See who's on your website

A strong buying signal is that they came to you. Put one small script on your own website and Diurno HQ watches for visits from the networks of the companies you track — each match becomes a scored, first-party signal on that account, feeding the same brief, queue, and alerts as everything else.

1. 1Turn it onSettingsIntegrationsYour websiteTurn on (owner/admin). You get a one-line snippet (always an https:// address).
2. 2Paste the snippet on your siteOne script tag, anywhere before </body> — most site builders (Webflow, Framer, WordPress, Squarespace) have a “custom code” box for exactly this. Nothing else to configure.
3. 3Watch matches landThe card shows the last 7 days: how many visitors were matched to a tracked company's network and how many visitors there were in total, a table of the matched companies (with pages viewed and the last page), and when the snippet last phoned in. A visitor is counted once per day, so one network that comes back on three days counts three times. A matched visit appears on the account as a “Visited your website” signal — one per day, refreshed as they browse, adding up all of that account's visitors that day — and raises the Event score via a scoring rule you can tune like any other.

How matching works — and where it stops

• A confirmed network name that belongs to the account. A visit matches only when the visitor's address has a reverse-DNS name that is confirmed — the name resolves back to the same address, because whoever controls an address block can write any reverse text they like — and that name's domain is the account's domain (vpn3.acme.com → acme.com), or the account's site is a subdomain of it. If an address carries reverse names that belong to several different companies (a shared host), nobody is matched. Matching compares domains; it doesn't use fuzzy name matching or probabilistic “identity graphs.”
• A lookup can be late or missing. A reverse-DNS lookup gets about a second and a half. If it times out, or the system declines to start one because many lookups are already running or because earlier lookups for that network were slow, that page view stays unmatched and the visitor is looked up again on their next page view.
• It identifies a network, not a person. The signal says someone on that company's network viewed your site, not that a particular employee did: an office, a VPN or a guest network can be anyone.
• Shared networks are skipped. Home broadband, mobile carriers, cloud providers, job boards, site builders and social sites are on a maintained list and stay anonymous counts. The list isn't exhaustive, so a shared network that isn't on it can still be read as a company.
• What is stored. A salted one-way hash of the visitor's network — the IPv4 address, or the IPv6 /64 network, so one household or site counts as one visitor however many temporary addresses it uses (different in every workspace; the address itself isn't kept) — which means “visitors” are networks, not devices or people. Also stored: the network's registrable domain (not the full reverse-DNS name, which on consumer networks can embed the address), page and referrer addresses cut to origin and path (no query string or fragment), and the date and counts. Visitor rows are deleted after 90 days by default (the “Visited your website” signals they produced stay). The visitor's address is the one our hosting proxy saw, not a header the browser can set.
• Older visit records. The shortened form above applies to visits recorded by this version. Records written by earlier versions may still hold the longer network name and the full referrer until the nightly clean-up rewrites them to the same shortened form; it works in batches, so a large backlog can take more than one night. Every visitor row is deleted after the retention period either way.
• Safety ceilings. New visitors are counted up to 1,000 an hour from one network block (an IPv4 /24 or an IPv6 /48) and 20,000 an hour per workspace; past a ceiling, further new visitors (from that block, or in that workspace) aren't recorded until the hour is up. A business website shouldn't reach them.
• Expect partial coverage. Only office networks and VPNs resolve to company domains, so most visitors stay anonymous counts — that's the honest ceiling of this technique, whatever other tools imply.

Email opens and clicks

The same first-party idea applies to email you send from Diurno HQ, once the service's tracking is switched on: outbound messages carry an invisible open beacon, and links route through a signed redirect (the visible link text is the real destination, and the recipient always reaches it). An open shows as a quiet chip only: mail apps auto-open messages, so Diurno HQ displays opens but doesn't score them. A click is scored and shows as a green Clicked chip on the Sent tab— unless it looks automated. Security scanners and link previews fetch every link in a message within seconds of delivery, so a request is treated as automated when it comes from a known scanner, bot or HTTP-library user agent (or none), is a prefetch, arrives within 20 seconds of the send, or is one of several different links fetched together. Those are recorded as “Email link fetched automatically (not counted as a click)” with no points and don't set Clicked. It's a heuristic: a gateway that behaves exactly like a browser can still be counted, and the next genuine click is counted even if an earlier one was recognized as automated. Words such as bot, crawler or preview in a user agent count only as whole words or in the standard Name-bot/1.0 form, so a phone whose model name merely contains the letters “bot” isn't mistaken for a scanner.