A site crawlers can't read, and AI engines don't know exists

SEO Audit: DC Creator

DC Creator is a SaaS for recruiters. It transforms a candidate's CV into an optimized “dossier de compétences” (skills dossier) in one click, generating structured skills so recruiters can evaluate candidates faster. The product is real and useful. The SEO, on the other hand, starts at zero — there is essentially none in place yet, which makes this a foundation audit: where to begin, and which actions move the needle first.

So we cast a wide net and focus on the highest-impact work. We go through the technical setup, Ahrefs, Google Search Console, content, backlinks, and AI visibility. One issue sits above all the others and has to be fixed before anything else can work: the site is a single-page application that renders nothing without JavaScript. To a search crawler it is slow and partial; to an AI crawler it does not exist.

Current performance

The core problem: DC Creator is not losing a ranking battle — it is not on the field yet. The SPA architecture keeps crawlers and AI engines from reading the site, there is almost no indexable content, and Google is not even sure what the product is about: the competitors it suggests by default are scheduling tools and freelancing platforms. The entire opportunity here is foundation work, and the upside is large because the keyword difficulty in this niche is close to zero.

Finding #1: The site doesn't exist without JavaScript

Open DevTools, disable JavaScript, refresh the homepage — and the page goes blank. That confirms a single-page application (SPA) with no server-side rendering: the content is assembled in the browser, not delivered by the server.

This is the one issue that blocks everything else. Google's crawler renders JavaScript, but it does so on a delay and inconsistently, so indexing is slow and partial. AI crawlers, LLM bots, and agents are stricter: most of them do not execute JavaScript at all. To ChatGPT, Perplexity, and the rest, the site is effectively empty. You cannot rank, get cited, or build AI visibility on a site the machines cannot read.

The fix is the absolute prerequisite: render the content server-side. DC Creator was built custom by its founders, so the team has the skills to do this. Nothing else in this audit pays off until it is done.

Finding #2: There is almost nothing to index

Search Console shows one indexed page and roughly twelve excluded. The sitemap lists only six URLs, and several of them should not be there at all (the OG image, the terms-of-sale page, an MP4 video). The homepage is long, the blog has a single post. In practice, the site has no indexable content yet.

My rule of thumb is that SEO really starts around 60 pages of unique, indexable, quality content. So the North Star here is roughly 70 useful pages. Some of them are basic and you can ship them fast:

• A pricing page, a “how it works” page, an FAQ.
• A team / story page. This one matters more than it looks: showing who you are and why you're the right people to build this tool feeds conversion, credibility, and the E-E-A-T signals that both Google and AI engines reward.

Those pages alone move you a long way toward the 70-page target. The rest comes from cornerstone content, which is Finding #3.

Finding #3: Own the “dossier de compétences” cluster

There is already a first blog post (“modèle de dossier de compétences”), which is the right instinct. “Dossier de compétences” is the key topic for this domain. The volume is modest (roughly 400 cumulative across variants) but the keyword difficulty is near zero and the intent is high, which makes it both winnable and worth a lead magnet to capture emails.

This is where the real competitors finally appear. Turbo CV and Qonex rank on the topic, and Turbo CV is a clean benchmark: it runs a proper cornerstone cluster — a referent parent page on “dossier de compétences” with deeper child pages nested beneath it (one per job: technical assistant, network engineer, data analyst, and so on). Build the same structure, then go one level better:

1. 1One parent page on “dossier de compétences,” with child pages per job role linked beneath it.
2. 2Better content than the benchmark. More depth, more data, and especially unique data. In the AI era, information that does not already exist on the web carries the most weight. If you have insight from your own product and customers that competitors don't, publish it.
3. 3Make it downloadable and rich. Infographics, a downloadable PDF template, a topic FAQ where it makes sense. These add value for the reader, the crawler, and the LLM at the same time.

That single cluster is already a strong content pipeline, and it points straight at the 70-page goal.

Finding #4: Win one primary keyword on the homepage

Right now the homepage is not optimized for a single clear target. The on-page work is surgical:

• Pick one primary keyword and make sure it appears in both the meta title and the meta description. Keep the meta description under ~150 characters and the title around 50 to 55.
• Drop the meta keywords tag. It carries no ranking value today, reads as keyword stuffing, and hands your target keywords straight to competitors.
• Fix the two H1s. There are currently two <h1> tags on the homepage, one of them nearly hidden. There should be exactly one H1 per page — remove the extra one.
• Push the homepage past 1,200 words. It sits around 1,100 today. Aim a little higher, but only where it genuinely adds meaning.
• Keep homepage content unique. As you add pages and reuse components, make sure the bulk of the homepage copy lives only on the homepage, so you don't dilute it into duplicate content.

Finding #5: Mine Search Console for question intent

Even with one indexed page, you already receive clicks and impressions, almost entirely on “dossier de compétences.” That's a good sign, and it's the seed of a proper keyword study.

The high-value move is long-tail question queries. Filter your queries with a regex for interrogatives, and the how-to intent jumps out: "comment remplir un dossier de compétences", "comment faire un dossier de compétences". These are recruiters trying to understand how to build a dossier, which is exactly your product's job. They are also the same phrasings people type into ChatGPT and Claude, which makes them the best queries to own for classic and AI search at once. Build a long, genuinely detailed how-to landing page for that intent — video included if you can.

Finding #6: Bootstrap backlinks in three waves

You start from zero, which actually makes the early wins easy. Run it in order:

1. 1Directories first. As a software product, get listed in reputable software and niche directories. Product Hunt alone (DR ~91) is a dofollow link from an extremely strong domain, and it moves a brand-new Domain Rating from 0 toward 20–30 quickly. Add GitHub, a published Notion page (your careers page, for example) linking back, and a Medium article. This is the bootstrapping layer.
2. 2Competitor backlinks. Study who links to the players ranking ahead of you. Turbo CV is small, but bigger, semantically relevant sites in French HR and recruiting (Welcome to the Jungle, Révolution RH, and similar) carry real weight in your category. A link from a recognized job board is a strong signal.
3. 3Manual, surgical outreach. Use Google search operators like "dossier de compétences" -site:dc-creator.com to surface every blog and partner already writing about the topic, then pitch them on linking to your tool. It's repetitive done by hand, but outreach tools can speed it up.

Two cautions: avoid the backlink equivalent of keyword stuffing (bulk paid links unrelated to your theme), and investigate that sudden referring-domain spike around the April core update to confirm it isn't spam pointing at you.

Quick technical fixes

Important, but not where you should spend most of your time at this stage:

• Broken OG image. The image shown when your site is shared is broken. Quick fix, real impact on social and AI previews.
• Sitemap cleanup. Remove the junk URLs and consider splitting into a French sitemap and an English one.
• robots.txt and locale paths. There's an fr-ca path that doesn't resolve and redirects to fr. Clean that up.
• Hreflang and canonicals. Today there's a single canonical. Add self-referencing canonicals plus hreflang alternates so Google knows which language version is authoritative (English is likely your default, flagship market). This also prevents duplicate-content issues across language versions.
• Mobile performance. PageSpeed shows the heaviest asset taking around 8 seconds to load and a slow First Contentful Paint, worse on mobile — which is what crawlers simulate. Optimize the above-the-fold assets and scripts, but in a second pass.
• Schema. You already have Software and FAQ schema in place, which is solid. Run a quick pass (validate at validator.schema.org) to see what else applies.

AI visibility

This is the differentiator, and the SPA issue from Finding #1 is exactly why it's weak today. In our live snapshot, Claude and ChatGPT tell two very different stories:

The verdict

DC Creator has a real product in a niche where the keyword difficulty is low and the competitors are beatable. The work is pure foundation, and the order matters.

Clean the technical issues along the way, but don't let them eat the time. You're a bootstrapped team, so base your priorities on what the competitors who already rank are doing — if Google ranks them, they're answering the query well, and that's your shortcut. Do that, and DC Creator stops being the recruiting tool the machines can't see, and becomes the one they recommend first.