perk.com — AI Site Grade

Analysis

The cold-knowledge gap is total: frontier LLMs describe perk.com as a defunct consumer rewards platform (acquired by Swagbucks in 2018), while the actual site is a $2.7B corporate travel-and-spend unicorn that rebranded from TravelPerk in November 2025.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended — receive a full 200 response with identical byte-size content (~1.4 MB) to a browser. No UA-based blocking exists. The site runs on AWS CloudFront behind an API Gateway, with no WAF rules discriminating by user-agent. The robots.txt at perk.com/robots.txt contains zero AI-bot-specific directives — only a generic * rule blocking /wp-admin/, /en-us/, and career subpaths. The llms.txt at perk.com/llms.txt returns a 404 (serving the full JS application shell, ~445 KB). The sitemap at perk.com/sitemap/global-index.xml is well-structured with 9 locale-specific sub-sitemaps covering blog, guides, resources, media, pages, case studies, and legal content.

Cold-Knowledge Gap

Asked cold, a frontier LLM describes perk.com as a consumer rewards platform where users earned "Perk Points" watching videos and taking surveys, acquired by Prodege/Swagbucks in 2018, with the domain now redirecting to Swagbucks. This is the entirely wrong company. The actual site is Perk (formerly TravelPerk, rebranded November 2025), a B2B SaaS platform combining corporate travel booking, expense management, invoice processing, and event management — valued at $2.7 billion after a 2025 Series E led by SoftBank. The Wikipedia page for "Perk (company)" correctly documents the TravelPerk history, but the cold LLM conflates the domain with the defunct rewards brand that previously owned perk.com. This is a domain-history collision: the old perk.com rewards platform left a stronger training-signal footprint than the new B2B entity.

Schema Posture

The site uses WebSite and Organization JSON-LD on every page, with alternateName: ["TravelPerk", "Yokoy"] and sameAs pointing to Wikipedia, Wikidata, and LinkedIn. The pricing page adds a FAQPage schema with 5 questions about billing. The platform page also carries FAQPage schema with 6 questions about travel-and-spend management. However, no page uses Product, SoftwareApplication, or BreadcrumbList schema — surprising for a SaaS platform with tiered plans. The blog has 394+ articles but no Article or BlogPosting schema detected on the blog index page.

External Signals

The site prominently links to CNBC, Forbes, Bloomberg, and Skift coverage of its fundraising and acquisitions (AmTrav in 2024, Yokoy in 2025). The customers page features case studies from Lush, On Running, Fabletics, Nord Security, PureGym, Emma Sleep, Caudalie, and Storyblok — strong social proof. The DNS TXT records reveal integrations with OpenAI, Anthropic, Intercom, Stripe, Notion, Figma, MongoDB, and 1Password — signaling deep enterprise tech stack adoption. The footer references a 2026 copyright date and an Audi F1 Team sponsorship.

Content Architecture

The homepage is a JS-rendered Astro application (~1.4 MB delivered) that nonetheless serves full HTML content to crawlers — no hydration wall. The content is rich but repetitive: the same feature blocks ("Seamlessly book and manage trips", "Make trip changes in seconds") appear 4+ times across the page in different sections. The blog has 394 posts across categories (Spend Management, AI & Automation, Sustainability, etc.) but the blog index page delivers only 158 words of visible text — mostly titles and category labels. The site maintains 9 locale-specific versions (en-ca, uk, de, de-ch, es, fr, nl, it) with separate sitemaps, suggesting strong international SEO investment.

Findings

1. Cold-knowledge gap: LLMs describe perk.com as a defunct rewards platform High

   Frontier LLMs incorrectly identify perk.com as a consumer rewards platform acquired by Swagbucks in 2018, while the actual site is a $2.7B corporate travel-and-spend platform rebranded from TravelPerk. This domain-history collision severely undermines AI visibility for the correct entity.

   What to change: Publish structured data (e.g., Organization schema with sameAs and alternateName) on the homepage and key pages to disambiguate from the old rewards brand. Submit the correct entity to Wikidata and Wikipedia. Consider a PR campaign to update training data sources.

2. Missing llms.txt file returns 404 Medium

   The llms.txt file at perk.com/llms.txt returns a 404 error, serving the full JS application shell instead of a helpful AI resource. This is a missed opportunity to guide AI crawlers with accurate context about the site.

   What to change: Create an llms.txt file at the root that provides a brief description of the site, links to key pages, and disambiguates from the old rewards brand.

3. Missing Product and SoftwareApplication schema on SaaS pages Medium

   No page uses Product or SoftwareApplication schema, which is surprising for a SaaS platform with tiered plans. This limits the ability of AI crawlers to understand the software offering and pricing.

   What to change: Add SoftwareApplication schema to the platform page and Product schema to the pricing page, including offers, applicationCategory, and operatingSystem.

4. Missing BreadcrumbList schema across the site Low

   No page uses BreadcrumbList schema, which helps AI crawlers understand site hierarchy and navigation.

   What to change: Add BreadcrumbList structured data to all pages to improve navigation understanding for AI crawlers.

5. Missing Article schema on blog index and posts Medium

   The blog index page delivers only 158 words of visible text and lacks Article or BlogPosting schema, reducing the discoverability of 394+ blog posts by AI crawlers.

   What to change: Add Article schema to each blog post and BlogPosting schema to the blog index. Ensure each post has proper headline, datePublished, and author fields.

6. Repetitive content on homepage reduces information density Low

   The same feature blocks appear 4+ times across the homepage in different sections, reducing the unique information density for AI crawlers.

   What to change: Consolidate repetitive sections to provide more unique content per page, improving the signal-to-noise ratio for AI crawlers.

7. Robots.txt blocks AI bots from /en-us/ and career subpaths Low

   The robots.txt file contains a generic rule blocking /en-us/ and career subpaths, which may inadvertently block AI crawlers from accessing localized content or career pages.

   What to change: Review the robots.txt rules to ensure AI crawlers are not blocked from important content. Consider allowing /en-us/ if it contains unique content.

What's working

• All major AI crawlers receive full 200 responses — All tested AI crawlers (GPTBot, ClaudeBot, PerplexityBot, etc.) receive identical content to a browser, with no user-agent-based blocking. The site runs on AWS CloudFront with no discriminatory WAF rules.
• Well-structured sitemap with locale-specific sub-sitemaps — The sitemap at /sitemap/global-index.xml contains 9 locale-specific sub-sitemaps covering blog, guides, resources, media, pages, case studies, and legal content, indicating strong international SEO investment.
• Organization and WebSite JSON-LD present on every page — Every page includes WebSite and Organization JSON-LD with alternateName (TravelPerk, Yokoy) and sameAs links to Wikipedia, Wikidata, and LinkedIn, helping AI crawlers understand the entity.
• FAQPage schema on pricing and platform pages — The pricing page includes FAQPage schema with 5 questions about billing, and the platform page includes FAQPage schema with 6 questions about travel-and-spend management, providing structured Q&A content.
• Strong external signals from media coverage and customer case studies — The site prominently links to CNBC, Forbes, Bloomberg, and Skift coverage, and features case studies from well-known brands like Lush, On Running, and Fabletics, providing strong social proof and backlink potential.
• JS-rendered Astro application serves full HTML to crawlers — Despite being a JS-rendered Astro application, the homepage serves full HTML content to crawlers with no hydration wall, ensuring content is indexable.
• DNS TXT records reveal deep enterprise tech stack integrations — DNS TXT records show integrations with OpenAI, Anthropic, Intercom, Stripe, Notion, Figma, MongoDB, and 1Password, signaling strong enterprise adoption and technical credibility.