Enter password to view Website Audit

Analysis

Website

Neo4j

Analysis

Website

Neo4j

Analysis

Website

Neo4j

Published on

2026-03-18

For

Neo4j

Score

43

Neo4j is the world's leading graph intelligence platform, providing graph database, graph analytics, and graph AI infrastructure for building connected data applications and agentic AI systems. Powers fraud detection, knowledge graphs, recommendation engines, network topology, drug discovery, and real-time AI. Commands 44% graph database market share. Revenue surpassed $200M (2024). Backed by Eurazeo, GV (Google), Deutsche Telekom, Goldman Sachs. $581M total raised.

Market

Graph Database / Graph Analytics / Graph AI / Knowledge Graphs / GraphRAG / Agentic AI Infrastructure

Audience

Enterprise architects, data engineers, data scientists, ML/AI engineers, developers building connected data applications, fraud detection teams, and agentic AI developers

HQ

San Mateo, CA

CopyPerformanceCopyNavigationSEOCopyPerformanceCopyNavigationCopy

Copy

38

Performance

35

Copy

42

Navigation

44

SEO

47

Copy

45

Performance

40

Copy

52

Navigation

48

Copy

36

Copy

Hero H1 'Build Intelligent Apps and AI For Real-Time Recommendations' Is One Use Case, Not the Platform

Score

38

Severity

High

Finding

The homepage H1 reads: 'Build Intelligent Apps and AI For Real-Time Recommendations.' This is a single use case headline — real-time recommendations — positioned as the defining purpose of the world's leading graph intelligence platform. Neo4j's customer base uses the product for fraud detection (BNP Paribas), agentic AI memory (Merck), traffic management (Transport for London), supply chain mapping (Intuit), drug discovery (QIAGEN), and network topology (Uber) — none of which are 'real-time recommendations.' The sub-label 'AI-ready data by design' is better, but the H1 actively misdirects enterprise buyers evaluating the platform for anything other than recommendation engines. The page title ('Neo4j Graph Intelligence Platform') and the hero H1 don't agree on what the product does.

Recommendation

Replace the H1 with a platform-level claim that reflects the full use case breadth: 'The Graph Intelligence Platform Powering AI That Knows What It Knows.' Or align with the agentic AI narrative Neo4j announced in October 2025: 'Give Your AI a Memory. Build Agentic Systems on the World's Leading Graph Platform.' The current H1 was likely written for a product marketing campaign around recommendation engines and was never updated when Neo4j pivoted its messaging toward agentic AI and GraphRAG. The TechCrunch headline from November 2024 is a better homepage H1 than the current one: 'The platform that makes AI accurate, explainable, and connected.'

Copy

Hero H1 'Build Intelligent Apps and AI For Real-Time Recommendations' Is One Use Case, Not the Platform

Score

38

Severity

High

Finding

The homepage H1 reads: 'Build Intelligent Apps and AI For Real-Time Recommendations.' This is a single use case headline — real-time recommendations — positioned as the defining purpose of the world's leading graph intelligence platform. Neo4j's customer base uses the product for fraud detection (BNP Paribas), agentic AI memory (Merck), traffic management (Transport for London), supply chain mapping (Intuit), drug discovery (QIAGEN), and network topology (Uber) — none of which are 'real-time recommendations.' The sub-label 'AI-ready data by design' is better, but the H1 actively misdirects enterprise buyers evaluating the platform for anything other than recommendation engines. The page title ('Neo4j Graph Intelligence Platform') and the hero H1 don't agree on what the product does.

Recommendation

Replace the H1 with a platform-level claim that reflects the full use case breadth: 'The Graph Intelligence Platform Powering AI That Knows What It Knows.' Or align with the agentic AI narrative Neo4j announced in October 2025: 'Give Your AI a Memory. Build Agentic Systems on the World's Leading Graph Platform.' The current H1 was likely written for a product marketing campaign around recommendation engines and was never updated when Neo4j pivoted its messaging toward agentic AI and GraphRAG. The TechCrunch headline from November 2024 is a better homepage H1 than the current one: 'The platform that makes AI accurate, explainable, and connected.'

Copy

Hero H1 'Build Intelligent Apps and AI For Real-Time Recommendations' Is One Use Case, Not the Platform

Score

38

Severity

High

Finding

The homepage H1 reads: 'Build Intelligent Apps and AI For Real-Time Recommendations.' This is a single use case headline — real-time recommendations — positioned as the defining purpose of the world's leading graph intelligence platform. Neo4j's customer base uses the product for fraud detection (BNP Paribas), agentic AI memory (Merck), traffic management (Transport for London), supply chain mapping (Intuit), drug discovery (QIAGEN), and network topology (Uber) — none of which are 'real-time recommendations.' The sub-label 'AI-ready data by design' is better, but the H1 actively misdirects enterprise buyers evaluating the platform for anything other than recommendation engines. The page title ('Neo4j Graph Intelligence Platform') and the hero H1 don't agree on what the product does.

Recommendation

Replace the H1 with a platform-level claim that reflects the full use case breadth: 'The Graph Intelligence Platform Powering AI That Knows What It Knows.' Or align with the agentic AI narrative Neo4j announced in October 2025: 'Give Your AI a Memory. Build Agentic Systems on the World's Leading Graph Platform.' The current H1 was likely written for a product marketing campaign around recommendation engines and was never updated when Neo4j pivoted its messaging toward agentic AI and GraphRAG. The TechCrunch headline from November 2024 is a better homepage H1 than the current one: 'The platform that makes AI accurate, explainable, and connected.'

Performance

Customer Logo Strip Loads the Same 22 Logos Four Times in DOM — 88 Image Elements

Score

35

Severity

High

Finding

The homepage customer logo carousel contains 22 distinct customer logos (Uber, Cisco, Walmart, Santander, Comcast, BNP Paribas, UBS, Airbus, BT Group, BMW, Boston Scientific, Novo Nordisk, Dun & Bradstreet, Intel, Intuit, Adobe, Levi's, NASA, NBC News, Siemens, Transport for London, Merck, eBay). The HTML source shows these 22 logos loaded exactly four times each — 88 total <img> elements — to create an infinite-scroll animation effect. This is the identical DOM-duplication pattern flagged in audits of coco.delivery, durable.com, and PandaDoc: CSS-based infinite carousels that clone the entire DOM rather than using CSS transforms on a single set. 88 image HTTP requests for a logo strip that could be served with 22 requests plus CSS animation is a significant performance overhead on first paint.

Recommendation

Refactor the logo carousel to use a single set of 22 logo images animated with CSS transforms (translateX) and the animation-iteration-count: infinite property, rather than duplicating the full set four times in the DOM. The duplication is only needed in CSS carousel implementations that lack proper infinite-loop support — modern CSS animations do not require DOM duplication. Audit with Lighthouse; the 88 image requests will flag as a render-blocking and LCP issue. At minimum, add loading='lazy' to all logo images beyond the first visible row, and ensure all logo SVGs are served inline or from a single sprite sheet.

Performance

Customer Logo Strip Loads the Same 22 Logos Four Times in DOM — 88 Image Elements

Score

35

Severity

High

Finding

The homepage customer logo carousel contains 22 distinct customer logos (Uber, Cisco, Walmart, Santander, Comcast, BNP Paribas, UBS, Airbus, BT Group, BMW, Boston Scientific, Novo Nordisk, Dun & Bradstreet, Intel, Intuit, Adobe, Levi's, NASA, NBC News, Siemens, Transport for London, Merck, eBay). The HTML source shows these 22 logos loaded exactly four times each — 88 total <img> elements — to create an infinite-scroll animation effect. This is the identical DOM-duplication pattern flagged in audits of coco.delivery, durable.com, and PandaDoc: CSS-based infinite carousels that clone the entire DOM rather than using CSS transforms on a single set. 88 image HTTP requests for a logo strip that could be served with 22 requests plus CSS animation is a significant performance overhead on first paint.

Recommendation

Refactor the logo carousel to use a single set of 22 logo images animated with CSS transforms (translateX) and the animation-iteration-count: infinite property, rather than duplicating the full set four times in the DOM. The duplication is only needed in CSS carousel implementations that lack proper infinite-loop support — modern CSS animations do not require DOM duplication. Audit with Lighthouse; the 88 image requests will flag as a render-blocking and LCP issue. At minimum, add loading='lazy' to all logo images beyond the first visible row, and ensure all logo SVGs are served inline or from a single sprite sheet.

Performance

Customer Logo Strip Loads the Same 22 Logos Four Times in DOM — 88 Image Elements

Score

35

Severity

High

Finding

The homepage customer logo carousel contains 22 distinct customer logos (Uber, Cisco, Walmart, Santander, Comcast, BNP Paribas, UBS, Airbus, BT Group, BMW, Boston Scientific, Novo Nordisk, Dun & Bradstreet, Intel, Intuit, Adobe, Levi's, NASA, NBC News, Siemens, Transport for London, Merck, eBay). The HTML source shows these 22 logos loaded exactly four times each — 88 total <img> elements — to create an infinite-scroll animation effect. This is the identical DOM-duplication pattern flagged in audits of coco.delivery, durable.com, and PandaDoc: CSS-based infinite carousels that clone the entire DOM rather than using CSS transforms on a single set. 88 image HTTP requests for a logo strip that could be served with 22 requests plus CSS animation is a significant performance overhead on first paint.

Recommendation

Refactor the logo carousel to use a single set of 22 logo images animated with CSS transforms (translateX) and the animation-iteration-count: infinite property, rather than duplicating the full set four times in the DOM. The duplication is only needed in CSS carousel implementations that lack proper infinite-loop support — modern CSS animations do not require DOM duplication. Audit with Lighthouse; the 88 image requests will flag as a render-blocking and LCP issue. At minimum, add loading='lazy' to all logo images beyond the first visible row, and ensure all logo SVGs are served inline or from a single sprite sheet.

Copy

Hero Stat '80+ Fortune 100 Customers' — Conflicts With Other Public Claims

Score

42

Severity

High

Finding

The homepage hero stats bar shows: '300k Developers Building', '80+ Fortune 100 Customers', '170+ Partner Ecosystem'. However Neo4j's own press releases and TechCrunch coverage (November 2024) state '84% of the Fortune 100' — which at 100 Fortune 100 companies means 84 customers, not 80+. An earlier Series F press release (2021) stated '75% of the Fortune 100.' The homepage claim of '80+' sits between these two figures. The discrepancy is likely a rounding artefact — '84% of Fortune 100' rounds to 84, which is both '80+' and lower than '84%' — but the headline stat and the press release stat use different framing that creates ambiguity. More importantly, the 84% figure is significantly more impressive-sounding than '80+' and should replace it.

Recommendation

Update the hero stat to match the most recent verified claim: '84% of the Fortune 100' or 'Trusted by 84 Fortune 100 companies.' The '84%' framing is more persuasive than '80+' and aligns with the TechCrunch coverage that is actively indexed. While updating stats, also verify the '300k developers' figure — the GraphAcademy nav link references '100K+ Neo4j experts' for certified developers specifically, which conflicts with the '300k developers building' hero claim. A CTO evaluating Neo4j who sees '300k developers' on the homepage and '100K+ Neo4j experts' in the nav sub-menu will notice the inconsistency.

Copy

Hero Stat '80+ Fortune 100 Customers' — Conflicts With Other Public Claims

Score

42

Severity

High

Finding

The homepage hero stats bar shows: '300k Developers Building', '80+ Fortune 100 Customers', '170+ Partner Ecosystem'. However Neo4j's own press releases and TechCrunch coverage (November 2024) state '84% of the Fortune 100' — which at 100 Fortune 100 companies means 84 customers, not 80+. An earlier Series F press release (2021) stated '75% of the Fortune 100.' The homepage claim of '80+' sits between these two figures. The discrepancy is likely a rounding artefact — '84% of Fortune 100' rounds to 84, which is both '80+' and lower than '84%' — but the headline stat and the press release stat use different framing that creates ambiguity. More importantly, the 84% figure is significantly more impressive-sounding than '80+' and should replace it.

Recommendation

Update the hero stat to match the most recent verified claim: '84% of the Fortune 100' or 'Trusted by 84 Fortune 100 companies.' The '84%' framing is more persuasive than '80+' and aligns with the TechCrunch coverage that is actively indexed. While updating stats, also verify the '300k developers' figure — the GraphAcademy nav link references '100K+ Neo4j experts' for certified developers specifically, which conflicts with the '300k developers building' hero claim. A CTO evaluating Neo4j who sees '300k developers' on the homepage and '100K+ Neo4j experts' in the nav sub-menu will notice the inconsistency.

Copy

Hero Stat '80+ Fortune 100 Customers' — Conflicts With Other Public Claims

Score

42

Severity

High

Finding

The homepage hero stats bar shows: '300k Developers Building', '80+ Fortune 100 Customers', '170+ Partner Ecosystem'. However Neo4j's own press releases and TechCrunch coverage (November 2024) state '84% of the Fortune 100' — which at 100 Fortune 100 companies means 84 customers, not 80+. An earlier Series F press release (2021) stated '75% of the Fortune 100.' The homepage claim of '80+' sits between these two figures. The discrepancy is likely a rounding artefact — '84% of Fortune 100' rounds to 84, which is both '80+' and lower than '84%' — but the headline stat and the press release stat use different framing that creates ambiguity. More importantly, the 84% figure is significantly more impressive-sounding than '80+' and should replace it.

Recommendation

Update the hero stat to match the most recent verified claim: '84% of the Fortune 100' or 'Trusted by 84 Fortune 100 companies.' The '84%' framing is more persuasive than '80+' and aligns with the TechCrunch coverage that is actively indexed. While updating stats, also verify the '300k developers' figure — the GraphAcademy nav link references '100K+ Neo4j experts' for certified developers specifically, which conflicts with the '300k developers building' hero claim. A CTO evaluating Neo4j who sees '300k developers' on the homepage and '100K+ Neo4j experts' in the nav sub-menu will notice the inconsistency.

Navigation

Nav Product Menu Labels 'Aura Agent' as 'A single console to manage all your DB instances' — Wrong Description

Score

44

Severity

High

Finding

In the Products mega-menu under 'GRAPH AI', Neo4j Aura Agent is described as: 'A single console to manage all your DB instances.' This is the description for Fleet Manager — the actual product positioned as 'A single control plane to manage all your DB instances' in the GRAPH TOOLS section of the same menu. The Aura Agent is Neo4j's AI agent-building product: it allows customers to 'build and deploy agents on their own data in minutes' and is one of the flagship October 2025 product launches. The copy-paste error swaps the descriptions of two completely different products in the primary navigation. Any developer clicking 'Aura Agent' expecting the AI agent product and reading 'manage all your DB instances' will be confused before they even land on the page.

Recommendation

Fix the Aura Agent nav description immediately: it should read something like 'Build and deploy AI agents on your graph data in minutes' — matching the October 2025 product launch positioning. Audit the entire Products mega-menu for description accuracy: Fleet Manager should retain its 'single control plane' copy; Aura Agent needs its own distinct AI agent framing. This is a live production copy error in the primary navigation of neo4j.com — the most-visited section of the site for product evaluators. Establish a nav copy review checklist that fires whenever a new product is added to the menu.

Navigation

Nav Product Menu Labels 'Aura Agent' as 'A single console to manage all your DB instances' — Wrong Description

Score

44

Severity

High

Finding

In the Products mega-menu under 'GRAPH AI', Neo4j Aura Agent is described as: 'A single console to manage all your DB instances.' This is the description for Fleet Manager — the actual product positioned as 'A single control plane to manage all your DB instances' in the GRAPH TOOLS section of the same menu. The Aura Agent is Neo4j's AI agent-building product: it allows customers to 'build and deploy agents on their own data in minutes' and is one of the flagship October 2025 product launches. The copy-paste error swaps the descriptions of two completely different products in the primary navigation. Any developer clicking 'Aura Agent' expecting the AI agent product and reading 'manage all your DB instances' will be confused before they even land on the page.

Recommendation

Fix the Aura Agent nav description immediately: it should read something like 'Build and deploy AI agents on your graph data in minutes' — matching the October 2025 product launch positioning. Audit the entire Products mega-menu for description accuracy: Fleet Manager should retain its 'single control plane' copy; Aura Agent needs its own distinct AI agent framing. This is a live production copy error in the primary navigation of neo4j.com — the most-visited section of the site for product evaluators. Establish a nav copy review checklist that fires whenever a new product is added to the menu.

Navigation

Nav Product Menu Labels 'Aura Agent' as 'A single console to manage all your DB instances' — Wrong Description

Score

44

Severity

High

Finding

In the Products mega-menu under 'GRAPH AI', Neo4j Aura Agent is described as: 'A single console to manage all your DB instances.' This is the description for Fleet Manager — the actual product positioned as 'A single control plane to manage all your DB instances' in the GRAPH TOOLS section of the same menu. The Aura Agent is Neo4j's AI agent-building product: it allows customers to 'build and deploy agents on their own data in minutes' and is one of the flagship October 2025 product launches. The copy-paste error swaps the descriptions of two completely different products in the primary navigation. Any developer clicking 'Aura Agent' expecting the AI agent product and reading 'manage all your DB instances' will be confused before they even land on the page.

Recommendation

Fix the Aura Agent nav description immediately: it should read something like 'Build and deploy AI agents on your graph data in minutes' — matching the October 2025 product launch positioning. Audit the entire Products mega-menu for description accuracy: Fleet Manager should retain its 'single control plane' copy; Aura Agent needs its own distinct AI agent framing. This is a live production copy error in the primary navigation of neo4j.com — the most-visited section of the site for product evaluators. Establish a nav copy review checklist that fires whenever a new product is added to the menu.

SEO

Footer Contains Link to 'NODES 2025' — Stale Event in March 2026

Score

47

Severity

Medium

Finding

The footer Learn section contains a link to 'NODES' pointing to neo4j.com/nodes-2025 — a URL clearly dated to 2025. In March 2026 the same footer also advertises 'NODES AI 2026' as a featured event in the nav Learn section, creating a direct contradiction: the footer links to NODES 2025 while the nav promotes NODES AI 2026. A visitor who clicks the footer NODES link reaches a 2025 event page rather than the current 2026 event. This is the exact stale-event-URL pattern flagged in the Spryker audit (Autumn 2025 product update still in nav in March 2026) and the Quentic audit (EXCITE 2025 still live). The Events section also shows three GraphSummit / Workshop / NODES events on March 17 2026 — today — suggesting the events section is current but the footer has not been updated.

Recommendation

Update the footer NODES link from /nodes-2025 to /nodes-ai (or the 2026 equivalent URL) and remove the dated URL slug. Establish a footer maintenance rule: any link containing a year-specific URL slug (/nodes-2025, /graphsummit-2025) must be updated within 30 days of the referenced event passing. The discrepancy between the nav (promoting NODES AI 2026) and the footer (linking to nodes-2025) creates a two-truths problem for the same visitor in the same session.

SEO

Footer Contains Link to 'NODES 2025' — Stale Event in March 2026

Score

47

Severity

Medium

Finding

The footer Learn section contains a link to 'NODES' pointing to neo4j.com/nodes-2025 — a URL clearly dated to 2025. In March 2026 the same footer also advertises 'NODES AI 2026' as a featured event in the nav Learn section, creating a direct contradiction: the footer links to NODES 2025 while the nav promotes NODES AI 2026. A visitor who clicks the footer NODES link reaches a 2025 event page rather than the current 2026 event. This is the exact stale-event-URL pattern flagged in the Spryker audit (Autumn 2025 product update still in nav in March 2026) and the Quentic audit (EXCITE 2025 still live). The Events section also shows three GraphSummit / Workshop / NODES events on March 17 2026 — today — suggesting the events section is current but the footer has not been updated.

Recommendation

Update the footer NODES link from /nodes-2025 to /nodes-ai (or the 2026 equivalent URL) and remove the dated URL slug. Establish a footer maintenance rule: any link containing a year-specific URL slug (/nodes-2025, /graphsummit-2025) must be updated within 30 days of the referenced event passing. The discrepancy between the nav (promoting NODES AI 2026) and the footer (linking to nodes-2025) creates a two-truths problem for the same visitor in the same session.

SEO

Footer Contains Link to 'NODES 2025' — Stale Event in March 2026

Score

47

Severity

Medium

Finding

The footer Learn section contains a link to 'NODES' pointing to neo4j.com/nodes-2025 — a URL clearly dated to 2025. In March 2026 the same footer also advertises 'NODES AI 2026' as a featured event in the nav Learn section, creating a direct contradiction: the footer links to NODES 2025 while the nav promotes NODES AI 2026. A visitor who clicks the footer NODES link reaches a 2025 event page rather than the current 2026 event. This is the exact stale-event-URL pattern flagged in the Spryker audit (Autumn 2025 product update still in nav in March 2026) and the Quentic audit (EXCITE 2025 still live). The Events section also shows three GraphSummit / Workshop / NODES events on March 17 2026 — today — suggesting the events section is current but the footer has not been updated.

Recommendation

Update the footer NODES link from /nodes-2025 to /nodes-ai (or the 2026 equivalent URL) and remove the dated URL slug. Establish a footer maintenance rule: any link containing a year-specific URL slug (/nodes-2025, /graphsummit-2025) must be updated within 30 days of the referenced event passing. The discrepancy between the nav (promoting NODES AI 2026) and the footer (linking to nodes-2025) creates a two-truths problem for the same visitor in the same session.

Copy

$200M Revenue and $100M GenAI Investment — Neither Mentioned on Homepage

Score

45

Severity

Medium

Finding

Neo4j publicly stated in November 2024 that revenue had surpassed $200M — double from three years prior. In October 2025 the company announced a $100M investment in GenAI-native startups over 12 months, and launched Aura Agent plus an MCP server. Neither figure appears anywhere on the homepage. The $200M revenue milestone is a rare self-reported growth proof point for a private company that signals financial health to enterprise procurement teams. The $100M GenAI startup investment is a category-defining ecosystem move that positions Neo4j as the default graph layer for the AI agent wave. Both belong on the homepage — either in the hero trust strip, a dedicated ecosystem section, or a news banner.

Recommendation

Add a homepage trust strip below the hero containing the three most important proof points: '84% of Fortune 100 · $200M+ revenue · $100M invested in GenAI startups.' Alternatively, add a news banner for the $100M GenAI investment (confirmed October 2025) as the homepage has a 'New Agentic Innovation' banner already linking to the blog — elevate this to include the $100M figure explicitly. The MCP Server availability (confirmed 'by end of year' per October 2025 coverage) should also be surfaced if now GA, as it directly competes with PandaDoc and other MCP-native product announcements in this audit series.

Copy

$200M Revenue and $100M GenAI Investment — Neither Mentioned on Homepage

Score

45

Severity

Medium

Finding

Neo4j publicly stated in November 2024 that revenue had surpassed $200M — double from three years prior. In October 2025 the company announced a $100M investment in GenAI-native startups over 12 months, and launched Aura Agent plus an MCP server. Neither figure appears anywhere on the homepage. The $200M revenue milestone is a rare self-reported growth proof point for a private company that signals financial health to enterprise procurement teams. The $100M GenAI startup investment is a category-defining ecosystem move that positions Neo4j as the default graph layer for the AI agent wave. Both belong on the homepage — either in the hero trust strip, a dedicated ecosystem section, or a news banner.

Recommendation

Add a homepage trust strip below the hero containing the three most important proof points: '84% of Fortune 100 · $200M+ revenue · $100M invested in GenAI startups.' Alternatively, add a news banner for the $100M GenAI investment (confirmed October 2025) as the homepage has a 'New Agentic Innovation' banner already linking to the blog — elevate this to include the $100M figure explicitly. The MCP Server availability (confirmed 'by end of year' per October 2025 coverage) should also be surfaced if now GA, as it directly competes with PandaDoc and other MCP-native product announcements in this audit series.

Copy

$200M Revenue and $100M GenAI Investment — Neither Mentioned on Homepage

Score

45

Severity

Medium

Finding

Neo4j publicly stated in November 2024 that revenue had surpassed $200M — double from three years prior. In October 2025 the company announced a $100M investment in GenAI-native startups over 12 months, and launched Aura Agent plus an MCP server. Neither figure appears anywhere on the homepage. The $200M revenue milestone is a rare self-reported growth proof point for a private company that signals financial health to enterprise procurement teams. The $100M GenAI startup investment is a category-defining ecosystem move that positions Neo4j as the default graph layer for the AI agent wave. Both belong on the homepage — either in the hero trust strip, a dedicated ecosystem section, or a news banner.

Recommendation

Add a homepage trust strip below the hero containing the three most important proof points: '84% of Fortune 100 · $200M+ revenue · $100M invested in GenAI startups.' Alternatively, add a news banner for the $100M GenAI investment (confirmed October 2025) as the homepage has a 'New Agentic Innovation' banner already linking to the blog — elevate this to include the $100M figure explicitly. The MCP Server availability (confirmed 'by end of year' per October 2025 coverage) should also be surfaced if now GA, as it directly competes with PandaDoc and other MCP-native product announcements in this audit series.

Performance

Homepage Loads Animated Background Images as Six Sets of Duplicate PNG Rows

Score

40

Severity

Medium

Finding

The hero background uses a tiled-row animation of graph nodes — implemented as six distinct PNG images (hp-row-1.png, hp-row-2.png, hp-row-3-1.png, row5.png, row6.png, row7.png), each loaded four times for a total of 24 background image requests, plus an additional set of the same images loaded a second time for the lower section of the background (another 24 requests — 48 total background image requests for a decorative animation). This is in addition to the 88 logo carousel image requests. The combined decorative image overhead of 136 <img> elements for non-content visuals on the homepage is a significant Largest Contentful Paint liability. On mobile connections this translates directly to slow first render times for a developer-focused product where page speed is a credibility signal.

Recommendation

Replace the tiled PNG background with a CSS-based SVG animation or a single repeating-background CSS pattern. The visual effect (scrolling graph node tiles) can be achieved with a single SVG tile and CSS background-repeat: repeat combined with a CSS keyframe animation — zero additional HTTP requests. If the PNG approach must be retained, serve a single copy of each row image and use CSS animation transforms to create the tiling effect rather than duplicating DOM elements. Run Lighthouse on the current homepage; the combined 136 decorative image requests will likely show as an LCP and TBT failure. For a product marketed to engineers, a slow homepage is a brand contradiction.

Performance

Homepage Loads Animated Background Images as Six Sets of Duplicate PNG Rows

Score

40

Severity

Medium

Finding

The hero background uses a tiled-row animation of graph nodes — implemented as six distinct PNG images (hp-row-1.png, hp-row-2.png, hp-row-3-1.png, row5.png, row6.png, row7.png), each loaded four times for a total of 24 background image requests, plus an additional set of the same images loaded a second time for the lower section of the background (another 24 requests — 48 total background image requests for a decorative animation). This is in addition to the 88 logo carousel image requests. The combined decorative image overhead of 136 <img> elements for non-content visuals on the homepage is a significant Largest Contentful Paint liability. On mobile connections this translates directly to slow first render times for a developer-focused product where page speed is a credibility signal.

Recommendation

Replace the tiled PNG background with a CSS-based SVG animation or a single repeating-background CSS pattern. The visual effect (scrolling graph node tiles) can be achieved with a single SVG tile and CSS background-repeat: repeat combined with a CSS keyframe animation — zero additional HTTP requests. If the PNG approach must be retained, serve a single copy of each row image and use CSS animation transforms to create the tiling effect rather than duplicating DOM elements. Run Lighthouse on the current homepage; the combined 136 decorative image requests will likely show as an LCP and TBT failure. For a product marketed to engineers, a slow homepage is a brand contradiction.

Performance

Homepage Loads Animated Background Images as Six Sets of Duplicate PNG Rows

Score

40

Severity

Medium

Finding

The hero background uses a tiled-row animation of graph nodes — implemented as six distinct PNG images (hp-row-1.png, hp-row-2.png, hp-row-3-1.png, row5.png, row6.png, row7.png), each loaded four times for a total of 24 background image requests, plus an additional set of the same images loaded a second time for the lower section of the background (another 24 requests — 48 total background image requests for a decorative animation). This is in addition to the 88 logo carousel image requests. The combined decorative image overhead of 136 <img> elements for non-content visuals on the homepage is a significant Largest Contentful Paint liability. On mobile connections this translates directly to slow first render times for a developer-focused product where page speed is a credibility signal.

Recommendation

Replace the tiled PNG background with a CSS-based SVG animation or a single repeating-background CSS pattern. The visual effect (scrolling graph node tiles) can be achieved with a single SVG tile and CSS background-repeat: repeat combined with a CSS keyframe animation — zero additional HTTP requests. If the PNG approach must be retained, serve a single copy of each row image and use CSS animation transforms to create the tiling effect rather than duplicating DOM elements. Run Lighthouse on the current homepage; the combined 136 decorative image requests will likely show as an LCP and TBT failure. For a product marketed to engineers, a slow homepage is a brand contradiction.

Copy

Aura Agent Product Copy Misdescription — 'Knowledge Graph' Section Uses Generic Stock Photo Description

Score

52

Severity

Low

Finding

The 'Transform Your Data Into Knowledge' section describes the knowledge graph use case with the sub-heading: 'Connect and organize your data with a knowledge graph to see the bigger picture.' This is accurate but generic — the same copy could describe any graph database vendor or even a relational database with JOIN tables. There is no specific mention of what differentiates Neo4j's knowledge graph from competitors (TigerGraph, Amazon Neptune, Memgraph): native graph storage vs index-free adjacency, the Cypher query language, the 44% market share, the 18-year track record, the Forrester 'essential' quote. The section's image is also a generic AI illustration (genai-graphic-600x331.webp) rather than an actual Neo4j knowledge graph visualisation — the kind of screenshot that would immediately demonstrate the product's visual differentiation.

Recommendation

Replace the generic 'see the bigger picture' copy with differentiated claims: 'The only graph database with native graph storage, purpose-built for connected data — not retrofitted from a relational engine.' Include a specific comparison point: 'Query relationships 100x faster than SQL JOINs at scale.' Replace the AI stock illustration with an actual Neo4j Browser or Bloom screenshot showing a real knowledge graph — this is the most visually distinctive aspect of the product and the homepage does not show it anywhere above the fold. The competitor comparison table (vs Amazon Neptune, vs TigerGraph) that likely exists somewhere in Neo4j's content library should be linked from this section.

Copy

Aura Agent Product Copy Misdescription — 'Knowledge Graph' Section Uses Generic Stock Photo Description

Score

52

Severity

Low

Finding

The 'Transform Your Data Into Knowledge' section describes the knowledge graph use case with the sub-heading: 'Connect and organize your data with a knowledge graph to see the bigger picture.' This is accurate but generic — the same copy could describe any graph database vendor or even a relational database with JOIN tables. There is no specific mention of what differentiates Neo4j's knowledge graph from competitors (TigerGraph, Amazon Neptune, Memgraph): native graph storage vs index-free adjacency, the Cypher query language, the 44% market share, the 18-year track record, the Forrester 'essential' quote. The section's image is also a generic AI illustration (genai-graphic-600x331.webp) rather than an actual Neo4j knowledge graph visualisation — the kind of screenshot that would immediately demonstrate the product's visual differentiation.

Recommendation

Replace the generic 'see the bigger picture' copy with differentiated claims: 'The only graph database with native graph storage, purpose-built for connected data — not retrofitted from a relational engine.' Include a specific comparison point: 'Query relationships 100x faster than SQL JOINs at scale.' Replace the AI stock illustration with an actual Neo4j Browser or Bloom screenshot showing a real knowledge graph — this is the most visually distinctive aspect of the product and the homepage does not show it anywhere above the fold. The competitor comparison table (vs Amazon Neptune, vs TigerGraph) that likely exists somewhere in Neo4j's content library should be linked from this section.

Copy

Aura Agent Product Copy Misdescription — 'Knowledge Graph' Section Uses Generic Stock Photo Description

Score

52

Severity

Low

Finding

The 'Transform Your Data Into Knowledge' section describes the knowledge graph use case with the sub-heading: 'Connect and organize your data with a knowledge graph to see the bigger picture.' This is accurate but generic — the same copy could describe any graph database vendor or even a relational database with JOIN tables. There is no specific mention of what differentiates Neo4j's knowledge graph from competitors (TigerGraph, Amazon Neptune, Memgraph): native graph storage vs index-free adjacency, the Cypher query language, the 44% market share, the 18-year track record, the Forrester 'essential' quote. The section's image is also a generic AI illustration (genai-graphic-600x331.webp) rather than an actual Neo4j knowledge graph visualisation — the kind of screenshot that would immediately demonstrate the product's visual differentiation.

Recommendation

Replace the generic 'see the bigger picture' copy with differentiated claims: 'The only graph database with native graph storage, purpose-built for connected data — not retrofitted from a relational engine.' Include a specific comparison point: 'Query relationships 100x faster than SQL JOINs at scale.' Replace the AI stock illustration with an actual Neo4j Browser or Bloom screenshot showing a real knowledge graph — this is the most visually distinctive aspect of the product and the homepage does not show it anywhere above the fold. The competitor comparison table (vs Amazon Neptune, vs TigerGraph) that likely exists somewhere in Neo4j's content library should be linked from this section.

Navigation

JavaScript Disabled Warning Shown as Body Content — Visible in Rendered HTML

Score

48

Severity

Medium

Finding

The homepage source contains a visible warning block: '### Warning: JavaScript is disabled on your browser. Parts of Neo4j.com will not work properly.' This is rendered as a markdown-formatted heading within the page HTML — appearing as '###' heading syntax in the raw source — suggesting it is a component-level noscript message that is being parsed as Markdown text rather than rendered as styled HTML. While this warning is likely hidden from JS-enabled browsers via CSS, its presence as a raw '### Warning:' markdown string in the HTML source means: (1) it is indexed by Google as page body text, adding 'Warning: JavaScript is disabled' to the homepage's indexed content; (2) it would be read by screen readers; (3) it appears prominently in the fetched markdown representation of the page.

Recommendation

Replace the noscript text block with a proper HTML <noscript> tag that renders styled HTML — not Markdown-formatted text: `<noscript><div class='js-warning'>For the best Neo4j experience, please enable JavaScript.</div></noscript>`. Ensure the noscript block is excluded from Google's indexable content by confirming it appears inside a <noscript> tag (which crawlers typically skip) rather than as a conditionally-hidden visible DOM element. Audit the page's Googlebot rendering (via Google Search Console's URL Inspection > View Crawled Page) to confirm the warning is not being indexed as homepage body text. The '### Warning' markdown string in rendered HTML source is unusual and suggests a component rendering pipeline issue.

Navigation

JavaScript Disabled Warning Shown as Body Content — Visible in Rendered HTML

Score

48

Severity

Medium

Finding

The homepage source contains a visible warning block: '### Warning: JavaScript is disabled on your browser. Parts of Neo4j.com will not work properly.' This is rendered as a markdown-formatted heading within the page HTML — appearing as '###' heading syntax in the raw source — suggesting it is a component-level noscript message that is being parsed as Markdown text rather than rendered as styled HTML. While this warning is likely hidden from JS-enabled browsers via CSS, its presence as a raw '### Warning:' markdown string in the HTML source means: (1) it is indexed by Google as page body text, adding 'Warning: JavaScript is disabled' to the homepage's indexed content; (2) it would be read by screen readers; (3) it appears prominently in the fetched markdown representation of the page.

Recommendation

Replace the noscript text block with a proper HTML <noscript> tag that renders styled HTML — not Markdown-formatted text: `<noscript><div class='js-warning'>For the best Neo4j experience, please enable JavaScript.</div></noscript>`. Ensure the noscript block is excluded from Google's indexable content by confirming it appears inside a <noscript> tag (which crawlers typically skip) rather than as a conditionally-hidden visible DOM element. Audit the page's Googlebot rendering (via Google Search Console's URL Inspection > View Crawled Page) to confirm the warning is not being indexed as homepage body text. The '### Warning' markdown string in rendered HTML source is unusual and suggests a component rendering pipeline issue.

Navigation

JavaScript Disabled Warning Shown as Body Content — Visible in Rendered HTML

Score

48

Severity

Medium

Finding

The homepage source contains a visible warning block: '### Warning: JavaScript is disabled on your browser. Parts of Neo4j.com will not work properly.' This is rendered as a markdown-formatted heading within the page HTML — appearing as '###' heading syntax in the raw source — suggesting it is a component-level noscript message that is being parsed as Markdown text rather than rendered as styled HTML. While this warning is likely hidden from JS-enabled browsers via CSS, its presence as a raw '### Warning:' markdown string in the HTML source means: (1) it is indexed by Google as page body text, adding 'Warning: JavaScript is disabled' to the homepage's indexed content; (2) it would be read by screen readers; (3) it appears prominently in the fetched markdown representation of the page.

Recommendation

Replace the noscript text block with a proper HTML <noscript> tag that renders styled HTML — not Markdown-formatted text: `<noscript><div class='js-warning'>For the best Neo4j experience, please enable JavaScript.</div></noscript>`. Ensure the noscript block is excluded from Google's indexable content by confirming it appears inside a <noscript> tag (which crawlers typically skip) rather than as a conditionally-hidden visible DOM element. Audit the page's Googlebot rendering (via Google Search Console's URL Inspection > View Crawled Page) to confirm the warning is not being indexed as homepage body text. The '### Warning' markdown string in rendered HTML source is unusual and suggests a component rendering pipeline issue.

Copy

Testimonial Carousel Slides Labelled 'First slide details / Current Slide / Second slide details' in DOM

Score

36

Severity

High

Finding

The testimonial carousel section contains visible accessible text labels in the HTML: 'First slide details.', 'Current Slide', 'Second slide details.', 'Third slide details.', 'Fourth slide details.' (appearing twice). These are ARIA carousel slide descriptions — placeholder accessibility labels that should contain actual content descriptions (e.g. 'Testimonial from Natalie Romanov, Merck Group about knowledge graphs') but instead contain generic boilerplate. Screen readers will announce 'First slide details' as the slide description to visually impaired users navigating the testimonial carousel. This is a standard accessible carousel implementation that was deployed with placeholder ARIA labels rather than content-specific descriptions — the equivalent of alt text reading 'image' instead of describing the image.

Recommendation

Update all ARIA slide description labels in the testimonial carousel to describe their actual content: aria-label='Merck Group testimonial: Agentic AI and knowledge graph quote from Natalie Romanov' for slide 1, etc. This is both an accessibility fix and an SEO fix — rich ARIA labels help search engines understand carousel content. Run the homepage through an automated accessibility checker (axe, Lighthouse Accessibility) to identify all placeholder ARIA labels, alt text issues, and other a11y gaps. For a product used by Fortune 100 engineering teams, a homepage with accessibility failures is a credibility problem that enterprise procurement accessibility reviews will flag.

Copy

Testimonial Carousel Slides Labelled 'First slide details / Current Slide / Second slide details' in DOM

Score

36

Severity

High

Finding

The testimonial carousel section contains visible accessible text labels in the HTML: 'First slide details.', 'Current Slide', 'Second slide details.', 'Third slide details.', 'Fourth slide details.' (appearing twice). These are ARIA carousel slide descriptions — placeholder accessibility labels that should contain actual content descriptions (e.g. 'Testimonial from Natalie Romanov, Merck Group about knowledge graphs') but instead contain generic boilerplate. Screen readers will announce 'First slide details' as the slide description to visually impaired users navigating the testimonial carousel. This is a standard accessible carousel implementation that was deployed with placeholder ARIA labels rather than content-specific descriptions — the equivalent of alt text reading 'image' instead of describing the image.

Recommendation

Update all ARIA slide description labels in the testimonial carousel to describe their actual content: aria-label='Merck Group testimonial: Agentic AI and knowledge graph quote from Natalie Romanov' for slide 1, etc. This is both an accessibility fix and an SEO fix — rich ARIA labels help search engines understand carousel content. Run the homepage through an automated accessibility checker (axe, Lighthouse Accessibility) to identify all placeholder ARIA labels, alt text issues, and other a11y gaps. For a product used by Fortune 100 engineering teams, a homepage with accessibility failures is a credibility problem that enterprise procurement accessibility reviews will flag.

Copy

Testimonial Carousel Slides Labelled 'First slide details / Current Slide / Second slide details' in DOM

Score

36

Severity

High

Finding

The testimonial carousel section contains visible accessible text labels in the HTML: 'First slide details.', 'Current Slide', 'Second slide details.', 'Third slide details.', 'Fourth slide details.' (appearing twice). These are ARIA carousel slide descriptions — placeholder accessibility labels that should contain actual content descriptions (e.g. 'Testimonial from Natalie Romanov, Merck Group about knowledge graphs') but instead contain generic boilerplate. Screen readers will announce 'First slide details' as the slide description to visually impaired users navigating the testimonial carousel. This is a standard accessible carousel implementation that was deployed with placeholder ARIA labels rather than content-specific descriptions — the equivalent of alt text reading 'image' instead of describing the image.

Recommendation

Update all ARIA slide description labels in the testimonial carousel to describe their actual content: aria-label='Merck Group testimonial: Agentic AI and knowledge graph quote from Natalie Romanov' for slide 1, etc. This is both an accessibility fix and an SEO fix — rich ARIA labels help search engines understand carousel content. Run the homepage through an automated accessibility checker (axe, Lighthouse Accessibility) to identify all placeholder ARIA labels, alt text issues, and other a11y gaps. For a product used by Fortune 100 engineering teams, a homepage with accessibility failures is a credibility problem that enterprise procurement accessibility reviews will flag.

Frequently asked

What kind of companies do you work with?

We work with ambitious tech companies typically Series A and B at the moment where the brand and website haven't kept pace with the business.

You've found product-market fit. Now you need to look the part, communicate clearly, and move fast enough to stay ahead.

That's the problem we're built for.

What does a typical project look like?
We've had bad experiences with agencies before. What's different?
Why Framer over other platforms?
How do we get started?
How does pricing work?

Recent work

V7 Labs
Enzai
Utila
Centific
Buena
trawa
Portex Global
Othello AI
Echo
Pools
Contentcloud
Wilson

Perspectives & Insights

Blazing fast brands &

Blazing fast brands &

Blazing fast brands &

websites for startups

websites for startups

websites for startups