dext.com — AI Site Grade

Analysis

I have enough data to write a sharp audit. Let me compile the findings.

---

Dext.com has a major cold-knowledge gap: the LLM model knows about Dext Precision, Dext Commerce, and Receipt Bank's rebranding — but the live site has completely buried or removed those products.

Crawler Access

Every major AI crawler — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, anthropic-ai, Applebot-Extended, Bytespider — receives a full 200 response with identical byte-size content (477 KB) on the homepage. No UA-based blocking exists. The site serves from AWS CloudFront (AmazonS3 origin) with HSTS, X-Frame-Options: SAMEORIGIN, and X-Content-Type-Options: nosniff. The robots.txt has a single User-agent: * Allow: / rule with no AI-bot-specific directives and a Content-Signal: ai-train=no, search=yes, ai-input=yes header. The llms.txt exists (22 KB) but is a raw crawl4AI dump of the navigation — not a curated AI-friendly content map. Blog article URLs from the listing (e.g., /en/blog/ai-in-accounting-from-hype-to-practical-impact-for-modern-firms) return 404 with a thin JS shell, meaning the blog listing page links to articles that do not resolve.

Cold-Knowledge Gap

The LLM's prior knowledge describes Dext Prepare (formerly Receipt Bank), Dext Precision (automated bookkeeping/reconciliation), and Dext Commerce (e-commerce data integration) as core products. The live site mentions none of these by name. The homepage and product pages only reference "Dext" as a unified bookkeeping automation platform with features like "AI Assist" and "Payments." The /en/receipt-bank page redirects to a thin page saying "Receipt Bank is now Dext" with no product detail. Dext Precision and Dext Commerce are entirely absent from the crawled site — no sitemap entry, no page, no mention. The LLM also knows Dext processes "over 100 million documents annually" and was founded in 2010 as Receipt Bank; the site claims "over 700,000 customers worldwide" and "over a decade" but does not state document volume or founding year explicitly.

Schema Posture

Every page carries a rich Organization schema with global contact points (UK, US, France, South Africa, APAC), sameAs links to 8 social profiles, and a MobileApplication schema for the iOS app with an AggregateRating of 4.7 from 2,500 ratings. The homepage and pricing page include FAQPage schema. However, no Product schema exists for any of the software tiers (Prepare, Precision, Commerce), and the WebApplication schema is present but generic. The blog listing page has zero schema markup — no BlogPosting, no CollectionPage, no Article.

External Signals

The DNS TXT records reveal OpenAI domain verification (openai-domain-verification=dv-dnsw6eu5oIzWx35wLU3VGkB0), Anthropic domain verification (anthropic-domain-verification-t8w7na=...), and Cursor domain verification, confirming Dext has proactively registered with all three AI platforms for potential retrieval-augmented generation access. The site also has a hidden /en/iris-intent-to-acquire-dext-faq page confirming IRIS Software Group's intent to acquire Dext — a material corporate event that the LLM's cold knowledge does not mention. The FAQ states Dext will remain a separate operating unit under IRIS, CEO Sabby Gill stays, and product identity is maintained. This acquisition is not referenced anywhere on the homepage, pricing page, or blog.

Content Fragmentation

The site uses a geo-redirect architecture: https://dext.com immediately 302s to /uk (or /en depending on locale). The sitemap contains 9 regional sub-sitemaps (UK, US, AU, CA, FR, ZA, etc.) plus a blog sitemap, totaling 495+ URLs. The blog listing shows articles dated as far forward as May 2026 — future-dated content that may confuse freshness signals. The /en/news page lists press releases but contains no actual press release content, only a contact email. The /en/careers page exists in the sitemap but was not crawled; the /en/diversity page exists. The llms.txt is auto-generated crawl output rather than a curated summary, undermining its utility for AI consumption.

Findings

1. Core products Dext Precision and Dext Commerce absent from live site High

   The LLM's prior knowledge describes Dext Precision (automated bookkeeping) and Dext Commerce (e-commerce data integration) as core products, but the live site does not mention either by name. No sitemap entry, page, or content references these products, creating a cold-knowledge gap.

   What to change: Add dedicated product pages for Dext Precision and Dext Commerce, include them in the sitemap, and reference them on the homepage and pricing page.

2. IRIS acquisition FAQ exists but is not linked from main site High

   A page at /en/iris-intent-to-acquire-dext-faq confirms IRIS Software Group's intent to acquire Dext, but this page is not referenced on the homepage, pricing page, or blog. The LLM's cold knowledge does not mention this acquisition.

   What to change: Prominently link the acquisition FAQ from the homepage or about page, and add structured data (e.g., NewsArticle) to signal the event to AI crawlers.

3. Blog listing links to articles that return 404 High

   The blog listing page includes links to articles such as /en/blog/ai-in-accounting-from-hype-to-practical-impact-for-modern-firms that return 404 with a thin JS shell. This wastes crawl budget and degrades user experience.

   What to change: Remove or redirect broken blog article links, and implement proper 301 redirects for any removed content.

4. No Product schema for software tiers Medium

   Despite having Organization, FAQPage, and MobileApplication schema, the site lacks Product schema for its software tiers (Prepare, Precision, Commerce). This limits rich result eligibility in AI-generated answers.

   What to change: Add Product schema markup for each software tier, including name, description, offers, and aggregate rating where applicable.

5. Blog listing page has zero schema markup Medium

   The blog listing page at /en/blog lacks any structured data such as BlogPosting, CollectionPage, or Article schema, reducing its visibility in AI-driven search results.

   What to change: Add BlogPosting and CollectionPage schema to the blog listing page, and ensure individual articles have Article schema.

6. llms.txt is an auto-generated crawl dump, not a curated summary Medium

   The llms.txt file (22 KB) contains a raw crawl4AI dump of the navigation rather than a curated, AI-friendly content summary. This undermines its utility for AI consumption and may confuse crawlers.

   What to change: Replace the llms.txt with a curated summary of key content, including product descriptions, pricing, and important pages, formatted for AI consumption.

7. Blog listing includes articles dated May 2026 Low

   The blog listing shows articles with future dates (May 2026), which may confuse freshness signals for AI crawlers and search engines.

   What to change: Correct the publication dates on blog articles to reflect actual publish dates, and avoid publishing content with future dates.

8. News page contains no press release content Low

   The /en/news page lists press releases but contains no actual press release content, only a contact email. This provides no value to AI crawlers or users seeking news.

   What to change: Include full press release content on the news page, or link to external press release pages with proper schema.

What's working

• All major AI crawlers receive full access with no blocking — Every major AI crawler (GPTBot, ClaudeBot, PerplexityBot, etc.) receives a 200 response with identical content. No UA-based blocking exists, and robots.txt allows all.
• Domain verified with OpenAI, Anthropic, and Cursor for RAG access — DNS TXT records include verification tokens for OpenAI, Anthropic, and Cursor, indicating proactive registration for retrieval-augmented generation access.
• Rich Organization schema with global contact points and social profiles — Every page carries Organization schema with global contact points (UK, US, France, South Africa, APAC), sameAs links to 8 social profiles, and MobileApplication schema with aggregate rating.
• FAQPage schema present on homepage and pricing page — The homepage and pricing page include FAQPage schema, which can enable rich results in AI-generated answers.
• llms.txt file is present and accessible — An llms.txt file exists at the root, providing a starting point for AI crawlers to discover content, even though it is not curated.
• Sitemap covers 495+ URLs across 9 regional sub-sitemaps — The sitemap includes 9 regional sub-sitemaps (UK, US, AU, CA, FR, ZA, etc.) plus a blog sitemap, ensuring broad coverage of localized content.
• Receipt Bank redirect page acknowledges rebranding — The /en/receipt-bank page redirects to a page stating 'Receipt Bank is now Dext', helping users and crawlers understand the rebranding.