Usage analytics and privacy

Saved coaching work

For verified accounts, conversations, source references and commitments you save are stored in the app’s D1 database so you can return after logout or on another device. They are available through your account; Mark does not automatically receive them. New conversation preserves earlier conversations. Delete conversation removes its stored content; delete a commitment to remove that plan item. Backend backups may retain earlier copies for the provider’s recovery period. These records are separate from analytics. Password tester conversations and plan changes remain temporary in page memory and are not saved to this database. For longer saved conversations, the coach also receives selected verbatim older user turns (up to eight turns and 8,000 characters), scoped to that conversation’s saved revision. This incomplete working digest excludes assistant suggestions, may miss details, and is recomputed without a separate AI call or stored memory. Recent corrections take priority; please correct outdated context. It never carries information between conversations or changes sharing consent. Selected source excerpts, this older context and recent conversation context are sent to the configured AI provider when you ask for coaching; summary generation sends the conversation you choose to summarize. Email sharing creates a separate copy only after your explicit review and confirmation.

We use content-free usage counts through PostHog and our existing Segment-to-Google Analytics integration. These measure access attempts, completed coaching requests, source opening, commitment saves, brief preparation/copy/download, summaries, booking-link intent and technical failures. Only fixed categories, aggregate count/latency buckets and app configuration fields are allowed. A booking-link click is not a booking.

Prompts, responses, commitments and dates, drafts and edits, names, email/recipient identities, source topics/passages, safety classifications and URL queries/fragments are excluded. Commitment status changes, conversation resets, summary edits and email-sharing actions are not remotely collected; the current version does not store local analytics counters for them either.

Random analytics identifiers stay in page memory, with no cross-visit identity, advertising, recording or automatic button/form tracking. Content-free random attempt IDs may accompany existing sign-in redirects for up to 15 minutes; they do not authenticate you and are removed from the callback URL. Segment requests omit cookies/referrers and specify a zero IP override. Receiving services still see the connecting network address. Clearing this app cannot recall data already received by a service.

Do Not Track and Global Privacy Control suppress collection. The preference below can disable both product analytics paths in this browser. Turning analytics off stops queued and future sends. Other existing product opt-outs remain honored. Conversation and email-sharing permissions are separate and are not changed by this preference.

Return to the site

Tester access records

For password-protected tester access, the site owner records your self-reported email, the app and each visit time in a private table. This does not verify your identity, subscribe you to marketing or record reflections or coaching messages. A protected cookie remembers this browser for 180 days. A return after 30 minutes away counts as a new visit. These first-party access records are separate from optional third-party analytics; cookie expiry does not delete the visit records.