Skip to content
llmoptimisation.fr

Audit

LOOP audit and checklist

40 checkable items grouped by the four LOOP dimensions. Each item carries a weight. Sum your points to get a score out of 100 and prioritise the work.

Mise à jour : 14 April 2026 9 min de lecture

Dimension L

Legibility (10 points)

Is your content extractable without friction by an AI engine?

24 pts

  1. 01

    Each page has a unique H1 carrying its primary query.

    3 pts
  2. 02

    H2s segment distinct intents, not decorative sections.

    3 pts
  3. 03

    Paragraphs run 3 to 6 lines, one idea per paragraph.

    2 pts
  4. 04

    Sentences are standalone, readable out of context.

    3 pts
  5. 05

    Every acronym is defined at its first local occurrence.

    2 pts
  6. 06

    Content contains at least one table or bounded list per pillar page.

    2 pts
  7. 07

    Publication and update dates are visible and authentic.

    3 pts
  8. 08

    Figures are dated and sourced.

    3 pts
  9. 09

    WCAG AA accessibility is respected (contrast, focus, alt, landmarks).

    2 pts
  10. 10

    Critical images carry a descriptive alt (not "image.png").

    1 pt

Dimension O

Ontology (10 points)

Is your brand identifiable, disambiguated and recognised as an authority?

24 pts

  1. 01

    The brand entity is named consistently across the whole site.

    2 pts
  2. 02

    A schema Organization with sameAs to LinkedIn, Wikipedia, Wikidata is in place.

    3 pts
  3. 03

    A factual About page establishes identity, dates and activities.

    2 pts
  4. 04

    The site covers a topic cluster (1 pillar + 5 to 10 satellites).

    3 pts
  5. 05

    Pages reference domain markers (competitors, tools, methods) with nuance.

    2 pts
  6. 06

    Internal linking ties pillar and satellites through varied semantic anchors.

    3 pts
  7. 07

    The site accrues backlinks from topically aligned sources.

    3 pts
  8. 08

    External mentions (podcasts, press, LinkedIn) converge on the same identity.

    2 pts
  9. 09

    A contextual (not generic) FAQ exists on pillar pages.

    2 pts
  10. 10

    A glossary or definitions page exists for the domain’s key terms.

    2 pts

Dimension O

Operations (10 points)

Can AI bots crawl and technically understand your site?

22 pts

  1. 01

    Explicit robots.txt for GPTBot, PerplexityBot, ClaudeBot, Google-Extended.

    3 pts
  2. 02

    WAF / Cloudflare rules verified: AI bots not blocked by default.

    3 pts
  3. 03

    Critical content is present in HTML without JS execution (SSR/SSG).

    3 pts
  4. 04

    An XML sitemap is generated and submitted to GSC.

    2 pts
  5. 05

    Every page carries a coherent absolute canonical.

    2 pts
  6. 06

    schema.org structured data is validated (Rich Results Test).

    2 pts
  7. 07

    Core Web Vitals: LCP < 2.5 s, CLS < 0.1, INP < 200 ms.

    3 pts
  8. 08

    HTTP/2 or HTTP/3, gzip/brotli compression, HTTPS.

    2 pts
  9. 09

    Images in AVIF/WebP, loading="lazy" below the fold.

    1 pt
  10. 10

    llms.txt and llms-full.txt published and consistent with actual site structure.

    1 pt

Dimension P

Performance (10 points)

Are you steering the effort with metrics and a cadence?

23 pts

  1. 01

    Multi-engine baseline on 10 to 20 strategic queries.

    3 pts
  2. 02

    Google Search Console connected; AI Overviews report reviewed.

    2 pts
  3. 03

    AI citation monitoring tool configured (Profound, Peec, Otterly, Scrunch) or manual bi-monthly monitoring.

    3 pts
  4. 04

    Server logs analysed for at least GPTBot, PerplexityBot, ClaudeBot.

    2 pts
  5. 05

    Monthly citation review compared against the baseline.

    2 pts
  6. 06

    Quarterly pillar-page review with genuinely updated dates.

    3 pts
  7. 07

    Internal log of behavioural changes observed.

    2 pts
  8. 08

    Target KPIs defined (positions, impressions, share of citations).

    2 pts
  9. 09

    Simple reporting shared with leadership or the team.

    2 pts
  10. 10

    Documented editorial cadence: publishing frequency, refresh frequency.

    2 pts

Next

You have your score. Now what?

  1. 1. Run the 6-step method.
  2. 2. Use the LOOP framework to prioritise.
  3. 3. Check the use cases page to adapt to your sector.
  4. 4. Re-test in 4 to 6 weeks, compare, iterate.

À lire ensuite