Effective November 12, 2026
Privacy notice
This notice covers free study viewing, the Agent API, Consumer Recreation Demand API access requests and accounts, and commercial data requests submitted to Fron Research.
Information we collect
The Prairie, BOWL360 and Swish Lab studies are free to read without submitting an email. We record study document requests and send traffic alerts to Luke's business inbox. These include the request time, fixed study URL, bounded campaign tags, referring domain and only a safely allowlisted public path when supplied, device and browser class, coarse country and region code when Cloudflare provides them, and whether a recognizable bot, link preview, prefetch or mail security link scanner requested the page. Unknown visitors are not verified human readers. Reloads may count as new requests.
For these alerts, we store the validated request IP address supplied by our hosting platform, when available, with a request event ID. We also store the city and the network operator that our hosting platform associates with that address (usually an internet provider or mobile carrier, or a company that runs its own network), and the user-agent string your browser sends. The event ID helps distinguish repeated delivery of an alert from separate requests. IP addresses may be shared, proxied or change; we do not use them to deduplicate people. Links we send to a specific person or company may carry a reference code. When such a link opens a study, the alert records which recipient we sent that link to. A forwarded link is recorded the same way, so this does not show who actually read the study. We do not store referrer queries, other URL queries, precise location, cookies, fingerprints or a persistent visitor identifier for these alerts, and we do not buy or look up third-party data about visitors. No analytics cookie or persistent tracking ID is added. Historical study requests and access tokens remain separate; any prior access cookie is no longer required to read the study.
We collect the work email, company, organization type, stated use case, delivery preference, timing, and consent that you submit. The application does not store your IP address, user agent, or referring page with the request.
We also store privacy-minimized product events. An event may include an allowlisted event name, normalized page path, channel, campaign, offer, and server time. These event rows do not contain contact details, free text, a raw referrer, IP address, or user agent, and are not linked to a contact request. Financial events are sent only by authenticated server systems. External transaction identifiers are hashed before storage.
The Agent API (/api/agent), which answers questions from AI agents and other programs, records each request so we can see what is asked and improve the answers. A record holds the request time, the question or document requested, the optional statement of why the question is asked (the intent or context field), what the API returned (number of results, the documents and section headings shown with their scores, whether evidence was found), the response status and time, the client name taken from the user-agent string (also kept, cut to 400 characters), the name and version an MCP client gives itself and its protocol version, the referring host name and the coarse country code Cloudflare provides. To tell which agent or company sent a request, it also holds the network the request came from as Cloudflare reports it (the autonomous system number and its organization name, such as a cloud provider or an AI company, never the address), Cloudflare's verified-bot category and bot score when available, the Web Bot Auth signature headers an agent sends to identify itself (Signature-Agent, and whether the request was signed), and these request headers: Accept, Accept-Language, Sec-CH-UA, the Sec-Fetch headers and the HTTP version. Each product site (heatcheck.tech, bowlingdata.co, golfsimdata.com, trampolineparkdata.com and CourtWatch) serves the same API at its own /api/agent and passes the request to it here; the record then also holds which site it came in through, and the network fields above are those Cloudflare reported for the original request at that site. Each search answer carries a random link id, added to every page address it returns as ?via=agent&lid=; when someone opens one of those pages, the page records one visit: the link id, the site and path opened, the time and the host name of the page they came from (such as an AI assistant’s site), then removes the two parameters from the address. That ties a visit to the question that produced it, not to a person. Requests our own checks send are marked internal, and requests that say they are tests are marked as such. Connection set-up messages from MCP clients (connecting, listing tools, keep-alive pings) and refused attempts to open an event stream are not recorded one by one: they are added to an hourly count per site, kind of message, client name, MCP client name, network and country. It does not hold the IP address, cookies or any visitor identifier. To stop one caller from flooding the API, Cloudflare and the API count requests per IP address for about a minute and refuse the excess; that count is kept only in memory and is never stored or logged. Send the Agent API only public questions; do not include personal or confidential information. These records are removed after 400 days.
Anyone can send Fron Research a note through the Agent API (POST /api/agent with "mode": "contact"). We keep the note, the reply email address if one is given, the time, which site it came in through, and the same client and network fields listed above. It does not hold the IP address. We use a note only to read and answer it: it is also posted, with the email address, the site, the network name, the country and the user-agent string, to Fron Research’s own Google Chat space so a person sees it promptly. Notes are removed after 400 days.
When you ask for Consumer Recreation Demand API access on the Consumer Recreation Demand API page, we keep the email address you enter and the time, and we email them to Luke’s business inbox so a person can reply. Nothing else is recorded with the request: no IP address, user agent or referring page. We use the address only to answer the request. Requests without a key to the Consumer Recreation Demand API itself are counted per endpoint and day, without an IP address.
The same form, and the Agent API’s "mode": "key", also give a free API key at once. We mail a six-digit code to the address you enter (the message is sent through Mailgun, which receives the address and the message) and keep, with the address: the time, whether you asked on the page or through the Agent API, a hash of the code (never the code) until it is used or expires after 30 minutes, how many times a code was tried, and, once a key exists, the ids of the account and key that were made. The key itself is shown to you once and is not stored by this site; the Consumer Recreation Demand API keeps only a hash of it. With the key you get a one-time link to set a password for an account at /account/, where your keys are listed and can be revoked; we keep only a hash of that link, for 7 days. We use the address to send the code and to answer you about access.
Consumer Recreation Demand API customers sign in at /account/, by invitation. For an account we keep the email address, a salted hash of the password (never the password), the customer account it belongs to and its role, and when it was created. A session cookie (__Host-fron_account) keeps you signed in for up to 12 hours; we store only a hash of it. Invitation links are stored only as a hash and expire after 7 days. We keep a log of changes made through the account page (who created or revoked an API key, invited or removed a person, or changed a plan, and when); it never holds a key or a password. To slow password guessing, sign-in attempts are counted for 15 minutes against a hash of the request IP address, then deleted. API requests made with a key are counted per key, endpoint and day. The Pro plan is paid by card on Stripe’s own checkout page: Stripe receives your card, billing details and the account’s email address, and we never see the card. From Stripe we keep the customer and subscription identifiers, the subscription’s status, monthly or yearly, and when the paid period ends, to keep the plan in step with your payments. These account records are kept while the account is open and removed on request.
How we use it
We use this information to understand research traffic, evaluate product demand, assess whether our evidence can support the stated use case, respond to qualified requests, and improve our commercial products. Study views are not leads or evidence of buyer qualification. Submitted requests do not subscribe you to a mailing list. We do not sell submitted contact information.
Alerts are asynchronous and access remains free if recording or email delivery fails. To prevent abuse, at most five alerts are queued per minute, 30 per hour and 150 per UTC day. Aggregate counters retain requests exceeding these limits; not every request produces an email. Delivery retries stop after five attempts, with failed/discarded status retained. Duplicate delivery is possible after an interrupted send. Assets, HEAD requests and our own monitoring checks do not generate alerts.
Hosting and retention
Cloudflare hosts the site and request database and may process standard network and security logs under its own terms. We retain submitted requests for up to 24 months unless a longer period is required for security or legal obligations.
Study-view details, including request IP addresses, city, network operator, user-agent strings and event IDs, in the application database are automatically removed after 30 days; aggregate traffic counters after 90 days. Emailed alert copies remain in the business inbox until deleted by its owner. Network or database failures may prevent recording some requests. Browser caching, offline copies and links never fetched from our server are not observable.
Your choices
To access or delete a submission, email luke@fronresearch.com. We will verify your email before acting.