Questions and answers
How the internet gets a page to you, why it's slower in some places, and what websites share about you, answered plainly.
What is a traceroute?
A traceroute lists every router a request passes through on its way to a website, with how long each one took to reply.
It works by sending packets that are allowed to cross one more router each time, so each router in turn reports back. Packet Voyage runs a real traceroute from a probe in the city you choose and draws the result on a globe: each hop is a stop, ocean crossings follow under-sea cables, and routers that don't reply are shown in fog.
How do you work out where each router is?
Mostly from the router's name, then from an IP location database, and every guess is checked against the speed of light.
Backbone routers are often named after the nearest airport (for example a name containing "lon2" is in London). If there's no usable name, an IP database gives a guess, but those often report a network's head office rather than where the box is. Light in fibre covers roughly 200 km per millisecond, so a reply in 3 ms can't have come from 5,000 km away; impossible guesses are thrown out, and the app says which method placed each hop.
Why do some hops show as "in the fog"?
Those routers didn't reply to the traceroute, or gave no clue to where they are.
Many routers are set to ignore or slow down traceroute replies, so a gap doesn't mean the request stopped there. Packet Voyage places a silent hop between its neighbours and labels it honestly rather than inventing a location. A hop that replied very quickly but can't be placed is shown near the start, with the distance its reply time allows.
Why does google.com (or any big site) answer from nearby?
Because big sites use content delivery networks (CDNs) that keep copies of the site in many cities, so your request is answered by the closest one.
That's why a trace to a large site often ends in your own city after a handful of hops. To see a long journey, trace a site that is hosted in one place, such as a university website on another continent.
What is time to first byte (TTFB), and what is a good number?
Time to first byte is how long it takes from starting a request until the first byte of the response arrives; it includes DNS lookup, connecting, TLS setup and the server's own thinking time.
Google's web.dev guidance treats 0.8 seconds or less as good for a full page load in a browser, and more than 1.8 seconds as poor. Packet Voyage measures from probes rather than browsers, without redirects or browser work, so its numbers are usually lower; what matters most is how much they vary between regions.
Why is my website slow in some countries?
Usually because the nearest copy of your site is far away, the page isn't being served from cache there, or every request has to travel back to one origin server.
Distance adds unavoidable delay: a round trip from Europe to Australia takes roughly 250 to 300 ms before your server does any work. A CDN helps only if it has an edge near your visitors and the page is cacheable; cache misses and uncacheable pages go back to the origin every time. Slow DNS can add to it too. Packet Voyage's worldwide check shows which edge answered each city, whether it was a cache hit, and how the time splits between DNS, connecting, TLS and waiting.
What is a CDN edge, and what do cache hit and cache miss mean?
An edge is one of a CDN's locations around the world; a cache hit means it answered from its own stored copy, and a miss means it had to fetch the page from your origin server first.
Hits are fast because the response comes from nearby. Misses are slower, sometimes much slower, because the edge has to go back to the origin. Packet Voyage reads each CDN's own response headers (Vercel, Cloudflare, CloudFront, Fastly, Netlify and the standard Cache-Status header) to show the edge and cache result for every city.
Which companies does a website tell about me before I accept cookies?
Often several: many sites load analytics, advertising and consent-management services as soon as the page opens, before you've clicked anything on the cookie banner.
Packet Voyage opens the page in a real browser as a first-time visitor in London, records every company the page contacts, then clicks the banner's Accept button and records who else joins in. The result separates "before you agreed" from "after". Some early contacts are legitimate, such as the cookie banner itself, fonts or security services.
Is it allowed to load trackers before I give consent?
In the UK, the ICO's guidance under PECR is that a site must get your consent before storing cookies or similar technologies on your device, unless they're strictly necessary for the service you asked for.
So trackers that set cookies or identifiers before consent may be a compliance problem, but not every early request does that, and the rules differ by country. Packet Voyage shows what was contacted, not whether it was lawful. This isn't legal advice.
How accurate are the results?
They are real measurements of one moment from one place, so treat them as a snapshot rather than a verdict.
Routes can change minute to minute and differ between networks. The data check visits from London once, so other countries, logged-in visitors or later page views can see different trackers. Companies are named from a hand-made list, and anything unknown is shown by its domain.
Where do the measurements come from?
Traces and worldwide checks run on Globalping, a free, open network of measurement probes run by jsDelivr; page visits run in headless Chrome on Vercel in London.
Router and server locations come from router names and free IP geolocation services. Under-sea cable routes come from TeleGeography's Submarine Cable Map.
Is Packet Voyage free?
Yes, it's free to use, with fair-use limits.
Each visitor (or API key) can run a handful of checks every 10 minutes, and everyone shares an hourly budget of Globalping probe measurements. When a limit is reached, the app says when you can try again. Saved reports and replays are free to view.
Can I use Packet Voyage from Claude, Cursor or another AI assistant?
Yes: it's an MCP server, so any MCP-capable assistant can run the checks with a free key from the "Use it from your AI" page.
The tools are check_speed_everywhere, who_does_it_tell, trace_route, full_report and get_report. Each returns a plain-text summary the assistant can reason over, plus a link to the visual version.
What do you store about my checks?
Saved reports are kept for 30 days so their links work; your IP address is used only to apply rate limits and expires with them.
Page lookups are cached for an hour so repeat views are instant. API keys are stored only as a one-way hash.