Support and tickets
A normal CRM gives you a queue and some statuses.
The queue is the easy part. What costs a support team its credibility is two agents replying to the same customer within a minute of each other, a ticket that should have been three, three that should have been one, and an SLA nobody notices until it has already gone. All four of those are surfaces here: tickets carry live viewer presence including whether somebody is typing, they merge and split, they escalate and auto-route by rule, and the SLA counts down on the ticket with breaches recorded as their own rows.
How it works
The four things a queue does not do
Presence, shape, routing and time.
Collision detection
While a ticket is open the client heartbeats, carrying whether that agent is typing, and the ticket reports back who else is on it. Leaving is explicit rather than inferred from silence. It is a small mechanism that prevents the single most embarrassing failure in shared support.
Merge, split and escalate
A duplicate merges. A ticket that turned out to be three separate problems splits. Anything that needs to go up, escalates. The shape of a ticket is allowed to change, because in real support work it does.
Routing and an SLA that keeps score
Routing rules pick the agent by rule rather than by whoever happens to be watching the queue. SLA policies drive a live countdown on the ticket, and a breach is a stored record with its own type, so “how often do we miss first response” is a question with an answer rather than an impression.
In the product
The queue, with time visible on it.
The mechanism
Presence, routing and the clock, in more detail
Three of the four things above are mechanisms rather than features, and a mechanism is only worth buying if you know what it does at the edges.
Presence is a heartbeat with a two-minute memory
An open ticket beats while an agent has it in front of them, carrying whether that agent is currently typing, and the ticket reports back who else is present. Anyone last seen more than two minutes ago drops out, so a closed laptop stops claiming a ticket instead of holding it indefinitely. Leaving is also an explicit action rather than being inferred from silence, which is what makes the common case — an agent who read it and moved on — clear immediately rather than after the timeout.
Routing is three strategies and a priority order
A routing rule matches on the ticket’s own fields and then assigns by one of three strategies: rotate evenly through a pool, match the ticket’s category to agents carrying that skill, or give it to whoever currently has the fewest open tickets. Rules are evaluated in priority order and the first one that matches wins, so a narrow rule placed above a broad one is how an exception gets expressed.
Two outcomes are refusals rather than fallbacks, and both report themselves. A ticket matching no rule is left alone rather than assigned to a default nobody chose, and a matched rule that finds no available agent assigns nobody rather than picking one anyway. Worth being exact about the trigger as well: routing is run against a ticket, not fired on every ticket the moment it arrives. It decides who, not when.
Two breach types, counted against business hours
A policy sets a first-response target and a resolution target, and those are the two things a breach can be. A breach is stored as a row naming which of the two it was, when the target fell, when it was actually crossed and by how many minutes. That is what turns SLA attainment into something you can group and report on rather than something the team has an impression about.
The clock respects working hours by default, with the week’s hours held as data rather than assumed, so an overnight ticket is not recorded as a failure nobody could have prevented. Policies are matched to a ticket by its priority and the customer’s tier.
One limit belongs on the page rather than in a footnote. A policy can carry an escalation contact and a period, and both are stored and shown — but nothing moves a ticket up on its own when that period lapses. Escalating is an action a person takes, and when they take it the ticket goes to urgent, the assignee is notified, an activity is written to the record and a webhook fires. Everything after the decision is automatic; the decision is not.
What the suggestion will not do
A suggested reply is drawn from the ticket’s own text and the published knowledge-base articles that match it, and an agent sends it or does not — nothing goes out on its own. Suggesting a new article works the other way round and is stricter: it refuses outright unless the ticket has actually been resolved and has comments to draw on. An article manufactured from an unresolved ticket would be a guess written down permanently, and a knowledge base is exactly the wrong place to keep one.
Who it is for
For teams where support and sales share a customer
Which is most of them, and is exactly where a separate helpdesk starts to hurt.
- A services business whose support desk and account managers talk to the same people.
- A support lead who has to report on SLA attainment rather than assert it.
- A small team where two people genuinely can pick up the same ticket.
- Anyone tired of asking a customer for context their CRM already holds.
Related
Questions
The things people actually ask.
How does collision detection work?
An open ticket heartbeats while an agent is viewing it, including whether they are typing, and reports who else is present. Leaving is an explicit action, so presence reflects who is actually there.
Can a ticket be split?
Yes, and merged. Support work arrives in the wrong shape routinely, and a system that cannot change the shape just gets worked around in the notes.
What does the SLA actually do?
Policies drive a countdown visible on the ticket, and breaches are recorded as rows carrying the breach type, which is what makes attainment reportable instead of anecdotal.
Does it suggest replies?
It can suggest a response and the knowledge-base article to attach, drawn from the ticket’s own text. An agent decides what is sent; nothing goes out on its own.
Is this a full helpdesk?
It is a support module inside the CRM, with tickets, SLA, a knowledge base, routing and chat. Its advantage is not feature count — it is that the customer’s deals, calls and history are in the same database as their tickets.
Open the same ticket twice, from two accounts.
The second window should tell you somebody is already there.