related-feature blocks.
Months 6–7
The help centre moves to /ayuda/ with 900 one-to-one 301s, noindex removed and a new sitemap index submitted.
Months 8–14
Weekly log checks, the render test in CI, and every new guide on line H linked in from the day it goes live.
06Arrivals
Organic trial sign-ups a month
+207%
46 → 141 organic trial sign-ups a month
Indexed pages
12 → 1,260
of the 1,260 pages that belong in Google’s index
Pages in Google’s top 10
1 → 214
pages ranking for at least one query, up from the homepage alone
Help-centre visits from search
0 → 18,400 a month
help articles now answer what vets type into Google
Modelled scenario for a fictional business — not a client result. Rankings and traffic are never guaranteed; see our Terms.
| Query (Argentina) | Month 1 | Month 14 |
|---|---|---|
| software veterinario | not in top 100 | #3 |
| software para veterinarias | not in top 100 | #4 |
| sistema de gestión para clínicas veterinarias | not in top 100 | #2 |
| historia clínica veterinaria digital | not in top 100 | #2 |
| agenda de turnos para veterinaria | not in top 100 | #5 |
| recordatorio de vacunas para mascotas | not in top 100 | #7 |
| factura electrónica veterinaria | not in top 100 | #3 |
| nodo austral | #1 | #1 |
Positions on google.com.ar, modelled. Before the work, 7 of these 8 queries had no Nodo Austral page in the top 100.
If your marketing site is a single-page app, the Signal plan starts where this one did: a render test of every template and a look at what Googlebot actually receives.