OxalisAdmin

Share a single link on LinkedIn, WhatsApp, or email to let anyone sign up for a time-limited trial account. Trial users land in the shared Trial org with viewer-only access.

Logins by user

Who's actually using it, how often, from where. Trial users get a chip; device column is the (OS · browser) mix seen across their sessions; acquisition is where they came from (newsletter / direct / referrer).

Org adoption

Per-organisation activity. lonely seat = paid for >1 user but only one has logged in. no logins = seat exists but nobody's ever signed in. Trial orgs flagged with a chip.

Acquisition

Where sessions came from. Built from the ref= URL param on newsletter links, UTM tags on share URLs, and the bare HTTP Referer host as a fallback.

Top pages

Raw user activity (legacy)

Inventory counters for the entity-event graph. These were moved off the front-of-house FSI dashboard so it reads as an editorial intelligence brief rather than a bulk-extract preview.

Stripe wiring status. Set STRIPE_SECRET_KEY + STRIPE_WEBHOOK_SECRET, point the Stripe Dashboard webhook at /api/billing/webhook, then bind each plan's Stripe price below. Prices & amounts live in Stripe — here you set the price binding, trial length, and what happens when a trial ends.

Plans & packaging

A plan becomes self-serve sellable once it's active and has a Stripe price ID. Trial days > 0 starts every self-serve checkout with a free trial; auto-renew converts to paid at trial end (card required), cancel ends with no charge.

Offer links (marketing)

Build a shareable link encoding a deal — fixed seats, trial length, and whether a card is required at sign-up. Paste it into outreach. A cold prospect gets a brand-new workspace; a signed-in customer can apply it to theirs. e.g. "5 seats, 14-day trial, auto-bill" vs "10 seats, 30-day trial, no card".

Curate which of the ~41k FCA firms are in scope. target_tier1 + target_tier2 appear in the newsletter and refresh weekly; passive is the default for unknown firms; excluded is dropped from the pipelines entirely. Manual overrides always win over auto-rules. The DB is now the source of truth — target_firms.json is a bootstrap seed only; use Export for snapshots, version control, or restoring a fresh instance.

Backfill from curated list

Reads data/target_firms.json and stamps every matching firm with the right tier. Idempotent. Skips rows where you've set a manual override.

Bulk exclude

Mark a set of firms as excluded so they stop appearing in the newsletter and refresh queues. Always shows a preview first.

Triage queue — biggest untriaged firms first

Firms with no priority row (or computed status of passive) ordered by headcount. Click a button to mark each one in place. The list refreshes after every action so you can rip through the long tail.

Manual overrides

Firms where you've manually set include / exclude. These always win over the auto-rules.

Daily Postgres operational dump (04:00 UTC) and weekly DuckDB snapshot (Mondays 04:15 UTC). Both are GPG-encrypted on the server and rsync'd to the Synology NAS, which POSTs to /api/backups/checkin after each successful pull. A green pulled badge means the off-site copy is fresh; red means the server has the encrypted dump but the NAS hasn't fetched it yet. Read-only view.

Recent runs — newest first

Every server-side 5xx and every client-side JS error lands here, regardless of whether the Discord webhook succeeded. The Discord column shows whether an alert was scheduled (yes = sent or deduped; no = no webhook configured or scheduling failed). Top errors lets you spot the single hot loop causing most of the noise. Auto-pruned after 90 days.

Top errors (by count)

Recent (newest first)

ready.