Replayowl vs Highlight.io

	Highlight (self-hosted) / LaunchDarkly	Replayowl
Status	Acquired by LaunchDarkly (announced April 2025). highlight.io, its pricing page and its docs redirect to launchdarkly.com. The repository is not archived and still gets commits	Actively developed
Scope	Session replay, error monitoring, logging, distributed tracing	Web session replay with console errors and search
Hosted option	LaunchDarkly observability. Developer plan free: 5,000 session replays, 5,000 errors, 10M logs and 10M traces a month. Foundation is pay as you go above the same allowances; per-unit rates are not printed on the pricing page	Free for 1,000 sessions a month; Pro $29 per month for 10,000; Business $149 per month for 100,000
Self-hosting	Hobby edition with Docker on Linux; an enterprise self-hosted edition under a separate license	Free, no session limits
Services to self-host	Backend, frontend, Postgres (pgvector), ClickHouse, Redis, Kafka, Zookeeper and an OpenTelemetry collector	One Cloudflare Worker with a bucket and D1 or Postgres; or one Node process with Postgres
Recommended hardware	At least 8 GB RAM, 4 CPUs and 64 GB disk for the hobby edition	No published minimum; a Worker has no server to size
Stated capacity	Hobby: under 10,000 sessions and 50,000 errors a month	No built-in limit when self-hosted
License	Apache-2.0 by default; enterprise/ and highlight.io/ folders under their own licenses	AGPL-3.0
Recorder	Based on rrweb (Highlight maintains a fork)	Its own recorder, about 8 KB gzipped, no dependencies
SDKs	Browser plus many backends (Go, Node, Python, Ruby, Java, .NET, PHP, Rust, Elixir and more)	One script tag for the browser
Error tracking	Yes, with grouping and alerts	Console errors and uncaught exceptions with stack traces on the session timeline; no grouping or alerts
Logs and traces	Yes	No
Retention	LaunchDarkly: one month on the published plans. Self-hosted: not stated in the README	Hosted: 30, 90 or 365 days by plan. Self-hosted: until you delete it
Where data lives	Self-hosted: your servers. LaunchDarkly: LaunchDarkly’s cloud	Self-hosted: your bucket and your database. Hosted: our storage, or your own bucket on Pro and Business

Sources: highlight/highlight on GitHub, Highlight license, docker/compose.yml, LaunchDarkly acquires Highlight, LaunchDarkly pricing, LaunchDarkly billing docs.

What happened to Highlight

LaunchDarkly announced it was acquiring Highlight in April 2025 and now sells session replay, errors, logs and traces as part of its own platform, through its observability and session replay SDK plugins. The highlight.io website, pricing page and documentation now redirect to launchdarkly.com. The open source repository is still there: on 20 August 2026 a commit redirected the landing page, and there were commits on 1 October and 21 October 2026, including a Next.js and Node upgrade. The most recent tagged Docker release is docker-v0.5.6, from August 2025.

The practical effect for self-hosters is that the code is maintained at a low pace, and the documentation the README links to for self-hosting no longer loads. We cover the options in more detail on moving off Highlight.

What self-hosting each one takes

Highlight’s hobby edition is started with cd docker && ./run-hobby.sh on a Linux machine with Docker. The script starts the infrastructure first, then the backend and frontend containers. The infrastructure in docker/compose.yml is Postgres with pgvector, ClickHouse, Redis, Kafka with Zookeeper, and an OpenTelemetry collector. The README recommends at least 8 GB of RAM, 4 CPUs and 64 GB of disk, and describes the hobby edition as good for fewer than 10,000 sessions and 50,000 errors a month. Beyond that, Highlight pointed to its enterprise self-hosted edition, which is under a separate license that requires agreeing to Highlight’s subscription terms for production use.

That stack exists because Highlight ingests logs and traces as well as replays, and ClickHouse and Kafka are what make that searchable at volume. If you use those features, the weight is justified. If you only open Highlight to watch replays, you are running all of it for one feature.

Replayowl runs as one Cloudflare Worker with an S3-compatible bucket (such as R2) and a D1 or Postgres database, or as one Node process with Postgres, writing recordings to disk or a bucket. There are no containers to keep running on Workers, and tables are created on the first request. Recordings are plain gzipped JSON files in your bucket, so backing up and exporting use the tools you already have. See the Cloudflare Workers guide or the Node guide.

The self-hosted Replayowl has gaps a Highlight self-hoster may notice: it has a single admin password, with no multiple users, roles or single sign-on, and no automatic retention limit (you can use bucket lifecycle rules and a scheduled delete instead). The hosted Replayowl service has team members with roles and retention by plan.

Session replay, feature by feature

• Recorder. Highlight’s replay is powered by rrweb, through Highlight’s own fork. Replayowl uses its own recorder, about 8 KB gzipped with no runtime dependencies. It captures same-origin stylesheets into the snapshot, CSS-in-JS rules written with insertRule, constructed stylesheets and open shadow roots.
• What sits next to the replay. Highlight shows network requests and console logs with the replay, and links replays to errors, logs and traces from your backend SDKs. Replayowl records console.error, console.warn, uncaught exceptions and unhandled promise rejections with stack traces. It does not record network requests and has no backend SDKs.
• Privacy. Replayowl masks all inputs, email addresses and runs of 12 to 19 digits in the browser by default, with data-ro-mask, data-ro-block and data-ro-unmask for individual elements. See masking and privacy.
• Finding sessions. Replayowl filters by user ID, any page visited, date, duration, page count, errors, rage clicks and device. See search.
• Not in Replayowl. iframe contents, canvas contents, heatmaps and recording only when an error happens. The full list is on limitations.

Pick Highlight (or LaunchDarkly) if

You use errors, logs and traces as well as replay and want them connected in one product. If you are happy with a hosted service, LaunchDarkly’s free Developer plan includes 5,000 replays and 5,000 errors a month and is the maintained successor to Highlight’s cloud. If you already run the self-hosted Highlight stack and it does what you need, it keeps working; weigh how much you depend on future fixes from a repository that now gets occasional commits.

Pick Replayowl if

You mainly used Highlight to see what a visitor did before something broke, and you want that on infrastructure you control without running ClickHouse, Kafka and Redis. Replayowl keeps console errors and stack traces on the session timeline, which covers the part of error tracking most teams use replay for. For alerting, grouping, logs and traces, pair it with a dedicated tool.

Moving over

Replayowl does not import Highlight recordings. Run both for a while: add the Replayowl script tag next to Highlight’s snippet, pass the same user ID to both with ReplayOwl.identify(), and compare replays of the same visits. When you are happy, remove Highlight’s recorder and keep the old deployment until you no longer need its recordings.

Compare replay quality in the live demo, read the deployment guide, and check what Replayowl does not do before deciding.