Privacy

Last updated: October 30, 2026.

This page explains what happens to data when you use louisbouchard.ai. The site includes optional analytics and A/B experiments, a contact form, public comments, an AI helper you can choose to use, and a few third-party media embeds. Each one handles different information.

Basic website and analytics data

Cloudflare hosts and protects the site, so it can process normal request data such as your IP address, browser, requested page, and security signals.

When allowed, Google Analytics helps identify useful pages and site actions. It can receive device, browser, referral, approximate location, page, and interaction data, and it may use cookies or similar browser storage.

Microsoft Clarity records how pages are used, such as clicks, scrolling, and mouse movement, to produce heatmaps and session replays that help improve the site. Like Google Analytics, it runs without cookies until you choose Allow, and Microsoft advertising storage remains denied regardless of your choice.

Poptin shows the site’s popups, such as an invitation to join the book waitlist. It loads unless you choose Decline and can use its own cookies or similar browser storage, for example to remember which popups you have already seen.

If you choose Allow, the site also loads the X (formerly Twitter) ads pixel. It tells X about your visits and about a completed signup to the book waitlist, so we can measure and improve our ads on X, and X can use these signals to build audiences for those ads. X may use cookies or similar browser storage and can receive device, browser, page, and ad-click data. Until you choose Allow, the X pixel does not load; if you decline later, it stops loading from the next page you open.

A new book waitlist signup also sends Google Analytics a lead event that follows your Google Analytics choice. To count each signup only once, the site’s server keeps a salted, one-way hash of the email address you submitted, which cannot be turned back into the address. The address itself is not stored or logged by the site. Neither Google nor X receives your email address or name from the site: the signup event carries only a random conversion number.

The site also keeps its own aggregate daily counts for page views, engaged views, reading progress, selected actions, outbound clicks, and the number of daily visitors. For page views it additionally counts the referring domain (for example google.com), the utm_source and utm_campaign tags from the link you followed, and a coarse device category, either mobile or desktop. This first-party counting is cookieless and does not depend on your analytics choice: events carry no identifier, and nothing is written to browser storage for it. To avoid counting the same visitor twice in one day, the server combines the network address, browser signature, the date, and a private salt into a one-way hash that changes every day and cannot be linked across days. Raw IP addresses, full referrer addresses, and raw browser details are not stored with these website totals. Preview pages and the private analytics dashboard are excluded.

A/B experiments

The site sometimes compares two versions of visible text or a control, such as a button label. A random experiment assignment in local browser storage keeps the version consistent. The assignment contains the experiment name, variant, and a random identifier. It does not contain your name or email address.

Experiment views and conversions are counted for every visitor, without cookies for measurement and without your name, email address, or any profile. The site stores aggregate daily counts in Cloudflare and keeps a one-way hash of the random assignment to avoid counting the same view or conversion more than once. Raw IP addresses, browser details, and referrers are not stored with experiment results. A temporary hashed network identifier limits automated abuse.

Contact form

The contact form asks for your name, email, topic, and message. Cloudflare Turnstile checks for automated abuse, and the site uses hashed browser and network identifiers for rate limiting. Accepted messages pass through a private Google Apps Script relay, then Google Workspace delivers them to my private inbox.

The site does not intentionally store contact-message contents in its own database. The delivered email remains in the Google Workspace mailbox according to its retention settings. Do not include passwords, API keys, health records, financial details, or other sensitive information.

Public comments

Comments show the name and text you submit publicly on the article. When you submit a comment, the site creates a random browser identifier and uses hashed browser and network identifiers for rate limiting and abuse prevention.

Comment text can be checked by DeepSeek or Gemini for spam, harassment, private information, and unsafe content. Moderation traces for accepted comments or comments held for review can be logged in Opik, a service from Comet, to help investigate failures. Honeypot, too-fast, rate-limited, and moderation-rejected submissions are not logged there. Approved comments remain with the article unless they are removed.

AI helper

The AI helper receives your message, the current page URL and title, recent messages in the same conversation, and a random browser identifier created when you use the helper. Browser and network identifiers are hashed before they are used for rate limiting or logs.

The helper can send this information to DeepSeek as the primary model or Gemini as a fallback. Accepted conversations and technical traces can also be logged in Opik to evaluate answers and investigate failures. Invalid and rate-limited requests are rejected without logging their raw content to Opik. Do not send confidential, personal, or regulated information to the helper.

Browser storage

The site uses local browser storage to remember your light or dark theme, analytics choice, a dismissed announcement, experiment assignments, random helper, comment, and contact identifiers, and the name you last used for a comment. After you join the book waitlist, the email address you submitted stays in that browser tab’s session storage only until the confirmation page checks it, then it is removed. First-party page counting stores nothing in your browser. You can clear this data in your browser settings.

Embedded services and external links

Some pages embed media from services such as YouTube, Spotify, Vimeo, Substack, Embedly, Descript, and Hugging Face. Loading or using an embed lets that provider receive normal request and browser data. External sites follow their own privacy terms.

Retention and your choices

Rate-limit counters stop affecting requests after their minute, ten-minute, hourly, or daily window and are cleared during later activity. Website analytics event and daily visitor hashes are removed after roughly eight days. Aggregate website totals and experiment assignment hashes are retained for up to roughly 400 days, while anonymous aggregate experiment totals can be retained for long-term comparison. Public comments stay with the article until they are removed. Delivered contact emails, Google Analytics, helper, moderation, and hosting data follow the retention settings of the services described above.

Your analytics choice controls Google Analytics and Microsoft Clarity cookies, and whether the X ads pixel and Poptin load. Declining keeps Google limited to cookieless measurements, and you can also block cookies or third-party scripts, clear local storage, or avoid the comments and AI helper. Anonymous first-party page and experiment totals are always collected because they contain no identifier that could single you out. To ask about data you submitted or request that a comment be removed, use the contact form and include enough detail to identify it.