TL;DR:

500 2,500 credits free every month, then $0.01/credit

5x the free tier for a limited time! 1 credit = $0.01. That's ~500 free observations a month on the standard model – you only pay for what your scanners run.

Get started – freeTalk to a human

Plans

Totally freePay-as-you-go
Price per credit$0No credit card required$0.01/credit1 credit = $0.01
Monthly free credits500 2,500500 2,500
Monthly volumeUp to 2,500 creditsUnlimited
All scanner types
All AI models
Custom monthly spending limit
Deep-link citations & confidence scores
Observations as queryable events
Get started – freeGet started – free

Calculate your cost

An observation is a single scanner watching one recording. It costs credits (1 credit = $0.01) based on its AI model. Pick a model and drag the slider to estimate your monthly cost.

Model

First 500 observations/mo500Free$0
500+0$0.05/observation$0
UsageObservations / mo

Observations

$0.05 each after the first 500

The first 500 observations are free, every month.

Estimated monthly cost

$0

PostHog vs...

An honest comparison, tl;dr:

Choose a competitor if...

  • You want self-hosting or more strict data residency
  • You want a dedicated UX-research tool with targeted clip capture and studies
  • You're not on PostHog and don't want to move session replay here

Go with if...

  • Point AI at any slice of recordings on a schedule and write the prompt yourself
  • Findings land as queryable observations next to your analytics, funnels, flags, experiments, and errors
  • What AI sees in your recordings becomes queryable data and signals that feed the self-driving loop, in the same platform as your analytics, flags, and experiments.
  • You'd rather keep sessions on PostHog than pay a second tool to scrape them out
PostHog's mascot and what competitor's mascots would look like if they had Lottie

Feature comparison

FullStory
compare
Contentsquare
Datadog
Mixpanel
compare
AI scanners
Point a scanner at a filtered recording set
Run AI analysis across a whole slice of sessions, not one recording at a time.
✓
10 per run
100 per run
✗
Playlist
Configurable scanner types
Choose the shape of the output rather than taking a single fixed job.
Monitor, classifier, scorer, summarizer
Fixed jobs (UI), schema via API
Planned analyses (Sense Analyst)
Fixed job
Fixed job
Custom prompt per analysis
Describe exactly what to look for in your own words for each scanner.
✓
API only, Enterprise/Advanced
Sense Analyst (beta)
✗
✗
Yes/no monitors (did this happen?)
Get a simple did-this-happen verdict on every session a scanner watches.
✓
API only
✗
✗
✗
Classify / tag sessions
Automatically bucket sessions by intent, outcome, or any label you define.
✓
Sentiment (UI), any label API-only
✗
✗
✗
Vendor-trained friction score
A trained, benchmarked numeric score for each session, out of the box.
✗
✗
0–100, ML-trained
✗
✗
Score sessions on your own criteria
Define what gets scored and on what scale, rather than taking a vendor-defined score.
✓
API only
✗
Custom metrics, not AI
✗
Cross-session theme summary
Roll many sessions up into the recurring themes and patterns across them.
✓
10 per run
100 per run
✗
Beta
Natural-language search over recordings
Find the recordings you need by describing them in plain language.
✓
✓
✓
Telemetry only
Via MCP
Scheduled / continuous runs
Scanners keep running on new sessions automatically, not just on demand.
✓
Fixed jobs only
✓
Fixed job
Fixed job
Sampling / coverage controls
Dial how much matching traffic gets scanned so cost stays predictable.
✓
✗
✗
✗
✗
Deep-link citations to the exact moment
Every finding links straight to the timestamp in the recording so you can verify it.
✓
✓
✓
✓
✓
Mobile replay AI
Run the same AI analysis over native iOS and Android session recordings.
✓
✓
✓
✗
✗
Results as data
Findings become queryable events
Each observation is emitted as an event you can query alongside the rest of your data.
✓
Via API
Frustration score only
✗
✗
Chart scanner output values over time
Chart and monitor the scanner's own output values, not just the underlying behavioural signals.
✓
✗
Frustration score only
✗
✗
Feed experiments and cohorts with scanner output values
Build cohorts from a scanner's output values and measure them in experiments.
✓
✗
Frustration score only
✗
✗
Proactive alerts / anomaly detection
Get notified when a scanner's output spikes or crosses a threshold.
✓
✓
✓
✗
✗
Access and agents
MCP / agent access
Author, run, and read AI analyses of recordings from any MCP client.
✓
Not documented
✓
RUM events only
✓
REST API for AI output
Pull scanner observations programmatically over a REST API.
✓
✓
✓
✗
✗
Feeds the self-driving loop
Findings from recordings raise signals that agents can research and turn into a pull request.
✓
✗
✗
✗
✗
Recordings and platform
Share recordings
Share a link to any recording with your team.
✓
✓
✓
✓
✓
Embed recordings
Embed recordings in docs, dashboards, or other tools.
✓
✗
Qualtrics only
Datadog notebooks only
✗
Export recordings
Export recordings out of the platform when you need to.
✓
✗
✗
Segments via API
✗
Feature-flag interlinking
Jump between recordings and the feature flags and experiments they hit.
✓
Self-instrumented
Via Optimizely
✓
✓
Product analytics in same platform
Analytics, replays, and AI findings all live in one product, not stitched together.
✓
✓
✓
✓
✓
PII redaction / masking
Mask sensitive content before it is ever recorded.
✓
✓
✓
✓
✓
Pricing
AI pricing model
How the AI analysis is billed.
Usage-based
Tier + add-on
Tier + add-on
Per 1k sessions
Tier + add-on

Still here?

The hedgehog has been watching this whole time.

It's free to start. Not "free trial" free – actually free. No card, no call, no "someone from our team will be in touch." Just PostHog.

Get started – freeTalk to a human

No sales call. No "tailored demo." No expiry.