In-depth two-product comparison
Amplitude vs PostHog
A practical Amplitude versus PostHog comparison for B2B SaaS teams, covering analytics governance, open source, event usage, MTU limits, ownership and a worked volume scenario.
Last updated: 2026-10-08
Short answer
Choose Amplitude when product managers, analysts and leadership need a governed analytics system with durable definitions, shared work and controlled access. Choose PostHog when engineers want one instrumented stack they can run in the cloud or self-host, with usage-based billing and free monthly allowances for each product. At a stated 12 million events per month, PostHog requires a usage calculation and an explicit billing limit; Amplitude requires a quote once a team moves beyond its published free allowance. The right owner is the deciding factor: Amplitude usually sits with product analytics, while PostHog often starts with engineering and grows into a shared product workflow.
Amplitude
Governed product analytics for teams that need reliable definitions and collaboration.
PostHog
A developer-friendly, open-source suite that puts more product capabilities in one place.
Amplitude fits a team that values governed product metrics and cross-functional analysis, while PostHog fits a developer-led team that wants analytics alongside replay, flags and experiments. Let the people who will own instrumentation and metric definitions run the evaluation before choosing a plan.
Quick comparison
| Decision area | Amplitude | PostHog | What to verify |
|---|---|---|---|
| Primary job | Help product and data teams analyze journeys, retention and cohorts with shared definitions. | Let an engineering-led team capture behavior and operate analytics, replay, flags and experiments together. | Ask who will own taxonomy changes, dashboard review and the response to a bad event. |
| Hosting | Vendor-hosted service with governance and access controls. | Cloud regions in the US or EU, plus a self-hosted MIT-licensed option. | Confirm whether your security team requires self-hosting or an approved cloud region. |
| Pricing basis | Free and Plus plans use event allowances; Growth and Enterprise are custom event-based proposals. Some teams still describe usage in MTUs, so define the unit in the order form. | Usage-based per product with a free monthly allowance, then pay for units used; no fixed public monthly plan price. | Model the same event, identified-user and replay volumes with each vendor before procurement. |
| Product breadth | Product analytics with experimentation, replay, guides and surveys in the wider platform. | Analytics, session replay, feature flags, experiments, surveys and data warehouse in one suite. | List the separate tools you would retire and the capabilities you still need from specialists. |
| Likely owner | Product analytics or product management, with data engineering and operations maintaining the model. | Engineering or developer experience at first, with product managers becoming regular consumers. | Name the person who can change instrumentation without breaking reporting. |
| Starting price | $0 monthly for 2M events on the Free plan; Plus starts at $0 with the first 2M events free. | $0 monthly within the free allowance; usage above each allowance is pay-as-you-go. | Treat $0 as the published free allowance, not as an estimate for a larger workload. |
Published free allowances are useful for a first model, but a 12M-event workload needs a vendor-confirmed quote or calculator result with the same event, profile, replay and retention assumptions.
Feature comparison
| Feature | Amplitude | PostHog |
|---|---|---|
| Event definitions | A stronger fit when product managers and analysts need a shared taxonomy, controlled definitions and a review process for important metrics. | HogQL and raw event access give engineers flexibility, but the team must create its own conventions, ownership and review discipline. |
| Cohorts and retention | Useful for structured behavioral cohorts, retention views and repeatable product questions across teams. | Supports behavioral analysis and SQL-like queries; expect more hands-on work when a PM needs a carefully governed metric. |
| Collaboration | Designed for shared analysis spaces, documented metrics and access controls across product, analytics and leadership. | Dashboards and notebooks can be shared, while larger teams should define naming, permissions and review practices themselves. |
| Predictive analysis | Evaluate Amplitude if predictive cohorts or modeled behavioral questions are part of the product analytics brief. | Focuses on descriptive and diagnostic product data; confirm any predictive workflow separately rather than assuming parity. |
| Feature flags | Amplitude’s experimentation offering can cover flags, but confirm which product and commercial scope are in your agreement. | Feature flags are built into the suite with targeting and rollout controls, so engineering can connect a flag to the same event stream. |
| Session replay | Session replay is available in Amplitude’s platform; confirm the allowance and retention that apply to your plan. | Replay is a central part of the product suite and can be inspected alongside captured events. |
| Experiments | A fit for organizations that want experimentation connected to governed product analysis and approval processes. | Experiments sit close to feature flags, which can shorten the path from code change to behavioral readout. |
| Surveys and warehouse | Guides and surveys support product feedback; check whether your desired warehouse workflow needs an additional data product. | The suite includes surveys and a data warehouse allowance, alongside analytics and replay. |
| Engineering responsibility | Engineering implements a stable event contract, while analytics or product operations usually governs the meaning of events. | Engineering can own instrumentation, flag rollout and self-hosting, which is attractive when developers want direct control of the stack. |
| Product management responsibility | PMs can work from shared reports and cohorts without writing queries for every recurring question, provided the taxonomy is maintained. | PMs can use dashboards and experiments, but the team should invest in templates and enablement so raw flexibility does not become inconsistent reporting. |
| Data residency | Review Amplitude’s contractual and regional controls with security before choosing it for regulated product data. | PostHog documents US (Virginia) and EU (Frankfurt) cloud regions and offers self-hosting for teams that need infrastructure control. |
Pricing and operating cost
Amplitude
Amplitude Free includes 2M events per month with no time limit. Plus starts at $0, includes the first 2M events free and can scale up to 70M events. Growth and Enterprise use custom event-based pricing.
A quote is needed when your event volume or governance requirements exceed the published free and Plus description. Budget for taxonomy stewardship, data quality checks and implementation time even when the subscription starts at $0.
PostHog
PostHog has a recurring free allowance on each product: 1M analytics events, 5K session recordings, 1M feature-flag requests, 1,500 survey responses and 1M data-warehouse rows are examples from the published plan. Usage stops at the free limit; paid usage is pay-as-you-go.
Usage can span several meters: analytics events, identified profiles, recordings, flag requests, surveys and warehouse rows. Self-hosting shifts hosting, upgrades, monitoring, backups and incident response to your engineering team.
| Tier or usage case | Amplitude | PostHog |
|---|---|---|
| Published free allowance | $0 monthly for 2M events Product analytics, 10K monthly session replays, limited experiments, guides and surveys. | $0 monthly within each product allowance Unlimited team members, one project and one year of retention; each product has its own allowance. |
| Growing workload | Plus starts at $0; event rates beyond the free allowance are not published in the research record. Plus lists up to 70M events, cohorts, alerts, heatmaps and two-year retention. | Pay-as-you-go; pricing on request for a specific mix of event, profile, replay and other product usage. Set billing limits per product. Paid plans list six projects, seven-year event retention and email support. |
| Governed or enterprise deployment | Growth and Enterprise: custom event-based pricing. Advanced exploration, SSO, project permissions, RBAC and expanded replay allowances vary by plan. | Annual plans and volume discounts are available through sales; pricing on request. Platform packages can add SSO, audit logs, custom roles and project permissions. |
The practical difference is governance versus control
Amplitude is the more natural home when the analytics operating model is led by product management, product analytics or a central data team. Those teams typically need a dependable vocabulary for activation, retention and conversion, then need to share it with executives and multiple product squads. The implementation still depends on engineering, but the day-to-day question is often “which behavior is true for this segment?” rather than “how do we wire another service into the stack?” Define the event contract and metric owners before inviting every team into the workspace.
PostHog suits a different starting point. Its open-source, MIT-licensed product can be self-hosted, and its cloud product places analytics, replay, feature flags, experiments, surveys and a data warehouse in one suite. That breadth is useful when an engineer wants to capture an event, gate a release and inspect the resulting session without stitching together several vendors. It also creates an operating obligation: decide who reviews event names, who owns flag cleanup and who approves changes to a shared query before a PM treats it as a company metric.
Neither ownership pattern is permanent. A small B2B SaaS may start with an engineer in PostHog and later add a product analyst. An enterprise team may start with Amplitude dashboards and later ask engineering to manage experimentation. During evaluation, put the person who will maintain the system in the demo. Ask them to add one event, correct one bad property, build one cohort and explain the change to a non-technical stakeholder.
A worked volume scenario: 12 million events and 8,000 active users
Consider a B2B SaaS with 8,000 monthly active users, 12 million product events per month and 300,000 identified-user events inside that total. The company wants feature flags, replay for debugging, two experiments per quarter and a warehouse view for account-level reporting. It has four engineers, two PMs and one data analyst. This is a planning example, not a quote: it shows which questions the buyer should put in front of each vendor.
Amplitude covers the first 2M events on its Free plan at $0 monthly. Its Plus description also starts at $0 and can scale up to 70M events, but the research record does not publish the rate for the additional 10M events in this scenario. The buyer should request the current event-based proposal and ask whether the team’s account reporting, replay allowance and experimentation needs fit the selected plan. If governance and shared definitions are the reason for buying, include analyst and PM workflow requirements in that proposal rather than comparing only an ingestion number.
PostHog supplies 1M analytics events free each month, then bills usage by product. Its pricing record lists a 1–2M base rate of $0.0000500 per event and a 2–15M base rate of $0.0000343 per event, while identified events also incur a person-profile charge after the first 1M profile events. The exact bill depends on the event mix and the other products: 5K recordings, 1M flag requests and 1,500 survey responses are separate allowances. The team can set a billing limit per product, which makes the first rollout easier to control, but it should still model the real mix rather than multiply one headline rate across every meter.
For this company, engineers would likely own SDK installation, flag cleanup and any self-hosted deployment; the analyst would own the event dictionary; PMs would own experiment questions and dashboard decisions. If the analyst spends most of the evaluation fixing inconsistent definitions, Amplitude’s governed workflow may justify its quote. If the urgent problem is replacing separate replay, flag and survey tools while retaining infrastructure control, PostHog may deliver more operational value even after hosting work is counted.
Migration and rollout checklist
Moving from Amplitude to PostHog usually means translating dashboards and event names into a more code-oriented workflow, then deciding which existing replay, flag or experimentation tools can be retired. Moving from PostHog to Amplitude usually means narrowing a flexible event stream into governed definitions and finding homes for capabilities that were bundled into PostHog. Preserve raw exports and a metric map so historical reports remain interpretable during the transition.
- Inventory events, properties, account identifiers, retention reports, flags, experiments, recordings and warehouse jobs.
- Choose five business-critical metrics and document their exact filters, owner, freshness expectation and accepted data window.
- Instrument a parallel sample for one release cycle; compare counts, identified-user handling and account joins before redirecting dashboards.
- Assign engineering ownership for SDKs and flags, analytics ownership for definitions, and PM ownership for questions and experiment decisions.
- Set PostHog product-level billing limits or obtain an Amplitude proposal that spells out event volume, replay, governance and support.
Questions to ask before signing
- Which unit is on the order form—events, MTUs, identified profiles, recordings or a combination—and what happens at the limit?
- Can a PM create a trusted cohort without engineering help, and who reviews a metric before it becomes a shared definition?
- For PostHog, which products are included in the cloud allowance, what can be self-hosted, and who owns upgrades and backups?
- For Amplitude, which replay, experimentation, guides and survey allowances are included in the quoted plan?
- How will account-level identity work for a B2B product with multiple users, workspaces and parent companies?
Pros and cons
Amplitude
Pros
- Strong fit for governed product analytics and shared metric definitions.
- Free plan includes 2M events per month with no time limit; Plus starts at $0.
- Useful collaboration and access-control story for cross-functional product teams.
Cons
- Growth and Enterprise pricing is custom event-based pricing.
- Some teams need a separate operating plan for developer-owned flags and deployment workflows.
- No self-hosted Amplitude deployment is listed in the research facts used here.
PostHog
Pros
- Open-source MIT-licensed product with a self-hosting option.
- Analytics, replay, flags, experiments, surveys and warehouse capabilities share one suite.
- Free monthly allowances and per-product billing limits make an incremental rollout possible.
Cons
- A large workload can span several usage meters and needs a volume model.
- Teams must create strong conventions for governance, flag cleanup and shared metrics.
- Self-hosting transfers infrastructure and maintenance work to engineering.
Who should choose what?
Governed product organization
Several squads need the same retention and activation definitions, with analysts reviewing changes.
Recommendation: Amplitude. Evaluate Amplitude first because governed analytics, collaboration and controlled access are the core operating need.
Developer-led product team
Engineers want analytics, replay, flags and experiments close to the code and may self-host.
Recommendation: PostHog. PostHog puts those capabilities in one open-source suite and lets the team choose cloud or self-hosting.
12M-event B2B SaaS
A four-engineer, two-PM team has 8,000 monthly active users and needs replay, flags and account reporting.
Recommendation: PostHog. Start by modeling PostHog’s event, profile and replay meters with billing limits, then compare the result with Amplitude’s event-based proposal.
The decision follows the operating model: governed product analytics under a shared definition owner, or a developer-controlled suite with open-source deployment choices.
Final verdict
Choose Amplitude if
- Product analytics needs governed definitions, durable cohorts and cross-functional collaboration.
- PMs and analysts should answer recurring questions without relying on raw SQL for every report.
- A custom event-based proposal is acceptable in exchange for stronger controls and shared workflows.
Choose PostHog if
- Engineering wants analytics, replay, flags, experiments, surveys and warehouse features together.
- Self-hosting or open-source transparency is a meaningful requirement.
- The team can model several usage meters and maintain its own conventions and infrastructure.
FAQ
Is PostHog a replacement for Amplitude?
It can replace core product analytics for teams comfortable owning more of the instrumentation and governance. Its broader suite can also replace separate replay, flag, experiment, survey and warehouse tools. Amplitude remains the stronger candidate when shared definitions, analyst workflows and controlled collaboration are the main requirement.
Does Amplitude price by MTUs or events?
Ask the vendor to define the unit in the current order form. The research record for this page describes the public Free and Plus allowances in events and Growth and Enterprise as custom event-based pricing. Some organizations still use MTU language in analytics planning, so do not compare a quoted MTU limit with an event allowance without reconciling the definitions.
Can PostHog be self-hosted?
PostHog documents an open-source MIT-licensed product and a self-hosting option. Self-hosting means your team owns infrastructure, upgrades, monitoring, backups and incident response. Cloud customers can select US (Virginia) or EU (Frankfurt) regions.
What does PostHog’s free plan include?
The published free allowances include 1M analytics events, 5K session recordings, 1M feature-flag requests, 1,500 survey responses and 1M data-warehouse rows, among other product meters. Usage stops at each free limit; paid usage is pay-as-you-go.
Who should own the implementation?
Engineering should own SDKs, identity plumbing and flag lifecycle. A product analyst or data owner should govern the event dictionary and shared metrics. PMs should own the product questions and experiment decisions. Name all three responsibilities even when one person covers more than one role.