About LatencyCost.com
A vendor-neutral reference for quantifying the business cost of latency across ecommerce, API, financial trading, cloud gaming, and SEO. Independent. No products to sell. The published research, consolidated and cited.
Why this site exists
The cost of latency is one of the most-cited topics in web performance, but the data lives in scattered places. Amazon's "100ms = 1% of sales" figure comes from a 2006 Greg Linden presentation. Google's "0.5s = 20% traffic loss" comes from a 2006 Marissa Mayer talk. Walmart's "1s = 2% conversions" comes from a 2012 WalmartLabs engineering blog. The Akamai retail performance reports update sporadically. The Deloitte / Google "Milliseconds Make Millions" mobile retail study dropped in 2020 and was barely linked again.
CDN comparisons are mostly vendor-funded. Each CDN publishes performance benchmarks that show their network as fastest, using methodology that favours their architecture. Independent third-party measurements from Cloudflare Radar, Catchpoint, and ThousandEyes exist but are diffuse and require analyst-level interpretation to translate to dollar cost.
The industry-vertical models for the cost of latency are different in each surface. Ecommerce uses a conversion-rate elasticity. API microservices use a P99 chain model. HFT trading uses a per-millisecond revenue model anchored in academic papers. Cloud gaming uses a churn-cliff threshold model. There is no single place that has all four side-by-side with their methodology documented.
LatencyCost.com is that place. The calculator runs the ecommerce model client-side with no email gate. The case studies are consolidated with primary-source attribution. The CDN comparison uses pricing from vendor public pages and latency benchmarks from independent measurement networks. The vertical pages each carry the math for that surface. The /methodology page documents every source and every calculation.
Who builds this
LatencyCost.com is built and maintained by Oliver Wakefield-Smith at Digital Signet, an independent research consultancy focused on web performance economics, cloud infrastructure cost, and operator-platform cost intersections.
This site is part of a sister-site cluster of vendor-neutral cost references, each focused on one corner of the operator-platform cost surface. Cousin sites include EgressCost.com (cloud egress and data transfer), PlatformEngineeringCost.com (internal developer platforms), MonitoringCost.com (observability stack pricing), and CICDCost.com (CI/CD platform pricing).
Editorial position
LatencyCost.com is an independent resource. It is not a CDN reseller, not an APM affiliate, not a hosting-vendor lead-generation funnel, and not a performance-consultancy front door. No vendor pays for inclusion. No outbound link carries affiliate tracking parameters. The CDN comparison includes Cloudflare, AWS CloudFront, Fastly, and Bunny.net because they are the most relevant vendors to the surface, not because one of them is paying.
We do not publish "the best CDN" verdicts. Latency is workload-specific. A site with 80 percent traffic from North America has different optimal CDN choices than a site with 60 percent traffic from Southeast Asia. The CDN comparison page presents pricing and benchmark data; the reader decides.
What this site covers
Twelve content pages. Every page links to its primary sources at the bottom or inline.
Latency Cost Calculator
Three models: ecommerce, API, SEO
Case Studies
Amazon, Google, Walmart and 6 more
Ecommerce Page Speed
7% conversion loss per second
API / Microservice Latency
P99 compounding math
HFT Trading Latency
$100M per millisecond economics
Cloud Gaming Latency
100ms threshold + churn cost
SEO Impact
Core Web Vitals 2026 thresholds
User Experience
Perception thresholds and bounce data
Latency Budget Framework
SLO + error budget allocation
Optimisation Techniques
10 techniques ranked by impact
CDN Pricing Comparison
Cloudflare, CloudFront, Fastly
Methodology
Sources, scope, calculation framework
Editorial principles
Source pattern
Every cost anchor traces back to a vendor public pricing page, a published research paper, or an aggregator with a methodology page. No insider rumour, no NDA-protected number, no LLM hallucination.
No paid placements
We do not take payment for CDN listings, APM mentions, or vendor inclusion in the comparison table. Every vendor appears or does not appear based on relevance to the latency-cost surface.
No affiliate parameters
No referral codes on any outbound link. If we link to Cloudflare, Fastly, AWS, or any other vendor, the URL goes to their public homepage or pricing page without tracking parameters.
Monthly verification
Vendor pricing pages drift. The first business week of every month we re-verify the CDN price tables, Core Web Vitals thresholds, and case-study numbers against their primary sources.
Single-source freshness
A single LAST_VERIFIED_DATE constant drives the footer stamp, the Article schema dateModified, and the verification badge on every page. No drifting hardcoded dates.
Conservative cost models
Calculator outputs use the lower bound of published research ranges. We would rather understate the cost of latency by 20 percent than overstate it. The directional answer matters more than precision.
Methodology in brief
The ecommerce calculator uses a conversion-rate elasticity of approximately 7 percent loss per additional second of load time, aggregated from Akamai retail performance reports, the Google / Deloitte mobile retail study, and Walmart, COOK, AutoAnything, and Mobify case studies. The API page uses a P99 sequential chain model with worked examples. The trading page uses Moallemi-anchored HFT latency cost economics. The gaming page uses ParksAssociates churn-cliff data at the 100ms and 200ms thresholds.
The full sources table, in-scope and out-of-scope decisions, and per-calculator math live on the methodology page.
Contact and corrections
Spot a number that has drifted, a source we should add, or a vendor we should include in the comparison? Email [email protected]. We aim to acknowledge corrections within 5 business days.
LatencyCost.com is an independent resource. We are not affiliated with any CDN, APM, hosting, or network provider. Data is sourced from published research, vendor-neutral studies, and publicly available benchmarks. Actual revenue impact varies by site type, audience, and product category.