Somebody mentioned it in a thread.
It never became a ticket.

The people who use an internal tool are close enough to tell you about a fault in passing, and that closeness is exactly why it is never filed. By the time it comes up they have found a way around it, and the workaround has become the process.

The problem

Faults in internal software get mentioned rather than reported. They surface as an aside at the end of a call, a message in a channel, a shrug from somebody demonstrating their own job to you — and each one is raised by a person who has already stopped being blocked by it. Very few of those become an entry on your board, and the ones that do arrive three retellings later with the specifics worn off. What that leaves behind is a backlog shaped by familiarity: the teams comfortable opening a ticket are over-represented, and the screens that hurt most tend to belong to the departments least likely to have ever logged into your tracker. Meanwhile the number of people who could usefully report something is the company headcount, which is a difficult number to pay per head for on software that generates no revenue of its own.

How Snag helps

Reporting moves into the tool, at the moment and on the screen where it goes wrong. Whoever is looking at the misbehaving grid opens the capture from that page — no second tab, no tracker to learn, no login they have never had — and what they send is the session rather than a recollection of it. That removes the two things that actually stop internal faults being raised: the context switch, and the sense that this is somebody else's job to write up. Because nobody is licensed to file, the coverage widens to the people you never hear from — warehouse, finance, field staff, the person covering a desk for a fortnight — and reading what arrives costs nothing either, since unlimited members on every plan. Where the widget appears is your decision per screen, and it can render nothing at all until one of your own controls calls it.

What you get back

A queue made of the screens people actually struggle with, each entry carrying the recorded session underneath it, so prioritising becomes a matter of watching rather than of interviewing. On a fix plan the change comes back as a pull request in the repository you connected, with the evidence it worked from attached. One honest boundary: this does not surface the workarounds nobody ever mentions — you still only hear about what somebody chose to raise. What it changes is how much that choice costs them, which for internal software is most of the reason the choice goes the other way.

Questions people actually ask

Our tool sits behind single sign-on. Does that complicate it?

Not really — the snippet is part of the application, so whoever is reporting is already through your sign-on and already on the screen in question. The capture is started by them and ends when they submit it. What does need attention is the address: recording only happens on domains you have verified, and verification is done by publishing a DNS record, so the tool needs to live on a name your organisation owns rather than on a hostname that exists only inside the network.

Our admin screens display customer records. What ends up in a recording?

What people type does not: input values are masked in the browser, sensitive headers are stripped and obvious secrets redacted, all before anything is uploaded. What the page displays is a different question, and the honest answer is that a replay reconstructs the page, so text rendered on screen is part of it. There is no per-region exclusion to configure today. If a screen in your tool shows records that should not be recorded, look at a capture of that screen before you put the snippet on it.

Can we put it on some screens and not others?

Yes, in two ways. The snippet only does anything on the pages you include it on, and its quiet mode renders nothing at all until one of your own controls opens it — so a fiddly workflow screen can carry a visible report-a-problem item while a settings page carries nothing.

Where these details come from

This page makes no claim about another product — every detail on it is Snag's own. Something out of date? Email admin@dvcllc.io and we will fix it.