Offen Fair Web Analytics vs Stormly

Opt-in only collection with first-party cookies

No analytics events are recorded until the visitor actively consents. Cookies are first-party and the tracker is meant to run on a same-site subdomain so third-party cookie restrictions and cross-site tracking models do not apply. Visitors who never opt in leave no usage trail—expect lower absolute volumes than cookieless tools that measure by default.

Browser-side end-to-end encryption of usage events

Clients encrypt usage data before it leaves the browser; the server stores ciphertext and cannot decrypt events alone. Only the visitor (via their cookie) and the matching operator account can open that visitor's data in the Auditorium. Practical impact: a compromised database or overly broad ops access does not yield plaintext browsing histories the way a typical self-hosted analytics DB would.

Auditorium for both operators and visitors

Operators see aggregates across pages where the installation is active (unique users/sessions, top pages, filters). Each opted-in visitor can open the same style of UI for their own data only, with plain-language metric explanations, and can delete data or fully opt out later. This is the fair design point: measurement is not a one-way glass.

Essential metrics without IP or User-Agent capture

Dashboards cover real-time activity, page views, unique users and sessions, bounce rate, returning users, top pages, referrers, UTM campaign/source, landing and exit pages, weekly retention, and load time. Location is country-level from timezone mapping; mobile share uses orientation capability—not IP geolocation or UA parsing. No heatmaps, session replay, or warehouse-grade product analytics.

Lightweight self-host: binary, Docker, SQLite or SQL

Production installs use a single binary (Linux/Windows/macOS) or the offen/offen image; docs also cover Heroku, Uberspace, and YunoHost. Default store is SQLite; MySQL and Postgres are supported. AutoTLS can request Let's Encrypt certificates. Config is environment variables or offen.env. You own uptime, backups, SMTP for password reset, and the subdomain layout.

Multi-site accounts, teams, and short retention

One installation can cover multiple websites with shared team access. Default retention is six months with automatic deletion; operators can shorten retention (e.g. 12 weeks, 30 days, 7 days) knowing shorter values purge older events on startup. Consent banner appearance is customizable; UI locales include EN, DE, FR, ES, PT, and VI.

AI agent for trends, anomalies, and root-cause style digs

Stormly positions an AI agent that watches connected e-commerce data for trends and anomalies (for example conversion drops or product spikes), then delivers plain-language findings—often to the inbox—with root-cause style exploration for questions like mobile conversion drops or regional return spikes. AI assistant access is plan-dependent; the public DPA documents Azure OpenAI (Microsoft) handling of assistant queries with a 30-day retention window and property stripping rules. Best for merchandising and growth leads who want narrative answers without waiting on a data team.

SKU-aware e-commerce report library (plus custom builds)

Ready reports target commerce questions: New Arrivals Performance (early SKU winners/laggards), Unviewed & Unsold Products, cart abandonment and revenue contribution by product/category, Conversion Journey and funnels by device/region, A/B test insights, Aha Moment Discovery, cross-selling analysis, and broader library items (retention, CLV, forecasting, SQL report, and more). Vendor materials state custom reports are included without a separate fee when a needed view is missing—useful for mid-market retailers that outgrow generic event charts.

Shopify, Adobe Commerce, Segment, and GTM-oriented connections

Homepage setup emphasizes linking Shopify, Adobe Commerce, or Segment quickly, with Google Tag Manager also listed among integrations. The DPA describes additional programmatic paths (JS library, import, custom endpoints, third-party sources). Fits stores already on common commerce stacks; teams on exotic storefronts should validate event coverage and product-feed depth in a pilot before replacing an existing analytics stack.

End-user tracking with IP anonymization defaults

Per the DPA, IP addresses for end-user analytics are not stored in full: the last octet is removed and geo/timezone are derived from the anonymized value. Web integrations automatically capture first-party cookie user ids, page/referrer/UTM context, and device metadata; app properties depend on client or CDP configuration. This is still full product analytics (not cookieless aggregate-only web stats)—clients remain responsible for consent banners and lawful basis for shopper tracking.

Public DPA, security architecture, and Dutch contracting

B2B buyers can download a Data Processing Agreement (effective 2023-09-06) and a Security Architecture page listing subprocessors, encryption expectations, backup retention (analytics backups up to six months), and logical tenant isolation. Contracts and DPA are governed by Dutch law with Amsterdam courts. Useful for EU procurement packages—but US-group subprocessors (AWS, Vultr, Azure OpenAI) still require transfer and CLOUD Act diligence.