Koyeb pricing teardown
https://www.koyeb.com/pricingKoyeb runs a hybrid model: three subscription tiers (Pro $29/mo, Scale $299/mo, Enterprise custom) that bundle included compute credit plus platform features, layered on top of granular usage-based pricing for compute (per-second/hour GPU and CPU instances) and Postgres storage. The page is dense and technical, aimed at infrastructure buyers who want transparency into raw instance pricing, but the interplay between subscription tiers and metered usage adds real cognitive load despite a solid comparison table and FAQ.
Tier structure
Pro
$29/mo + compute
Scale
$299/mo + compute
Enterprise
Custom + compute
Value metric
flat-rate subscription tier + usage-based compute/storage billed per second
How limits scale
Escalation logic
Tiers escalate by included usage credit, user seats, concurrency limits, regions, and support/SLA tier, with compute and storage always billed separately on top regardless of plan.
Psychological anchors
- Enterprise as high anchor — Enterprise tier explicitly states 'starting at $1000/mo', anchoring the custom plan against the $29/$299 tiers.
- Popular/recommended tags on usage SKUs — 'Popular' badges appear on specific GPU instances (A100) and Postgres instance sizes (Medium), nudging selection within the usage-based catalog rather than on the subscription tiers themselves.
- Free entry point on Postgres — A '$0/mo Free 5h' Postgres instance acts as a no-cost trial anchor for the database product.
- Included credit framing — Each subscription tier bundles a dollar amount of compute credit ($10/$100/$500), softening the perception of paying twice (subscription + usage).
- Charm/precision pricing — Granular per-second and per-hour prices (e.g. $0.000208/sec) signal fine-grained fairness and technical credibility to an infra-buyer audience.
What this page is optimizing for
Self-serve conversion for developers/startups (transparent instance pricing, instant 'Get started' CTAs, calculator) combined with enterprise lead-gen for large accounts (Contact us, custom pricing, compliance features) — the page serves both motions rather than picking one.
Red flags
- No visible free trial or trial period for the core compute product, only a small free Postgres tier, raising the barrier to first use.
- Pricing complexity is high: three subscription tiers each still require separate usage-based billing for compute, bandwidth, and storage, making total cost hard to predict at a glance.
- No annual billing discount is mentioned anywhere on the page, missing a common conversion/retention lever.
- The long list of 20+ GPU instance types and multiple pricing toggles (Month/Hour/Second, GPU/Standard/Eco) could overwhelm users trying to quickly compare plans.
- Feature comparison table mixes plan-level and usage-level details, requiring significant scrolling and reading to differentiate Pro vs Scale vs Enterprise value.
Best-practices scorecard
What works · 5
- Visible prices — Pro ($29) and Scale ($299) prices are shown plainly; only Enterprise is 'Custom' with a stated starting point ($1000/mo).
- Value metric matches usage — Per-second/hour billing for compute and per-GB storage pricing directly aligns cost with actual infrastructure consumption.
- FAQ or objection handling — FAQ section covers billing mechanics, bandwidth costs, custom domains, and cancellation policy.
- Trust signals present — Logo strip ('Trusted by the most ambitious teams') includes MongoDB, Neon, Meilisearch, and others.
- Clear CTAs per tier — Each tier has a distinct CTA ('Get started' for Pro/Scale, 'Contact us' for Enterprise).
Half measures · 2
- Clear recommended tier — No 'Most Popular' badge on the three subscription tiers themselves, though 'Popular' tags appear on specific GPU/DB instance choices.
- Tier differences are scannable — A detailed comparison table exists but is very long, mixing dozens of rows across compute, networking, security, and support.
What's missing · 1
- Annual discount offered — No monthly/annual toggle or discount is visible anywhere on the page.