Methodology

How every click is classified.

Many visits to a shared link aren't people. Apps open it to draw a preview, email systems scan it, crawlers index it. In the 30 days to 14 November 2026, 80% of visits to MiniSnip links came from these. This page explains how MiniSnip tells them apart, what it can't, and how click receipts make the result checkable.

The five classes.

People
Visits that passed every check: a real browser, no bot signature, not from a crawler or cloud network.
Shares seen
Link previews. An app such as WhatsApp, Facebook, Telegram or Slack opened the link to draw its preview, which usually means someone posted it.
Email and security scanners
Mail systems and security tools that open a link to check it before a person sees it.
AI crawlers and agents
AI services fetching the link, such as the crawlers of ChatGPT, Claude or Perplexity.
Other automated
Search engines, SEO tools, monitors, scripts, browser prefetching and traffic from cloud servers.

MiniSnip never calls a visit “certified human”. Following the Media Rating Council's guidance on filtered traffic, People means a visit that passed every check below, not a guarantee. Only People are counted as clicks in your dashboard.

The checks, in order.

1. The user agent

Every request names the software that sent it. MiniSnip matches it against more than 140 named bots: link-preview fetchers (WhatsApp, Facebook, Telegram, Slack, iMessage, LinkedIn and others), search engines, AI crawlers, SEO tools, uptime monitors, email security scanners, automated browsers and HTTP libraries. A request with no user agent counts as a script.

2. The request itself

A HEAD request is a link checker, not a visit. Requests marked as prefetch or preview by the browser are counted as prefetching. A request that claims to be Chrome, Safari, Firefox or Edge but sends no language list, or doesn't accept a web page, is a script posing as a browser.

3. The network it came from

Requests from published crawler networks (for example Facebook's, Telegram's and Google's) are matched by IP address. After the visitor has been sent on, MiniSnip looks up the reverse DNS name of the address: crawler hostnames name the bot precisely, and cloud and hosting providers (AWS, Google Cloud, Azure and others) mark traffic from servers rather than people.

4. Everything else counts as people

A visit that passes all three checks is counted under People, with its country, device, browser and source. Sources include the in-app browsers of Instagram, Facebook, TikTok, Telegram, LinkedIn and others, so app traffic isn't lost as Direct.

Each bot visit keeps its name and the reason it was flagged (user agent, request, network or reverse DNS), so you can see why it was left out. Visitor IP addresses are stored only as one-way hashes.

What it can't tell apart.

Click receipts.

Every link has a receipt: the last 30 days of visits split into the five classes above, the apps it was shared in, and where its people came from. Its owner can share it with a private link, for example with a sponsor or a client.

Each receipt is signed with Ed25519 over its payload serialised as JSON with keys sorted at every level. The public keys are published at /.well-known/minisnip-receipt-keys.json, so anyone can check a receipt with any Ed25519 library, or on the verify page. Changing any number or date breaks the signature.

Questions or a bot we're missing? Email support@minisnip.com. Read more in how MiniSnip counts clicks and who opens your link before any person does.