itstheflashpack.com — AI Site Grade

Analysis

The Flash Pack (itstheflashpack.com) — AI-Visibility Audit

The site's homepage is fully accessible to ChatGPT-User, OAI-SearchBot, PerplexityBot, and Perplexity-User, yet GPTBot, ClaudeBot, Google-Extended, and anthropic-ai all receive zero-byte disconnects — the server drops the TCP connection before sending any response. This is not a robots.txt block (the file is wide-open with User-agent: * Disallow:) but a WAF-level or nginx-level UA rejection. The result is that OpenAI's training crawler (GPTBot) and Google's AI-training crawler (Google-Extended) cannot read a single page, while the retrieval bots for ChatGPT and Perplexity can. This selective blocking means the site's content will not appear in the training data of the two largest frontier model families.

Crawler Access

The robots.txt at /robots.txt is a bare Yoast-generated file with a single User-agent: * Disallow: rule and a sitemap reference. No AI-specific directives exist — no GPTBot, ClaudeBot, Google-Extended, or anthropic-ai rules are defined. The /llms.txt returns a 404, meaning no AI-friendly content map exists. DNS records show the site is hosted on a single nginx server (35.214.54.37, likely a Vultr or similar VPS) behind AWS Route53 DNS, with no Cloudflare or major CDN visible. The x-cdn-c and x-proxy-cache headers suggest a caching proxy layer, but the server fingerprint is plain nginx. The selective bot blocking appears to be implemented at the application or proxy level, not via a standard WAF dashboard.

Cold-Knowledge Gap

A frontier LLM queried cold about "itstheflashpack" returned zero knowledge — the model had no record of the brand, its services, or any notable clients. This is a severe gap for a company that claims to have built the Oscars backstage photo booth for the 98th Academy Awards, executed activations for Netflix (Arcane Season 2), Samsung (the £500K Selfie Tour), New Balance, Burberry, Nike, and Liverpool FC. The site itself presents a strong portfolio of blue-chip clients, but none of this brand equity is reflected in what AI models know without live retrieval. The gap is compounded by the fact that GPTBot (OpenAI's training crawler) is blocked, so even future training runs will not absorb this information.

Content & Schema Posture

The homepage and all key pages use Yoast-generated JSON-LD schema with WebPage, BreadcrumbList, WebSite, and Organization types. The Organization schema includes sameAs links to Facebook, X, Instagram, LinkedIn, and YouTube. However, no Article, FAQPage, Product, VideoObject, or CreativeWork schema is used on case study pages — the Arcane activation, Oscars booth, and Samsung tour pages all lack structured data describing the project, client, or outcomes. Case study pages contain rich narrative content (300-550 words) with specific metrics ("10K+ selfies snapped, 100% content share rate") but no schema to surface these facts in AI-generated answers. The blog section ("The Lens") has 12 posts, the most recent being the Oscars photo booth article dated March 2026, indicating active publishing. No FAQ or comparison-table signals were detected anywhere on the site.

External Signals

Web searches for the brand name, its clients, and its projects returned zero indexed results from major news outlets, press releases, or review sites. The brand has no detectable press coverage, no third-party articles mentioning its work, and no external reviews. The only off-domain presence is its own social media profiles (Instagram, Vimeo, LinkedIn, YouTube) and a Google-site-verification TXT record. This absence of external citations means AI engines have no third-party signals to corroborate the brand's claims, making the site the sole source of truth — and that source is partially blocked to key crawlers.

Findings

1. Server drops GPTBot and Google-Extended connections High

   The server disconnects GPTBot, ClaudeBot, Google-Extended, and anthropic-ai without sending any response, while allowing ChatGPT-User, OAI-SearchBot, PerplexityBot, and Perplexity-User. This selective blocking prevents training crawlers from accessing any content.

   What to change: Remove the server-level UA-based blocking for GPTBot and Google-Extended, or configure the WAF to allow all legitimate AI crawlers.

2. No /llms.txt file published Medium

   The site returns a 404 for /llms.txt, missing an opportunity to provide an AI-friendly content map.

   What to change: Create an /llms.txt file listing key pages and summaries for AI crawlers.

3. Zero AI knowledge of the brand High

   A frontier LLM queried cold about 'itstheflashpack' returned no knowledge of the brand, its services, or its blue-chip clients (Netflix, Samsung, Oscars).

   What to change: Ensure GPTBot and Google-Extended are allowed, and build external press coverage to create third-party signals.

4. Case study pages lack structured data Medium

   Case study pages for Arcane, Oscars, and Samsung tours contain rich narrative content but no Article, CreativeWork, or Product schema to surface facts in AI answers.

   What to change: Add Article or CreativeWork JSON-LD schema to case study pages, including client name, metrics, and outcomes.

5. Zero external press or third-party citations High

   Web searches for the brand, its clients, and projects returned no results from news outlets, press releases, or review sites. No third-party articles corroborate the brand's claims.

   What to change: Pitch press releases and case studies to industry publications to generate third-party backlinks and citations.

6. No FAQ or comparison schema detected Low

   The site does not use FAQPage or comparison schema anywhere, missing opportunities to appear in rich AI answers.

   What to change: Add FAQ schema to relevant pages, such as services or case studies.

7. Robots.txt lacks AI-specific directives Low

   The robots.txt file only has a single User-agent: * rule and does not explicitly allow or disallow GPTBot, ClaudeBot, or other AI crawlers.

   What to change: Add explicit rules for GPTBot, ClaudeBot, and Google-Extended to ensure intended access.

What's working

• Homepage accessible to ChatGPT and Perplexity retrieval bots — ChatGPT-User, OAI-SearchBot, PerplexityBot, and Perplexity-User can access the homepage, enabling real-time AI retrieval.
• Organization schema with social profiles — The site uses Yoast-generated JSON-LD with Organization schema including sameAs links to Facebook, X, Instagram, LinkedIn, and YouTube.
• Active blog with recent posts — The Lens blog has 12 posts, the most recent dated March 2026, showing active content publishing.
• Sitemap index with 80 URLs — A sitemap index is available at /sitemap_index.xml, listing 80 URLs for crawler discovery.
• Detailed case study content with metrics — Case study pages contain 200-550 words with specific metrics (e.g., '10K+ selfies snapped, 100% content share rate'), providing valuable narrative for AI.