feat(vs/n8n): highlight persistence and resilience as Kestra differentiators

[elliotgunn]

Positioning context

n8n wins on visual simplicity: it's the right tool for connecting SaaS apps and prototyping AI agent workflows fast. Kestra wins on production reliability: persistent execution state, step-level checkpointing, language-agnostic execution, and deployment anywhere. This page draws that line clearly rather than competing on n8n's turf.

Summary

Positioning and copy updates to the vs/n8n page based on feedback from vendor partner Stefan Fritz (SI space, SMB segment), who has been advocating Kestra over n8n to his clients and shared what wasn't landing.

Changes from initial pass

• Adds a Crash recovery row to the feature comparison grid: Kestra checkpoints every completed task; n8n can reconstruct which nodes crashed vs. completed but crashed workflows must restart from scratch
• Adds a new Platform strength card — "Built for business-critical reliability" — explaining the checkpoint/restart distinction with concrete ERP examples
• Adds a new FAQ — "What happens if a workflow crashes mid-execution?" — directly addressing the checkpoint scenario
• Updates Decision guide with a sixth bullet: "Business-critical workflows where losing execution state mid-run is not acceptable"
• Updates businessValue footer text to close on the "prototyping vs. production" framing

Changes from Stefan Fritz call follow-up

• Reframes the AI row from a feature-for-feature comparison (where n8n's tooling is more specific) to a production reliability frame: Kestra leads with persistent state, step-level retries, and container isolation; n8n is honestly described as a visual agent builder built for prototyping
• Renames "Air-gapped deployment" → "Deployment" with copy that conveys "run it wherever you want" (own servers, any cloud, air-gapped) rather than a narrow air-gap checkbox — addresses feedback that "self-hosted" doesn't translate clearly to international audiences
• Sharpens the license row to name what Apache 2.0 means in practice ("no usage restrictions") and contrast it with n8n's commercial restrictions
• Fixes plugin count in integration FAQ: 1,200+ → 1,300+
• Removes two em dashes from seeHowBody and the AI FAQ answer

Fact-check corrections

• Corrected all claims about n8n "holding state in memory" — n8n does persist execution data and has a recovery service that reconstructs crashed states from event logs. The accurate differentiator is step-level checkpointing: Kestra can resume from the last successful task; n8n must restart crashed workflows from scratch.

[github-actions]

☁️ Cloudflare Worker Preview Deployed!

🔗 https://ks-fix-n8n-vs-page-resilience-mes-docs.kestra-io.workers.dev
🔗 https://c2869356-docs.kestra-io.workers.dev

## 🔦 Lighthouse Benchmark

Tested: https://ks-fix-n8n-vs-page-resilience-mes-docs.kestra-io.workers.dev on 2026-06-11 14:57 UTC
No baseline available — scores will appear after the first merge to main

Scores (0–100, higher is better)

Page	Performance	Accessibility	Best Practices	SEO
Home	72	83	56	92
Pricing	97	91	56	100
Enterprise	89	82	56	100
Cloud	95	87	56	100
About Us	86	91	56	100
Docs Landing	92	88	56	92
Contribute to Kestra (simple docs)	98	88	56	92
Flow (full featured docs)	92	90	56	92
Blog Index	62	91	56	100
Blog Post (sample)	94	87	56	100
VS Page (sample)	97	88	56	100
Plugins Landing	91	81	56	92
Plugin Page (sample)	93	87	56	100
Plugin Debug Page (sample)	90	87	56	100
Plugin Debug Return Page (sample)	90	88	56	100
Blueprints Landing	92	80	56	92
Blueprint Audit Logs CSV Export	69	86	56	100

Core Web Vitals (lower is better)

Page	LCP	FCP	TBT	CLS	Speed Index
Home	1.51 s	0.66 s	454 ms	0.001	1.74 s
Pricing	1.20 s	0.63 s	23 ms	0.001	0.80 s
Enterprise	1.28 s	0.58 s	36 ms	0.171	0.85 s
Cloud	1.45 s	0.52 s	35 ms	0.000	0.98 s
About Us	2.59 s	0.64 s	38 ms	0.000	0.84 s
Docs Landing	1.02 s	0.51 s	204 ms	0.000	0.98 s
Contribute to Kestra (simple docs)	1.16 s	0.54 s	48 ms	0.000	0.82 s
Flow (full featured docs)	1.30 s	0.57 s	167 ms	0.000	1.14 s
Blog Index	5.51 s	1.21 s	116 ms	0.002	60.84 s
Blog Post (sample)	1.64 s	0.56 s	17 ms	0.000	0.83 s
VS Page (sample)	1.15 s	0.63 s	24 ms	0.064	0.83 s
Plugins Landing	1.07 s	0.62 s	34 ms	0.000	3.11 s
Plugin Page (sample)	0.97 s	0.59 s	99 ms	0.051	2.06 s
Plugin Debug Page (sample)	0.77 s	0.43 s	189 ms	0.001	2.18 s
Plugin Debug Return Page (sample)	1.09 s	0.64 s	145 ms	0.025	2.41 s
Blueprints Landing	1.40 s	0.79 s	20 ms	0.000	2.05 s
Blueprint Audit Logs CSV Export	1.10 s	0.56 s	161 ms	0.485	2.24 s

Legend

🟢 improved  ·  🔻 regressed  ·  (blank) no significant change
Score threshold: ±10 pts  ·  Metric threshold: ±30% of baseline