Learn how to read Argos findings on a demo product page: understand the evidence, choose an action, then verify the correction.
An educational example prepared from the HTML files below, not a report from a complete Argos run. No score or performance gain is presented as measured. No customer site or user report is used.
Preview the Argos report
Fictional data: scores, coverage and mosaic values illustrate the layout, without a real run. The three HTML examples are explained below. illustrative-report-v2
Example · Fictional data
Audit report
Atelier Démo
https://atelier.example/produit/sac
Scope: this URL only. The whole domain has not been crawled.
The score averages measured checks; it measures neither business urgency nor missing checks.
Severity comes from the source check: “critical” may describe a tool threshold or an observed defect. Check the evidence and effect on the journey before deciding urgency.
The link opens a public summary. Evidence and corrections require sign-in. The report is excluded from indexing.
Where to start
The order follows the scoring model. Check evidence before choosing corrections. Confidence describes measurement reliability; effort still needs estimation on the site.
01
Important · SEO
Review the page description
Without a specific summary, search previews may lack context.
First step: Locate this URL’s meta description inside <head>.
Informative images without alternatives may lose their meaning in assisted reading.
First step: In Network → main document → Response, locate img elements without alt. If Argos retains no location, this investigation is necessary; img:not([alt]) helps inspect the rendered DOM, which may differ from received HTML.
An unclear title can make the page hard to recognise in tabs and results.
First step: In browser tools, open Network, select the main document, then Response and search for <title>. Compare it with the tab title if JavaScript changes it.
The score summarizes measured checks included in the method. Coverage shows how many planned checks were actually measured: interpret a high score with low coverage cautiously.
Performance: loading and execution. Security: observable technical protections. SEO: search engine access and understanding. Accessibility: automated checks, complemented by human review.
Severity helps prioritize a finding. Reliability describes confidence in detection; it measures neither the score nor severity. Metrics describe this execution and provide no guarantee about unaudited pages.
Share of measured check weights in the expected scoring policy, excluding checks that do not apply.
The success rate counts fully successful modules, excluding partial collections, failures and skipped modules. Terminal progress also counts those finished states; both values are distinct from weighted scoring coverage.
Confidence qualifies each finding; it cannot be inferred from the score or coverage.
Check states
Measured: 8
Unavailable: 0
Blocked by anti-bot protection: 0
Not applicable: 0
Coverage breakdown
100 % measured — 0 % not covered
2 / 2 · illustrative-report-v2
Sufficient coverage threshold: 80 %
Values frozen in this report; no recalculation using the current policy. Contributions are rounded individually.
The retained breakdown does not fully reconcile with the recorded total. Missing checks cannot be invented.
Sources present in the report
html
Available versions
Coverage: illustrative-report-v2
Displayed editorial catalogue: report-content-v11
Priority ranking: illustrative-report-v2
Scoring policy version not provided
Lighthouse scores are tool scores, separate from Argos domain scores.
Automated accessibility does not certify RGAA or WCAG compliance. Human checks are still necessary.
Results describe only the measured scope; they guarantee neither compliance nor business outcomes.
Confirm priority signals, estimate corrections, then re-measure under comparable conditions.
A fictional store with one product page, no accounts, orders or data collection. We control both versions and their assets. The before version has deliberately introduced issues; the links let you inspect the pages and their source in your browser. Demo pages are in French.
The title ‘Sac’ is only 3 characters long. The HTML check flags a very short title; it does not evaluate editorial relevance. A descriptive title helps identify a page in a browser tab or search result.
Before · HTML evidence
<title>Sac</title>
After · HTML correction
<title>Sac en coton réutilisable | Atelier Démo</title>
How to verify
Compare the title tags in both sources, then rerun the HTML check. Manually confirm that the title describes the product. This does not prove better Google rankings.
SEO: missing description
Related Argos check : html.meta.description.present
The before version has no meta description. The check can identify that absence. The correction provides a coherent summary, but a search engine may choose a different snippet.
Before · HTML evidence
<!-- No meta description -->
After · HTML correction
<meta name="description" content="Découvrez le sac en coton réutilisable Atelier Démo, ses dimensions et ses conseils d’entretien.">
How to verify
Find meta name="description" in the corrected source. Argos checks this technical signal; review the text for clarity and consistency with the content.
Accessibility: image without alternative text
Related Argos check : html.images.alt_coverage
The only image in the before version has no alt attribute: 1 out of 1 images lack an alternative. The HTML check detects the missing attribute. We add a description for this informative image; a decorative image would need alt="".
<img src="/demo/atelier/sac.svg" width="320" height="240" alt="Sac en coton écru avec deux anses">
How to verify
Compare image attributes: 0 out of 1 images have a missing alt attribute after correction. Then assess the text with a human review and a screen reader; an attribute alone does not prove accessibility.
How should you prioritize these corrections?
Confirm all three pieces of evidence on the product page and the checks actually completed. Here, no real purchase journey exists and no security risk has been assessed.
Add alternative text for the informative image, then write the title and description. These simple changes can ship together; their order depends on your visitors and goals.
Validate the delivered HTML, rerun an audit of the same page, then manually review content and keyboard access. On a real store, a blocked purchase would come before these improvements.
What this example does not demonstrate
Evidence covers only three HTML checks. HTTP headers, TLS, security services, browser measurements and Lighthouse scores are not measured here. These three pieces of evidence alone cannot produce a measured overall score. Both demo pages remain noindex to keep the fictional store out of search; this is intentional and unchanged after correction.
Report links open a public summary; detailed evidence and corrections require sign-in. Reports are excluded from indexing. This public page is a separate editorial resource with no access to user reports.
Put this diagnosis in context
This illustration shows how to read a report. Audit your own URL to obtain your findings.
An automated score does not certify security or accessibility. Argos and Lighthouse use different scoring policies.