Probant vs. LaunchDarkly
Both tools compared point by point, based on LaunchDarkly’s public documentation, accessed November 11, 2026. Every statement about LaunchDarkly links to its source.
What LaunchDarkly offers.
LaunchDarkly, a pioneer of feature management, now presents itself as the runtime control layer for software development.1 Founded in 2014 in Oakland, it serves large enterprises and startups alike.23 Its engine is sequential frequentist by default, with Bayesian as an option.456
Probant is for sites that want A/B test verdicts that hold up, with a plan set before launch and guardrails monitored through rollout. It measures what each conversion is worth: the margin on an order, a qualified or signed lead, a paid subscription.
The differences, point by point.
On the left, what LaunchDarkly’s public documentation says, each sentence with its source. On the right, what Probant does. “Not mentioned” means nothing was found in that documentation, which does not prove a feature is missing.
| Criterion | LaunchDarkly | Probant |
|---|---|---|
| Statistical method | Two approaches: frequentist (default) and Bayesian. In frequentist mode, sequential analysis is the default option; fixed horizon (z-test) remains available.456 | Frequentist, with a sequential plan set before launch: target effect, duration, and checkpoints. The verdict fits in one sentence, with a 95% range. Where to check |
| Peeking | Sequential mode relies on the confidence sequences of Howard (2021) and Waudby-Smith et al. (2024), called GAVI, and allows a decision at any time. It is on by default in frequentist mode.78 | Decisions are only made at planned checkpoints, with O’Brien-Fleming boundaries: very strict at first, close to the usual threshold at the end. During a rollout, ranges are always valid (mSPRT), so you can check them at any time. Where to check |
| Traffic split (SRM) | In an experiment, SRM is flagged by a health check (“Health errors”) without blocking results. In a guarded rollout, an SRM triggers a rollback, even when automatic rollback is not turned on.91011 | The traffic split is checked with a chi-square test every time the report is read. If the split is off, the test moves to “blocked” status and no winner is declared. Where to check |
| Guardrails | Secondary metrics act as guardrails in experiments, with no documented automatic stop. Automatic stop and rollback exist for “guarded rollouts,” which do not run at the same time as an experiment on the same flag.121314 | Guardrail metrics are tracked with their range; a guardrail is only breached if the whole range crosses the limit. A breached guardrail blocks the test’s verdict, stops a running test on its own, and during a rollout moves visitors back to the original automatically. Where to check |
| Gradual rollouts | Guarded rollouts increase traffic in steps, monitor metrics, and can roll back automatically on regression. The documentation ties them to the Enterprise plan with the Guardian add-on; the current pricing page includes them in all plans (limited on Developer).151617 | Steps of 25% → 50% → 100% by default, 10% → 25% → 50% → 100% in cautious mode. Automatic rollback when a guardrail is breached; an exposed visitor stays exposed in the following steps. Where to check |
| Multi-armed bandit | Multi-armed bandits reallocate traffic to the best variant with Thompson Sampling, at a chosen frequency. The pricing page includes them from the free Developer plan.181920 | Optimization test: Thompson sampling, reallocated at most once a day, with a minimum share per variant. It maximizes a conversion rate, or an amount per visitor such as revenue: purchase probability and average order value drawn from their distributions, with values capped at a percentile as in an A/B test. Optionally, shares per device (mobile, tablet, desktop): one draw per device, each with its own learning and floor. A visitor always keeps their variant: a new split only applies to visitors who arrive after it. The screen says so: it maximizes conversions (or the amount per visitor) during the test, without proving that a variant is better. The difference is read period by period (and device by device), with no declared winner; the CSV export gives the same reading. Where to check |
| Revenue and margin per visitor | Custom numeric metrics measure amounts (for example, spend per user), with a confidence interval (frequentist) or a credible interval (Bayesian).2122 Margin per visitor is not mentioned in their public documentation (as of November 11, 2026). | Revenue and margin per visitor, each with its range; the gain is stated in euros. Margin is calculated on the server from reconciled orders, the cost of each product (Shopify or CSV), and store fees; no margin verdict below 90% of revenue covered by known costs. Where to check |
| Winner’s curse | In Bayesian mode, variants get a shrinkage prior toward the control by default, built from the empirical distribution of lifts observed across all experiments on the platform. No equivalent is documented for frequentist mode, the default mode.2324 | The reported gain is corrected for the winner’s curse by Bayesian shrinkage toward the effects of the store’s past tests. Where to check |
| Bayesian mode | Bayesian mode with a shrinkage prior by default, or a non-informative prior. It shows the probability to beat control, the probability to be best, and the expected loss.252627 | Optional, off by default: a Bayesian reading next to the verdict, with the probability that the variant does better, the credible interval, and the expected loss of each choice. The prior is centered on zero and as wide as the effects measured by the store’s completed tests; while there are too few of them to estimate it, the Bayesian reading is not shown. The decision always follows the sealed plan; when the two readings disagree, the report explains why. Where to check |
| Multiple comparisons | Optional correction of frequentist results with Bonferroni or Benjamini-Hochberg, applied to variants, metrics, or both. It does not apply to results filtered by segment.2829 | Segment comparisons are corrected with the Holm method. With several variants, the error risk is split across comparisons (Bonferroni). Where to check |
| Multivariate testing | Multivariate flags allow A/B/n tests and a tutorial mentions “multivariate experiments”; a factorial design or interaction report is not documented, and a guide points to a factorial design you set up yourself.303132 | Full factorial design generated by Probant: two to four sections, each with two to four levels, up to sixteen combinations, each served as a variant. Main effects of each section across all traffic and interactions between two sections, corrected with the Holm method on the comparisons stated in the plan. Based on the workspace’s measured traffic, the plan says whether the test can reach a conclusion and suggests removing a level or a section; no best combination is declared without a check. Where to check |
| Variance reduction (CUPED) | CUPED (adjustment with pre-experiment covariates) is turned on when the experiment is created. It only applies to metrics analyzed as means (not percentiles) and not to results filtered by attribute.3334 | CUPED applied to live test reports, using each visitor’s history before their first exposure; without history, the difference stays unadjusted and the report says so. Where to check |
| Visual editor | An article on the official blog cautions against using visual editors for experimentation.35 A visual editor is not mentioned in their public documentation (as of November 11, 2026). | Two ways to edit, producing the same changes: a page preview loaded by Probant, without its scripts, or the live page on the site, scripts included. On the site, the editor is loaded by the script already installed, through a short-lived signed link for one draft and one variant, with no browser extension. It only sends the chosen changes to Probant. Where to check |
| Widgets | Ready-made widgets are not mentioned in their public documentation (as of November 11, 2026). | Banner, countdown, pop-up, badge, reassurance, action bar, delivery date, social proof, promo code, free shipping, and survey, set up without code in the editor and saved in the workspace library for other tests. A widget is just a set of ordinary changes: escaped text, links over https or to the site, and styles applied without a style attribute, so a strict CSP does not need loosening. The countdown follows the site’s time zone and disappears at the deadline; the pop-in opens on arrival, on exit intent, or after scrolling, closes with the keyboard, and does not reappear during the visit if consent allows. Their small modules only load for visitors in the variant. The delivery date is calculated in the visitor’s browser (cutoff time, business days, public holidays); social proof shows the exact number of purchases received over a rolling week, and only above a threshold. The survey asks up to 8 questions (closed answers, recommendation scale, open question) with conditional logic, on the page or through a link outside the site (email, QR code), based on 70 templates (recommendation, satisfaction, effort, checkout blockers, cancellation, concept test, multi-question surveys…); open answers are redacted before sending (email addresses, phone numbers, cards, IBANs), kept for 180 days, and grouped by theme; each question is read by variant, with its range and response rate, and comparisons across all questions are corrected together; a response through a link is attached to the variant the visitor saw, recalculated by Probant, or stays outside the test. Where to check |
| Script and flicker | No ready-made tag is documented: the JavaScript SDK is integrated in the code. Flicker is handled by the developer (earlier initialization, render blocking, bootstrapping); no script size is published.3637 | Script under 7 KB compressed, loaded asynchronously, that does not hide the page while it loads. Server-side assignment available, for a display with no flicker at all. Where to check |
| Granular permissions | Predefined project and organization roles; custom roles with granular per-resource policies are limited to the Enterprise plan.383940 | Four fixed roles, plus custom roles the owner builds from a catalog of named permissions: prepare, launch and stop, roll out, approve a launch, write variant JavaScript, manage flags, metrics, margin, members, and keys. No privilege escalation: billing, roles, transfer, and deletion of the workspace stay with the owner; no one can grant more than their own permissions, and an API key cannot exceed its author’s. Where to check |
| Data hosting | EU instance on AWS Frankfurt (eu-central-1), limited to new Enterprise accounts and trials; Datadog monitoring and the Fastly CDN remain in the United States. Two documentation pages differ on Experimentation in the EU: listed as supported on the EU page, marked unavailable on the Experimentation page.41424344 | Database and compute in Amsterdam, with Scaleway, in the European Union. Scaleway, which hosts the database and compute, is a French company. Published subprocessor list: Scaleway (Amsterdam, Netherlands (nl-ams region), in the European Union), Brevo (European Union: servers in France and Germany, encrypted backups in Belgium (Google Cloud), according to Brevo), Mistral AI (European Union by default (European API endpoint), according to Mistral), ClickHouse (option not used), Stripe (Contract with Stripe Payments Europe, Ltd. (Ireland); data transferred to Stripe, LLC (United States), according to Stripe), Sentry (European Union: Frankfurt (Germany), including backups, according to Sentry). Depending on the feature used, Mistral may temporarily transfer data outside the European Union to the subprocessors it publishes; its data processing agreement limits these transfers to countries the EU recognizes as adequate, or covers them with standard contractual clauses. Stripe transfers payment data to Stripe, LLC, in the United States, under the Data Privacy Framework and, failing that, standard contractual clauses. Whatever region is chosen, Sentry may keep the data of Probant’s Sentry account (members, settings, audit log, project descriptions) in the United States; reported errors stay in Frankfurt. Where to check |
| Consent by default | The SDKs set no cookies, but the client SDK generates an anonymous key stored in localStorage and sends events; no built-in consent mode is documented, and deferring until consent is left to the integrator.454647 | Consent required by default: no cookie, no assignment, and nothing sent before the visitor agrees. Consent given mid-page: the page does not change in front of the visitor; their variant starts on the next page. First-party cookie, lasting 390 days at most, never extended. Where to check |
| Pricing model | Usage-based pricing with no per-seat fee: server SDK connections and client-side MAU, with paid overage on Foundation and negotiated volumes on Enterprise. Published example: $10 per connection beyond 5 and $8.33 per thousand MAU beyond 1,000, billed annually.484950 | Creating a Probant workspace is free; paid plans cost €590 to €1,990 per month depending on the number of tested visitors (2 months free with annual billing), with no overage charges. Agency plan: 1,000,000 tested visitors per month, shared across client workspaces, from 3 workspaces, for €1,490 per month or €14,900 per year. Where to check |
What LaunchDarkly also does.
Probant shares these points with LaunchDarkly.
A sequential test
Sequential mode relies on the confidence sequences of Howard (2021) and Waudby-Smith et al. (2024), called GAVI, and allows a decision at any time. It is on by default in frequentist mode.78
A multiple comparisons correction
Optional correction of frequentist results with Bonferroni or Benjamini-Hochberg, applied to variants, metrics, or both. It does not apply to results filtered by segment.2829
Variance reduction (CUPED)
CUPED (adjustment with pre-experiment covariates) is turned on when the experiment is created. It only applies to metrics analyzed as means (not percentiles) and not to results filtered by attribute.3334
Revenue per visitor, with its range
Custom numeric metrics measure amounts (for example, spend per user), with a confidence interval (frequentist) or a credible interval (Bayesian).2122
Gradual rollouts
Guarded rollouts increase traffic in steps, monitor metrics, and can roll back automatically on regression. The documentation ties them to the Enterprise plan with the Guardian add-on; the current pricing page includes them in all plans (limited on Developer).151617
When to choose LaunchDarkly over Probant.
Based on LaunchDarkly’s public documentation.
Your tests live in code, behind feature flags
Unlimited feature flags with targeting by user, segment, or rule, from the free plan.51
You want to analyze in your warehouse
Experiments can use metrics computed in the customer’s warehouse (BigQuery, ClickHouse, Databricks, Redshift, Snowflake), then export the data back. Included in all plans according to the pricing page.5253
You want a free plan
Free Developer plan with no time limit, capped on usage (5 server connections, 1,000 client MAU), with experimentation included. A 14-day trial is also offered.54555657
What sets Probant apart.
These points are not in LaunchDarkly’s public documentation, accessed November 11, 2026. A missing mention does not prove a feature is missing; if these points matter to you, ask LaunchDarkly.
Margin per visitor
Probant measures revenue and margin per visitor, each with its range.
LaunchDarklyMargin per visitor is not mentioned in their public documentation (as of November 11, 2026).
Learn moreSRM that blocks the verdict
A traffic split imbalance moves the test to “blocked” status, and no winner is declared.
LaunchDarklyIn an experiment, SRM is flagged by a health check (“Health errors”) without blocking results. In a guarded rollout, an SRM triggers a rollback, even when automatic rollback is not turned on.91011 Automatic blocking of the test is not mentioned there (as of November 11, 2026).
Consent required by default
No cookie, no assignment, and nothing sent before the visitor agrees.
LaunchDarklyThe SDKs set no cookies, but the client SDK generates an anonymous key stored in localStorage and sends events; no built-in consent mode is documented, and deferring until consent is left to the integrator.454647 Consent required by default is not mentioned there (as of November 11, 2026).
A test stopped by its guardrails
A breached guardrail stops a running test on its own and blocks its verdict.
LaunchDarklySecondary metrics act as guardrails in experiments, with no documented automatic stop. Automatic stop and rollback exist for “guarded rollouts,” which do not run at the same time as an experiment on the same flag.121314 An automatic stop of the test on a guardrail is not mentioned there (as of November 11, 2026).
Learn more
Frequently asked questions.
How much does LaunchDarkly cost?
Usage-based pricing with no per-seat fee: server SDK connections and client-side MAU, with paid overage on Foundation and negotiated volumes on Enterprise. Published example: $10 per connection beyond 5 and $8.33 per thousand MAU beyond 1,000, billed annually.484950 Creating a Probant workspace is free; paid plans cost €590 to €1,990 per month depending on the number of tested visitors (2 months free with annual billing), with no overage charges.
Where does LaunchDarkly host data?
EU instance on AWS Frankfurt (eu-central-1), limited to new Enterprise accounts and trials; Datadog monitoring and the Fastly CDN remain in the United States. Two documentation pages differ on Experimentation in the EU: listed as supported on the EU page, marked unavailable on the Experimentation page.41424344 Probant hosts its database and application in Amsterdam, with Scaleway.
Does LaunchDarkly measure margin?
Margin per visitor is not mentioned in their public documentation (as of November 11, 2026). Probant measures revenue and margin per visitor, each with its range, from orders and the cost of each product.
Can you look at LaunchDarkly results while a test is running?
Sequential mode relies on the confidence sequences of Howard (2021) and Waudby-Smith et al. (2024), called GAVI, and allows a decision at any time. It is on by default in frequentist mode.78 Probant sets its checkpoints before launch (O’Brien-Fleming) and uses always-valid ranges during rollouts.
Sources.
Official pages from LaunchDarkly, accessed November 11, 2026. Quotes are reproduced in their original language.
- Source 1: launchdarkly.com/about-us
Today, LaunchDarkly is the runtime control layer for the AI era of software development, reflecting how we’ve evolved from pioneering feature management into a broader platform for controlling software releases, experimentation, observability, and AI agent behavior in production.
- Source 2: launchdarkly.com/about-us
LaunchDarkly was founded in 2014 in Oakland, California, by Edith Harbaugh and John Kodumal, to solve a clear problem in modern software delivery: Teams were releasing code faster in the cloud, but they still lacked safe, practical control over what happened after deployment.
- Source 3: launchdarkly.com/about-us
LaunchDarkly gives large enterprises and small startups the tools they need to maintain real-time control of AI code and agents.
- Source 4: launchdarkly.com/docs/guides/statistical-methodology/choosing
LaunchDarkly offers two different statistical approaches for analyzing your experiment: Bayesian and frequentist. Frequentist is the default approach in LaunchDarkly.
- Source 5: launchdarkly.com/docs/guides/statistical-methodology/choosing
For this reason, at LaunchDarkly we default the frequentist analysis to use sequential testing.
- Source 6: launchdarkly.com/docs/guides/statistical-methodology/methodology-frequentist
LaunchDarkly uses the industry- and scientific-standard z-test for fixed-horizon analyses on metrics based on means.
- Source 7: launchdarkly.com/docs/guides/statistical-methodology/methodology-frequentist
For sequential analysis LaunchDarkly uses the method of Howard (2021) and Waudby-Smith et. Al. (2024), sometimes referred to as generalization of always valid inference, or GAVI.
- Source 8: launchdarkly.com/docs/guides/statistical-methodology/choosing
In the sequential method, you can take action on your results at any time.
- Source 9: launchdarkly.com/docs/home/experimentation/health-checks
Health errors: There is at least one critical error with your experiment configuration, such as a flag or config toggled off, or an SRM.
- Source 10: launchdarkly.com/docs/home/experimentation/health-checks
The sample ratio mismatch check alerts you to any detected SRMs.
- Source 11: launchdarkly.com/docs/home/releases/managing-guarded-rollouts
LaunchDarkly rolls back a release with an SRM whether or not you have automatic rollback enabled.
- Source 12: launchdarkly.com/docs/guides/experimentation/designing-experiments
Often, they are “guardrail” metrics that can ensure an experiment’s success.
- Source 13: launchdarkly.com/docs/home/releases/guarded-rollouts
You can create a guarded rollout on any targeting rule, as long as no other guarded rollouts, progressive rollouts, or experiments are running on the flag or AgentControl config, and the flag is not a migration flag.
- Source 14: launchdarkly.com/docs/home/releases/guarded-rollouts
When LaunchDarkly detects a regression, the metric tile highlights the regression. If you enabled automatic rollback, LaunchDarkly also rolls back the release.
- Source 15: launchdarkly.com/docs/home/releases/guarded-rollouts
A guarded rollout progressively increases traffic to the new variation while monitoring selected metrics for regressions until the rollout reaches 100%. If LaunchDarkly detects a regression before the rollout reaches 100%, it can pause the rollout and send a notification.
- Source 16: launchdarkly.com/docs/home/releases/guarded-rollouts
Guarded rollouts are available to customers on an Enterprise plan with the Guardian add-on.
- Source 17: launchdarkly.com/pricing
Roll out gradually while watching guardrail metrics, and roll back automatically on a regression.
- Source 18: launchdarkly.com/docs/home/multi-armed-bandits
LaunchDarkly’s multi-armed bandit implementation uses Thompson Sampling.
- Source 19: launchdarkly.com/docs/home/multi-armed-bandits
Unlike traditional A/B experiments, which split traffic between variations and waits for performance results, MABs continuously evaluate variation performance and automatically shift traffic toward the best performing variation.
- Source 21: launchdarkly.com/docs/home/metrics/custom-numeric
Custom numeric metrics track changes to the amount of something, such as dollars spent or latency time.
- Source 22: launchdarkly.com/docs/home/experimentation
Depending on your statistical methodology, LaunchDarkly surfaces frequentist confidence intervals or Bayesian credible intervals so you can see how much evidence supports each variation, not just the raw metric values.
- Source 23: launchdarkly.com/docs/guides/statistical-methodology/choosing
By default, LaunchDarkly shrinks treatment results toward the control, so the numbers on the results page are not reflective solely of your observed data.
- Source 24: launchdarkly.com/docs/guides/statistical-methodology/methodology-bayesian
We generate this prior from the empirical distribution of relative differences between variations in all experiments on our platform using metrics of the same type (numeric or conversion) and aggregation function (average or sum).
- Source 25: launchdarkly.com/docs/guides/statistical-methodology/methodology-bayesian
Shrinkage priors (default): Each treatment variation uses a prior centered on the control variation's mean. This pulls treatment results toward the control when data is limited.
- Source 26: launchdarkly.com/docs/guides/statistical-methodology/methodology-bayesian
Expected loss represents the average potential downside of shipping a variation, quantifying how much one could expect to lose if it underperforms relative to the control variation.
- Source 27: launchdarkly.com/docs/home/experimentation/bayesian-results
The probability to beat control represents the likelihood that this variation performs better than the control variation for a given metric.
- Source 28: launchdarkly.com/docs/guides/statistical-methodology/mcc
LaunchDarkly gives you the option to correct your frequentist experiment results for multiple comparisons arising from having more than one metric or more than one treatment configured on your experiment.
- Source 29: launchdarkly.com/docs/guides/statistical-methodology/mcc
The two corrections we offer at LaunchDarkly, the Bonferroni correction and the Benjamini-Hochberg procedure, are designed to mitigate this risk by changing the threshold for declaring significance to avoid declaring significance too easily when multiple comparisons are at play.
- Source 30: launchdarkly.com/docs/home/experimentation
Compare multiple variations of one experience (A/B/n testing): Assign contexts to two or more flag or AgentControl config variations and measure which variation performs best on the metrics you attach to the experiment.
- Source 31: launchdarkly.com/docs/tutorials/ld-arch-deep-dive
Experimentation: Full-featured A/B testing and multivariate experiments.
- Source 32: launchdarkly.com/docs/guides/experimentation/ai-experiments-roi
If interaction effects are important to your use case, consider running a proper factorial experiment design.
- Source 33: launchdarkly.com/docs/home/experimentation/analyze
Applying CUPED to experiment results makes it easier to detect differences between variations. CUPED is compatible only with metrics that use the average (mean) analysis method.
- Source 34: launchdarkly.com/docs/guides/statistical-methodology/methodology-frequentist
CUPED is not compatible with percentile metrics.
- Source 35: launchdarkly.com/blog/have-you-outgrown-your-wysiwyg-experimentation-tool
This is exactly what happens when visual editor tools are used for experimentation.
- Source 36: launchdarkly.com/docs/sdk/client-side/javascript/default-values
If the page is initially rendered with one set of flag variations and then updated with a new set, this can cause a jarring switch between two views of the content.
- Source 37: launchdarkly.com/docs/sdk/client-side/javascript/default-values
You have several options for addressing this initial "flicker," including the following:
- Source 38: launchdarkly.com/docs/home/account/roles/role-create
LaunchDarkly provides both project and organization preset roles.
- Source 39: launchdarkly.com/docs/guides/teams-roles/custom-roles
Creating your own roles is only available to customers on select plans.
- Source 40: launchdarkly.com/pricing
Write your own roles with resource-level policies, group members into teams, and keep an unlimited history of who changed what.
- Source 41: launchdarkly.com/docs/home/infrastructure/eu
The EU instance resides in the AWS Frankfurt (eu-central-1) region and operates nearly the same as the US instance of LaunchDarkly. Operational monitoring with Datadog and edge delivery/CDN (Fastly) still operate from regions within the United States.
- Source 42: launchdarkly.com/docs/home/infrastructure/eu
The EU instance only supports net new Enterprise plan accounts and LaunchDarkly trials.
- Source 43: launchdarkly.com/docs/home/infrastructure/eu
Experimentation, including Snowflake native Experimentation
- Source 44: launchdarkly.com/docs/home/experimentation
This feature is not available on LaunchDarkly's European Union (EU) instance
- Source 45: launchdarkly.com/docs/sdk/concepts/client-side-server-side
LaunchDarkly SDKs do not store cookies.
- Source 46: launchdarkly.com/docs/home/flags/anonymous-contexts
The SDK generates a unique identifier for the context and persists it in local storage.
- Source 47: launchdarkly.com/docs/sdk/features/experimentation
You need identify() when the context's identity changes after the SDK has already initialized, for example, when a user on a public page logs in, when identity loads asynchronously after SDK startup, or when a user completes a cookie consent flow.
- Source 48: launchdarkly.com/pricing
No per-seat pricing. Invite your whole team; pricing is based on usage.
- Source 49: launchdarkly.com/pricing
5 connections over the free 5 at $10 each, plus 1K MAU over the free 1K at $8.33 per 1K, billed yearly.
- Source 50: launchdarkly.com/pricing
Keep going past the free amounts and pay only for the usage you add.
- Source 51: launchdarkly.com/pricing
Create as many feature flags as you need and target them by user, segment, or rule.
- Source 52: launchdarkly.com/docs/home/warehouse-native/warehouses
This category includes documentation about running LaunchDarkly experiments using metric events from your external warehouse.
- Source 53: launchdarkly.com/pricing
Warehouse-native integration (Snowflake, Redshift, Databricks, BigQuery and more)
- Source 55: launchdarkly.com/pricing
Each unique client-side user or device counts as one monthly active user. 1K free
- Source 56: launchdarkly.com/pricing
Every replica, worker and dev or staging environment counts. 5 free
Migrate from LaunchDarkly
Your LaunchDarkly history can be imported as a generic CSV (visitors and conversions per variant) and is reread by Probant’s engine: every verdict is recalculated from visitors and conversions.
- Open demo, no sign-up
- Free to start, no overage charges
- Database and compute in Amsterdam, with Scaleway