Fly.io pricing teardown
https://fly.io/docs/about/pricingFly.io's pricing page is a detailed infrastructure rate card — pure pay-as-you-go with no traditional SaaS tiers. It is highly transparent and technically thorough but is structured as documentation rather than a conversion-optimized pricing page, making it difficult for non-technical buyers or newcomers to quickly understand total cost or make a purchase decision.
Tier structure
Pay As You Go (no named tiers
usage-based across all resources)
Value metric
usage-based: per CPU/RAM second, per GB storage, per GB data transfer
How limits scale
Escalation logic
There are no escalating tiers; instead, customers compose costs from granular resource primitives (compute presets, volumes, network, Kubernetes, support add-ons). The only structured escalation is the optional support packages at $29, $199, and $2,500+/month.
Psychological anchors
- High-anchor enterprise support — Enterprise support starting at $2,500/month anchors the $29 and $199 support tiers as affordable by comparison.
- Reservation discount framing — Machine reservation blocks are presented as '40% discount' with upfront annual pricing, nudging higher-usage customers toward commitment.
- Free allowances for legacy plans — Legacy free resource allowances (3 VMs, 3GB storage, 100GB egress) are documented as discontinued, subtly reinforcing that the current model requires payment from the start.
- Region-based price variation — A region selector dynamically updates prices, making the cheapest regions (iad, ewr) visible as entry anchors while higher-cost regions (jnb) serve as implicit high anchors.
- Stopped machine pricing — Charging only rootfs ($0.15/GB/30 days) for stopped machines is framed as a cost-saving feature, reducing perceived risk of over-provisioning.
What this page is optimizing for
This page is optimized for developer self-serve transparency and trust-building among technical users who want to calculate exact costs before committing. The exhaustive rate tables, region selector, and documentation-style layout signal that Fly.io prioritizes informed adoption over conversion funnel optimization.
Red flags
- No recommended or highlighted plan — a first-time visitor has no guidance on where to start or what a typical bill looks like.
- Overwhelming volume of pricing tables (hundreds of CPU/RAM/region combinations) creates cognitive overload and makes cost estimation impractical without a calculator.
- No pricing calculator or cost estimator tool is linked or embedded, despite the complexity of the rate structure.
- No annual vs. monthly toggle for the overall plan — reservation blocks exist but are buried and require upfront cash commitment with no rollover.
- CTAs are absent from the pricing page itself — there is no 'Sign Up', 'Get Started', or 'Try Free' button visible on the page.
- Managed Postgres pricing is not shown inline — it redirects to a separate page, breaking the single-page cost estimation flow.
- GPU machines are marked deprecated with an August 1 end-of-life warning, which may alarm existing GPU users without a clear migration path presented.
- Legacy plan documentation (Hobby, Launch, Scale) adds noise and confusion for new visitors who encounter discontinued plan details.
- No trust signals such as customer logos, uptime SLA references, or testimonials are present on this pricing page.
- Support tiers are buried near the bottom with no feature comparison matrix, making it hard to justify the $199 vs $29 upgrade.
Best-practices scorecard
What works · 2
- Visible prices — All prices are fully public and granular, with per-second, per-hour, and per-month breakdowns across all regions.
- Value metric matches usage — Pay-per-second compute billing is well-aligned with infrastructure workloads that scale up and down dynamically.
Half measures · 2
- Annual discount offered — Reservation blocks offer a 40% annual discount but require upfront payment and are opt-in, not surfaced as a primary pricing option.
- FAQ or objection handling — Some billing edge cases are explained inline (e.g., stopped machines, snapshot billing, granular vs. non-granular transfer) but there is no dedicated FAQ section.
What's missing · 4
- Clear recommended tier — No tier is highlighted, badged as 'most popular', or recommended — there is no decision anchor for new customers.
- Tier differences are scannable — There are no tiers to compare — the page is a flat rate card with no side-by-side feature differentiation.
- Trust signals present — No customer logos, testimonials, uptime guarantees, or compliance badges appear on the pricing page.
- Clear CTAs per tier — No signup, trial, or contact CTAs are present on the pricing page itself — conversion is entirely dependent on navigation elsewhere.