Reading your account online.
The console is where history outlives any one machine, where a report becomes a link you can send, and where you see what your provider can read, one line at a time. Your own provider sees your line only once it has proved it runs your network, and one switch turns that off. This is what each screen means and what each control actually does.
What the account adds
Nothing you need to measure or diagnose. The agent does all of that with no account at all, and will keep doing it if you never create one. What an account adds is history across devices, a way to publish a report as a link, and the ability to share a line with a provider.
Signing a computer in to your account is what brings its measurements here. Sharing with your provider is a separate switch, on purpose: your own provider sees your line only once it has proved it runs your network, one switch turns that off, and nothing but you ever turns it back on.
Your connection
The same question the desktop answers, aggregated across every device you have running rather than just the machine in front of you.
- Latency, slow tail, jitter, loss, worst run — the five figures that decide quality, each with the sample behind it. Loss shows its counts always, and a rate too once 300 probes stand behind it.
- Good for, right now — calls, gaming and streaming as a word, a number and a colour together, never colour alone.
One deliberate difference from the app: the console does not apply the bufferbloat penalty, because there is no meaningful average of a loaded-latency test across a whole account. The panel says so on its face rather than letting two surfaces quietly disagree about one line.
How your line compares
Your median latency to common destinations beside the median for your country and for everyone running Helix — with the number of contributing subscribers next to each figure.
“Not enough data yet” is a real answer. Below a minimum number of distinct contributors a cell is absent rather than shown with a small sample or anonymised, because a rating built from two people is not a rating of a network. Withheld cells are counted rather than hidden, so you can tell “nobody on this ISP uses Helix” from “something is broken”.
These are other Helix subscribers, not a survey and not a representative sample of the country — the page says so where it shows the figures.
Devices
Every agent reporting into your account, when each was last seen, and which connection each is measuring. Removing a device stops it uploading; it does not stop it measuring locally.
Reports and public links
A diagnosis saved on the desktop can be uploaded to your account. From the console you can publish it as a link to send to a support desk.
Uploading and publishing are different acts. Uploading is private and undoable. Publishing mints a public URL, so it is only ever done by you, signed in, from here — never by the agent on your PC. A published link can be revoked, and once revoked it answers exactly the same way as a link that never existed.
The page carries no tracking and is not indexed by search engines.
Speed-test history
Every speed test your agents have run, kept rather than overwritten. Sharing one freezes the result into the shared copy rather than pointing at live data, so a link somebody was sent keeps showing what was sent. It names your ISP, because that is the point of sharing a result, and nothing else about your account.
ISP links and sharing
The most important screen in the console, and worth reading slowly.
- A link records that a provider is associated with one of your connections. Your own provider sees your line only once it has proved it runs your network.
- Sharing is a switch on each link. When one of your computers uploads from a provider that has verified with Helix, a link to that provider is made for you with sharing on. One switch turns it off, and Helix never turns it back on for you. A link you add yourself starts with sharing off.
- While sharing is on, the provider’s support desk can see that line’s quality history — latency, jitter, loss counts, outages, speed tests and path traces.
- Never shared with a provider, in any state: which applications used the network, which sites you visited, or anything about your traffic’s content. Which applications used the network never leaves your computer at all. The one list drawn from it that does — once a day, the services the computer uses most: a name where one is known, otherwise only the address’s network block, with nothing attached — goes to Helix and never to a provider. It is on by default, and the app’s Settings → Privacy & sharing has its own switch for it.
- Revocation bites on the next query. Nothing is cached on their side; the next thing they ask returns nothing. There is no stale copy to expire.
When a provider asks
A provider can assert that a service ID belongs to you, which arrives as a claim — through their channel, on their invoice or SMS, because Helix holds no contact list for other companies’ customers.
A claim grants nothing. It is one half of an assertion. Only your acceptance creates the link, and accepting asks you to pick which of your lines is theirs, because that choice is the scope of what they can see.
The screen shows the organisation’s registered and trading names and their published support contacts specifically so you can check out-of-band before accepting. If a claim surprises you, verify through a number you already had — never through one in the message that surprised you. Declining tells the provider the code was used and declined, and nothing about you.
Tickets
A support conversation with a provider you are linked to, where evidence can be attached rather than described. The ticket form can attach a diagnosis report you have sent to your account.
What your provider gets from an attached report: if they take tickets by email, the notice carries its headline and verdict, plus a link to the full report if you published it before raising the ticket. That link stops working once you revoke it, or after its 30 days. Their portal shows only the report’s reference, so if they need the full report, publish it and send them the link.
If your provider escalates to their own upstream carrier, what crosses is a frozen, redacted copy made at the moment of escalation, not a live view. It cannot become a back door into current data, and the carrier never sees your account, nor a stable identifier that would let them join your tickets into a history.
Retention and tiers
Tiers decide how far back you can read, not how long data is kept: Free reads back 3 days, Pro 30 and Enterprise 45. The stored window is the largest tier’s, for everyone — measurements are kept for 45 days. Upgrading therefore reveals history that was there all along rather than starting a new recording — and downgrading does not delete anything.
Leaving
Deleting your account removes the row rather than flagging it hidden, and takes the measurements with it. Read-time filtering would leave your identity in the database for any export or debugging endpoint to surface; deletion does not.
Terms defined in the glossary. For the app itself, see reading the desktop app. What is collected and what is not is set out on the privacy page.