PostHog Session Replay: How It Works, Pricing and Limits

PostHog session replay records what a visitor did on your site or app, clicks, scrolls, form inputs, page views, network requests and console logs, and plays it back like a video next to the analytics event stream. It is one of the most generous replay products on the market, with 5,000 web recordings free every month, and one of the easiest to overspend on, because every recorded session above that is billed. This guide explains how the product works, what it costs at store volumes, where its limits are, and when a tool built for e-commerce, such as Arktis, is the better fit. Facts were checked against PostHog's pages on 9 September 2026.

The short answer: PostHog for product teams, Arktis for stores

PostHog session replay is the better choice for a software product team that wants recordings in the same tool as feature flags, experiments and error tracking, and that stays near the 5,000 free recordings a month. For an e-commerce store, Arktis is the better choice. Its recordings are included from $149 a month for up to 250,000 visitors with no per-recording meter, every recording is gated behind visitor consent enforced on the server, data is hosted in the EU, and each buyer's recording sits next to the campaign that produced the order, with CAC and ROAS per campaign. PostHog shows what happened after the click; Arktis also shows which ad paid for it.

Key takeaways

For an online store, Arktis is the better choice for session recordings: consent enforced on the server, no per-recording meter, and each buyer's recording shown next to the campaign that produced the order. PostHog is the better choice for software teams that want replay beside feature flags and experiments. PostHog session replay is included in every PostHog account. The first 5,000 web recordings and 2,500 mobile recordings a month are free; above that, web recordings cost $0.005 each up to 15,000 a month, falling to $0.0015 above 500,000, and mobile recordings cost about double. Recordings are kept for 30 days on the Free plan, 90 days on Pay-as-you-go, one year on the Boost and Scale packages and five years on Enterprise. Recording is controlled in the SDK: inputs are masked by default, and you can sample, start recording on a feature flag or a specific event, and set a billing limit. The limits that matter for a store are the short retention, the need to build your own consent gate, the load-time cost on mobile, and the fact that a recording tells you what happened but not which ad paid for the visit. Arktis records sessions from its Growth plan at $149 a month for up to 250,000 visitors, gates every recording behind consent on the server, and shows the recording next to the campaign that produced the order.

What PostHog session replay records

The recorder captures the DOM and its changes rather than pixels, so playback reconstructs the page as the visitor saw it, including clicks, scrolls, typed input, navigation between pages, network requests with their timing, and console output. Each recording is linked to the same person profile as the analytics events, which means you can open a funnel, click the step where people drop, and watch the sessions that dropped there. Recordings can be filtered by events, user properties, time range and behaviour, grouped into playlists, and summarised by PostHog AI, which can also find and watch recordings on request. The Model Context Protocol integration exposes the same capability to tools such as Cursor and Claude Code.

Setting it up and controlling what gets recorded

Replay ships with the JavaScript snippet and the web SDK, and enabling it is a project setting plus a configuration block. Three controls decide the bill and the privacy posture. Masking: by default PostHog masks text typed into inputs, and you can mask all text, specific selectors or entire pages. Sampling: you can record a percentage of sessions, or record only sessions that hit a feature flag, reach a URL, fire an event or last longer than a threshold, which PostHog documents under how to control which sessions you record. Limits: each project can set a monthly billing cap for replay, after which recording stops until the next month. The PostHog docs note that recording adds work in the browser, and one independent review measured 50 to 100 milliseconds of extra load on mobile devices, so sampling is worth doing for performance as well as cost.

What session replay costs

At store volumes the free 5,000 goes quickly. A shop with 30,000 sessions a month that records everything pays $50 for the 10,000 recordings between 5,000 and 15,000 and $52.50 for the 15,000 above that, about $103 a month for replay alone, on top of whatever the same sessions cost in events. Recording only sessions that reach the cart brings a typical store back under the free allowance, which is the setting most PostHog store customers end up on. Retention is the other cost: on the Free plan a recording from the first of the month is gone by the end of it, and even Pay-as-you-go keeps 90 days, so a recording of last quarter's checkout bug does not exist unless you exported it to JSON. Our PostHog pricing guide works the full bill for two store sizes.

Where PostHog session replay falls short for stores

The product is built for software teams debugging their own app, and that shows in four places. Consent is yours to build: PostHog records as soon as the SDK loads unless you gate it, and under GDPR a replay of a shopper's session is personal data, which the French regulator's 2026 draft guidance says needs consent for optimisation and troubleshooting use. Our guide to session recording and GDPR covers what that gate has to do. Retention is short unless you pay for a platform package. Filtering and navigation draw the most consistent complaints in reviews, because the interface assumes you know which event to search for. And replay does not know where the visitor came from in the sense a store cares about: it shows what happened after the click, not which campaign paid for the click, because PostHog's marketing attribution is an opt-in beta without click ID matching or attribution models.

The alternative built for e-commerce: Arktis

Arktis is the product we make, so weigh this section accordingly. It pairs session recordings with attribution rather than with product analytics. The tracker captures fbclid and gclid on landing, imports Shopify orders and matches Stripe customers, assigns credit with switchable first-touch, last-touch, linear, time-decay and position-based models, and shows the session recording of each attributed buyer next to the campaign that produced the order. Every recording is gated behind explicit visitor consent enforced on the server, so nothing is stored before the shopper agrees, and data is hosted in the EU by default. Recordings are included from the Growth plan at $149 a month for up to 250,000 tracked visitors, with no per-recording meter, and the free plan for 1,000 visitors lets you see the attribution side before paying anything.

Where PostHog is still the right choice: a software product that wants replay inside the same tool as feature flags, experiments and error tracking, or any team comfortably under 5,000 recordings a month. For a store, the recording is only half the question, and the half that matters at budget time is which ad sent the buyer. Start free on Arktis to see attribution on your own traffic, and see the pricing page for the plan with recordings. For the broader comparison, read Arktis vs PostHog and what session recording is.

Sources

PostHog: Session Replay, accessed 9 September 2026

PostHog: Session Replay pricing, accessed 9 September 2026

PostHog docs: Session replay

PostHog docs: How to control which sessions you record

PostHog docs: Session replay data retention

PostHog: We've decided to make less money

Userpilot: PostHog session replay, a detailed review

CNIL, the French data protection authority