An email lands at billing@ at twenty past eight on a Monday. Before anybody has the kettle on, it is sitting with the accounts person with a billing tag on it, and nobody has read so much as a line of it. That is a ticket routing rule doing its job, one line of if-this-then-that the help desk runs on every new email as it comes in. A small desk wants about a dozen of them. We have written ours out further down, in the order we would run them, since the order turns out to matter nearly as much as the rules do.
When one of them sends a mail to the wrong person, the logic is hardly ever what went wrong, funnily enough. It is nearly always the field the rule was told to look at, and more often than not that was the subject line.
Why is the subject line the worst field to route on
Wednesday, ten to nine, and a customer cannot get logged in. Writing you a fresh email about it is the last thing on their mind. They go into Gmail or Outlook and find the last mail you sent them, which was their March invoice as it happens, and they hit reply, on account of it sitting at the top of their inbox with your address on it already. So the subject reads Re: Your invoice for March, and your billing rule, which looks for the word invoice in the subject, does exactly what it was told and sends a login fault to the accounts person. It happens most weeks.
Nobody went and did anything wrong there. The rule worked. The customer used their email the way everybody uses email, mind you, and the subject line simply stopped describing what was underneath it the moment they pressed reply.
Scroll back through a month of subject lines on any desk. You will find Help, and Question, and Urgent!! a few times over, and a good many with nothing in them at all. There will be one in Spanish. Somebody will have put the name of a product you stopped selling two years ago, and somebody else a ticket number out of Jira that means a great deal to your engineers and nothing whatever to a rule. Some of it is useful enough. A good share of it is noise all the same, and the rule cannot tell which is which by its self.
Other fields hold up a good deal better, mostly the ones a customer does not stop to think about. Somebody who mails billing@ rather than support@ has told you what the mail is about before writing a word of it, and a second address costs nothing to add on Google Workspace or Microsoft 365, so it is about the cheapest routing you will ever set up. The domain helps too, at least when people write from work, since it says which company is at the other end, and your list of key accounts is probably sitting in a CRM already. Whether a mail is a machine talking to a machine, or a customer reopening an old conversation, you can tell from the sender and the status without anybody having phrased a thing well.
Then there is the mail your own systems send, and keyword rules go further astray on that than on anything. Xero or QuickBooks sends invoices out in your name, and every reply to one of those comes back with invoice in the subject whatever the customer is actually writing about. Stripe sends its payout and dispute notices to whichever address it was given years ago, and that is very often support@, and Shopify does much the same with new orders. Add whatever Zapier was set up to forward before anybody can remember why. None of it is a customer asking for anything, mind you, and the lot of it wants a rule that knows the sender and hands it to whoever owns that system before a keyword rule gets near it.
What a week of routing looks like when somebody counts it
Take a made-up week of 200 mails at a small software company, split across two addresses, support@ and billing@, that both land in the one desk. Every figure in it is invented for the example, though we have not made any of them flattering.
Thirty eight came in to billing@. The rule on that address sent all 38 to the billing group, and 3 of them turned out to be something else, a login fault and two feature questions from people who had billing@ saved in their contacts and went and used it for everything.
The other 162 came in to support@. A rule on sender domain tagged 22 of them as coming from the four biggest customers, without moving any of them anywhere. Another 17 were out-of-office replies, bounced mail and messages from no-reply addresses, and a rule closed those off, which is 17 mails nobody on the team ever had to open. Six were the company’s own Stripe and Xero notices. Those went straight to whoever set the two systems up, which on a desk that size is usually the owner. The keyword rules picked up 58 off the subject line. Of those, 10 went to the wrong place, mostly people replying to an old thread with a brand new question in it.
That leaves 81 that no rule matched at all, about two in every five, and not one of them because anybody went and wrote a bad rule. A few asked two things in the one mail. One was a supplier chasing a delivery note, and one was somebody looking for a job, mind you. The whole lot went to one named person that week.
Put the two next to each other and the address got it wrong 3 times out of 38, a little under 8%, while the subject line got it wrong 10 times out of 58, near enough 17%. Which is the case for letting the address and the sender decide things early and leaving the words in the subject till the end of the list, where a misfire does the least damage.
Twelve routing rules you can copy
| # | When | If | Then | Why it earns its place |
|---|---|---|---|---|
| 1 | An email arrives | It was sent to billing@ | Assign to the Billing group; tag billing | The customer chose the address, so it misroutes least |
| 2 | An email arrives | The sender is one of your own systems: your invoicing, payment or shop platform | Tag system-mail; assign to whoever owns that system | Stops invoice copies and payout notices tripping the keyword rules |
| 3 | An email arrives | The sender’s domain is on your key-accounts list | Tag key-account; set priority High | Tags and prioritises without moving the mail, so it never fights an assigning rule |
| 4 | An email arrives | The sender is mailer-daemon, postmaster or a no-reply address, or the subject starts Automatic reply or Out of office | Tag auto-reply; close | Machines answering machines never need a person |
| 5 | An email arrives | The subject contains security, vulnerability or phishing | Assign to one named person; set priority Urgent; send an alert | A security report skips every queue |
| 6 | An email arrives | The subject contains data request, delete my data, subject access or GDPR | Assign to your named data-protection owner; set priority High; tag gdpr | GDPR gives you 1 month to answer a data request |
| 7 | An email arrives | The subject contains down, not loading, cannot log in, SSO or error | Set priority Urgent; tag possible-outage | Priority only, so it never fights the rule that assigns |
| 8 | An email arrives | The subject contains cancel, cancellation or close my account | Assign to the account owner; tag cancellation | The dearest mail to leave sitting in a general pile |
| 9 | An email arrives | It was sent to support@ and the subject contains invoice, refund or charged | Assign to the Billing group; tag billing | Keyword fallback for people who did not use billing@; expect misfires on old reply threads |
| 10 | A status changes | A closed conversation is reopened by the customer | Tag reopened; set priority High | A second contact about the same thing says the first answer did not land |
| 11 | Time passes | No reply from your side for 1 day | Notify the owner | A quiet conversation is the one nobody is thinking about |
| 12 | Time passes | Still unassigned 30 minutes after arriving | Assign to this week’s catch-all person | Puts a name on whatever no other rule claimed |
Copy the words in the keyword rules from your own mail rather than from ours, since your customers have their own words for things and a list of ours will not have them.
What happens when two rules match the same email
Now and then one mail suits two of your rules at once. A cancellation from one of the key accounts, say, or somebody writing to billing@ with the word security in the subject. Both turn up often enough.
What happens next depends on the tool, and plenty of people running a desk could not tell you off the top of their head which way theirs goes, to be honest with you. In some the first rule that matches wins and the rest are ignored, while others give every matching rule its turn, from the top of the list to the bottom. So the security rule hands the mail to your developer and then the cancellation rule comes along after it and gives it to the account owner, and nobody ever sees the first move happen. You find out which kind you have on a quiet afternoon, with two rules that overlap and one email sent from your own Gmail with both words in the subject.
What saves you is that most rules never need to move a mail anywhere at all. The key-account rule only sticks a tag on and bumps the priority, the outage rule does much the same, and the pair of them can land on the one email alongside anything else without a bother, since the mail stays exactly where it was. Only one rule should get to assign any given email to a person or a group, so the assigning rules want writing so they do not overlap, with the narrowest first where they must.
On most desks that puts the security rule above the billing one, and anything going to a named person above anything going to a group, since four people in a group can each quite reasonably assume one of the other three has it. Our help desk automation page goes into that bit, and into why we would switch a new set of rules on two or three at a time rather than all twelve in one afternoon.
Why the last rule matters more than the first eleven
Every desk has a pile of mail no rule was ever going to catch. In the made-up week it was 81 out of 200. With no routing at all it is the whole of the inbox, and adding forty rules makes it smaller and harder to see without ever quite making it go away.
So the last rule on the list just puts a name on it. This week the name is whoever drew the short straw on Monday. Anything still sitting there with no owner half an hour after it came in goes to them, and they answer it or walk it over to whoever it belongs to. It is dull, plainly said. No keyword could do it for them either, on account of the mail in that pile being the very mail that never said what it was about.
Somebody different gets it the next week, and on the Friday whoever had it goes and writes down how many came their way. It was 81 one week and 70 the next, which means the other rules are picking up more. When the number jumps, it is nearly always customers writing about something new, a price change in the spring, say, or a feature that went out on a Tuesday. You hear about it from a number on a Friday instead of from somebody saying it out loud a month later. Without the rule those same mails sit in a shared view with no name on them until somebody scrolls down far enough, and on a busy Tuesday nobody does.
The quiet-conversation nudge is the same idea from the other end. A mail can go to exactly the right person and still sit there, because the right person went off sick on the Thursday, and after a day of nothing the rule taps the owner on the shoulder, which beats the customer doing it, or the thing turning into an SLA problem and then a proper escalation.
What do these rules look like in Maxdesk
The billing@ one takes a minute or so over here at Maxdesk. An email arriving is what sets it off, the match is anything sent to billing@, and what it does is hand the mail to the billing group with a billing tag on it. You click the whole of it together without a line of code. The other eleven go together the very same way, since our rules read nothing fancier than the plain fields this page has been going on about, and as well as assigning and tagging, a rule can set a priority, send an alert or close a mail off. None of them writes a reply.
By the time anybody opens the shared inbox on a Monday the sorting is done, and the tags the rules left behind are still on every mail for searching later, which is handier than it sounds, truth be told. Search the auto-reply tag at the end of a month and you have a count of the mail that was machines talking to machines. On the made-up desk that was 17 a week, near enough 70 in a month that nobody had to open. The reopened tag is the more uncomfortable search of the two, since it turns up every answer that did not settle the thing first time, and a desk that sorts by hand has no way of pulling either list at all.
If you have already built support tiers, the tiers become groups here and the rules do the moving between them. Anyone working from a priority matrix will find the security, data-request and outage rules are where it goes in, though the matrix its self is still the thing that decides what the priorities mean.
Our pricing counts workspaces, never people. On this particular subject that means the catch-all duty can go round one and all, including whoever in accounts only looks in on a Monday, without a single one of them costing a seat.
The routing we do not do
On the Free and Pro plans, what our rules do is match words and fields, the address and the sender and the subject and the rest. They do not work out what a customer meant, so the Wednesday login fault in its invoice thread goes wherever its subject sends it, the very same as it would anywhere else. On the Elite plan our triage agent reads the issue its self, sets the priority and a category off what the mail actually says, and assigns by how many open tickets each person has live. That is load balancing rather than skill. Skill-based routing is not something we do on any plan, and tiers are not a built-in thing either, so you make them out of groups, tags and rules.
The free workspace also has a ceiling on how many rules can run together. You may well write all twelve and find you can only have some of them going until you move up a plan. If it comes to that, we would have the billing@ rule on first, then the two for machine mail and the catch-all, since on the made-up week those four between them dealt with 142 of the 200 mails.
And we handle email and nothing else. A desk taking chat or phone calls needs routing for those as well, and it will not get any bit of it from us.
One limit is not really ours, truth be told. Say two people are on the desk and ninety mails came in overnight. The rules will have every one of them in front of the right person by nine, and the pair of them will still be answering at four. That is a staffing question, and nothing we could build goes and answers it for you.
How do you work out which rules your own mail needs
What we would do first, before building a single rule, is go back through the last 100 emails that came in, on screen or in a CSV export if your desk will give you one, and sort them by hand into piles by who ought to have dealt with each. It takes an hour, near enough.
As you go, write one word beside each mail for what told you where it belonged, whether that was the address it came to, or who sent it, or a word in the subject line, or whether you had to go and read the whole of it before you knew. Nobody enjoys writing that last one down.
Anything with an address or a sender or a word written beside it can have a rule, and whichever of those piles came out tallest is the rule we would switch on first. The mail you could only place by reading the whole of it properly is the catch-all person’s week, give or take, and on most desks that pile comes out a good deal taller than anybody guessed before they sat down with the hundred. Find that number out your own self before anybody builds a rule at all, so somebody’s name is on it from the first day.
Common questions about ticket routing rules
What is ticket routing?
Ticket routing is deciding, automatically, which person or group a new support email goes to, usually with rules that read fields such as the address it was sent to, the sender’s domain or words in the subject line. Routing rules often set the priority and add tags at the same time.
What is the difference between ticket routing and ticket assignment?
Routing sends a ticket to the right place, often a group such as Billing or Engineering. Assignment puts one named person on it. A rule can do both at once by assigning to a person, but a ticket routed to a group still needs somebody in that group to take it.
Should routing rules use keywords or email addresses?
Addresses first. The address a customer writes to is a deliberate choice, so it misroutes least; in a made-up week of 200 mails, an address rule misrouted 3 of 38 (under 8%) while subject-keyword rules misrouted 10 of 58 (about 17%). Keep keyword rules for the mail that arrives at a general address, and run them last.
What is round-robin ticket assignment?
Handing new tickets to each person in turn, regardless of how busy they already are. Assigning by workload, where the ticket goes to whoever has the fewest open, evens things out better on a desk where some tickets take minutes and others take days. In Maxdesk, workload-based assignment is done by the AI triage agent on the Elite plan.
What should happen to email that no routing rule matches?
It should go to one named person, not sit unassigned in a shared view. A catch-all rule that assigns anything still unassigned after a set time, such as 30 minutes, to the person on duty that week makes sure every email has an owner. Count that pile weekly to see how much your other rules miss.
How many routing rules does a small support team need?
Fewer than most start with. An address rule, rules for machine mail and a catch-all usually carry most of the mail; add keyword rules only where a hand count of your own last 100 emails shows a pile big enough to be worth one.
Rules and product details reviewed on 29 September 2026 against the live Maxdesk site. The week of 200 mails is made up for the example; the GDPR one-month window is from Article 12(3) of the regulation.