Feature flags · Comparison
LaunchDarkly vs Statsig vs ConfigCat: Statsig leads at 8.7; 50,000 users from $0 to $416.50
We read the pricing, security and docs pages of three feature flag services on the same day and scored them on the same six indicators. How each one bills, what 50,000 monthly users cost, and who should pick which.
On this page8 sections
On six indicators read from each vendor's public pages, Statsig scores 8.7, LaunchDarkly 6.5 and ConfigCat 6.0 out of 10. For our reference scenario, a team of 10 with one web app of 50,000 monthly active users, one backend service, 2 environments and 30 flags, Statsig's free plan covers it at $0 if the flags create no metered events, ConfigCat's Pro plan costs $110 a month billed annually, and LaunchDarkly's Foundation plan comes to $416.50 billed yearly by our arithmetic. Statsig leads four of six items, with experiments and unlimited flag checks on its free plan. ConfigCat leads on team controls: SAML SSO and SCIM on every plan.
The three bill in different units, so the scenario figures rest on stated assumptions. The scores measure what the public pages commit to, not targeting depth, SDK quality or how a product feels to use, and we did not sign up for any of the three.
At a glance
| LaunchDarkly | Statsig | ConfigCat | |
|---|---|---|---|
| What you pay for | Flags, experiments and observability on one platform | Flags, experiments and product analytics | Feature flags and remote config |
| Billing units | Client-side MAU and service connections | Metered events (not flag checks) | Flat plans sized by downloads and traffic |
| Free plan | 1,000 client-side MAU, 5 service connections | 2M metered events/mo, unlimited flag checks | 10 flags, 2 environments, 5M downloads/mo |
| Reference scenario (50,000 MAU) | $416.50 (Foundation, yearly), computed | $0 (Developer), if no metered events | $110 (Pro, annual), if about 15M downloads |
| A/B testing | Every plan | Every plan, incl. CUPED | Not listed; via analytics tools |
| SOC 2 Type II / ISO 27001 | Both | SOC 2 Type II; ISO certificate is Amplitude's | ISO 27001:2022 |
| EU data region | EU instance, new Enterprise accounts | EU hosting on Enterprise | Main infrastructure in the EU |
| SAML SSO | Enterprise | Enterprise | Every plan, incl. Free |
| Seats | Unlimited | Unlimited | Unlimited |
| Score (public pages) | 6.5 Good | 8.7 Excellent | 6.0 Good |
What the public pages show
LaunchDarkly has a free Developer plan with unlimited seats and flags, 1,000 client-side MAU and 5 service connections a month, and a pay-as-you-go Foundation plan after a 14-day trial: $8.33 per 1,000 client-side MAU and $10 per service connection beyond the first five, billed yearly ($10 and $12 monthly). A/B testing, CUPED and sequential testing are marked on every plan. SSO, SCIM, audit logging and flag scheduling are marked Enterprise; the security page lists SOC 2 Type II, ISO 27001 and FedRAMP.
Statsig bills metered events, the events that power analytics: exposures for experiments and for flags with Metric Lifts on, plus logged events. Flag checks themselves are unlimited. Developer is free with 2 million events and unlimited seats; Pro is $150 a month per project with 5 million, then $0.05 per 1,000. SSO, EU hosting and a HIPAA BAA are on Enterprise. Statsig joined Amplitude on May 5, 2026, per its blog, and its trust link now opens Amplitude's trust center.
ConfigCat sells flat plans: Forever Free, Pro ($110 billed annually), Smart ($325), Enterprise ($900) and Dedicated ($4,500). SDKs cache a config file and evaluate flags locally, so only downloads of that file count; seats, MAU and flag reads are unlimited. 2FA, SAML SSO and SCIM are listed on every plan, the main infrastructure is in the EU, and the trust center lists ISO 27001:2022.
Strengths and things to know
LaunchDarkly. Strengths: experiments on every plan, unlimited seats, SOC 2 Type II, ISO 27001 and FedRAMP, open-source SDKs and OpenFeature providers. Things to know: client-side users drive the bill, SSO and audit logging are listed for Enterprise, the EU instance is for new Enterprise accounts, and a note on the pricing page about SSO on every plan does not match the plan table.
Statsig. Strengths: flag checks not billed, experiments, CUPED and product analytics on the free plan, rollout metrics and change logs on every plan. Things to know: SSO, EU hosting and the BAA are on Enterprise, each Pro subscription covers one project, the ISO certificate in its trust center is Amplitude's, and the product now sits within Amplitude.
ConfigCat. Strengths: SAML SSO and SCIM on every plan, no per-user billing, EU infrastructure, a JSON export and an on-premise option. Things to know: no built-in experiment analysis, a free plan of 10 flags and 2 environments, no SOC 2 or HIPAA listed, and a download count that depends on polling.
Pricing reality
The three bill on different models: LaunchDarkly by client-side users and server connections, Statsig by analytics events, ConfigCat by plan with download and traffic quotas. So we state one scenario and show the arithmetic. Reference scenario: a team of 10, one web app with 50,000 monthly active users plus one backend service, 2 environments and 30 flags, no experiments, at the lowest listed rates.
| Service | Plan | Monthly cost | Where the number comes from |
|---|---|---|---|
| Statsig | Developer | $0 | Free plan: flag checks are not metered, seats are unlimited |
| ConfigCat | Pro | $110 | Listed, billed annually: 100 flags, 3 environments, 25M downloads |
| LaunchDarkly | Foundation | $416.50 | Computed: 50 × $8.33 per 1,000 client-side MAU, billed yearly; 2 service connections within the 5 included |
Billed monthly, ConfigCat Pro is $120 and LaunchDarkly comes to 50 × $10 = $500. Seats cost nothing at any of the three, so the team of 10 adds nothing.
Each figure rests on something the pricing page cannot settle. Statsig's $0 assumes Metric Lifts stay off and no custom events are logged; if they were on, the free 2 million events would cover 40 deduplicated exposures per user a month (our arithmetic), and Pro would be $150 plus overage. ConfigCat's figure assumes its own docs example, 5 hours of use a month per user with 60-second polling, which gives 50,000 × 300 = 15 million downloads, within Pro's 25 million. LaunchDarkly's figure assumes every one of the 50,000 users is billed, because the page does not say whether Foundation includes free MAU, and that staging adds no new users; the backend's one instance per environment is 2 service connections.
How we scored them
Every product in this category is scored on the same six indicators, read from the vendor's own pricing, security or docs pages, with fixed formulas. The overall score is the plain mean. Our methodology page explains how these scores work.
| Indicator | LaunchDarkly | Statsig | ConfigCat |
|---|---|---|---|
| Pricing clarity | 6 | 10 | 8 |
| Free tier | 8 | 10 | 7 |
| Price for 50,000 monthly users | 2.4 | 10 | 6 |
| Experimentation and rollout safety | 9 | 10 | 1 |
| Compliance and data residency | 8 | 6 | 4 |
| Team and control | 5.5 | 6 | 10 |
| Overall | 6.5 | 8.7 | 6.0 |
LaunchDarkly loses most on the scenario price, because client-side users are billed, then on team and control, where SSO and audit logging are listed for Enterprise, and on pricing clarity: two units to estimate, an SSO note that does not match the plan table, and no word on whether Foundation includes free MAU. ConfigCat's experimentation score reflects a flag-only product that sends data to analytics tools rather than analysing tests itself. Statsig's free-tier and price scores rest on its flags creating no metered events.
The rubric rewards what a buyer can confirm before signing up. It has no item for targeting depth, SDK coverage or platform breadth, where LaunchDarkly also offers observability and AI configuration, and it cannot see how each product feels to use. Hands-on runs will add that later.
Who should choose which
- Choose Statsig if you want flags, A/B tests and product analytics in one tool and want to start free with the whole team. Look elsewhere if you need SSO or EU hosting without an Enterprise contract.
- Choose LaunchDarkly if you want experiments on every plan and a security review that asks for SOC 2 Type II, ISO 27001 or FedRAMP. Look elsewhere if many client-side users and a flat budget matter most.
- Choose ConfigCat if you only need flags and remote config at a flat price, with SAML SSO, SCIM and EU data on every plan. Look elsewhere if you want experiment analysis inside the flag tool.
Bottom line
From public pages alone, Statsig gives the most on a free plan: unlimited seats and flag checks, with experiments included. LaunchDarkly has the broadest platform and compliance list, and costs the most at 50,000 client-side users. ConfigCat's lower score reflects how this rubric counts experimentation and compliance; for a team that only needs flags with SSO and EU data, its flat plans are the simplest here.
How we checked
We read the vendors' pricing, security or trust, and documentation pages on Oct 15, 2026, including the billing switches on LaunchDarkly's and ConfigCat's pricing pages for monthly and annual prices. We did not sign up, install an SDK or evaluate a flag. Facts the pages did not state, such as whether LaunchDarkly's Foundation plan includes free MAU, are left out rather than guessed. Every figure is on the vendors' pages as of that date or, where marked, computed from them.
