What Is ACD? Call Routing Methods and Queue Design Explained
A plain guide to automatic call distribution: how calls reach the right agent, common routing methods, queue and skill design, and how to manage waiting time.
When a handful of people share one phone line, it is easy to decide who answers: whoever is nearest. Once call volume grows and questions become more varied, that stops working. One person gets every call, a new starter takes a complicated billing dispute, and callers in the waiting line have no idea whether they are next or fiftieth.

Automatic call distribution (ACD) solves this. It is the part of a phone system that takes an incoming call, finds an agent who can take it, and hands it over according to rules you set. If nobody is free, it keeps the caller in a queue and serves them in order. This guide explains how it works, compares the common routing methods, and shows how to design queues, skills and waiting experience. If you are new to menus, read What Is IVR? first.
What ACD does
Without ACD, a call either rings every phone at once or lands on a single reception desk where someone transfers it by hand. With ACD, the system does the following automatically:
- Finds agents who are available right now and offers them the call.
- Places the call in a queue when everyone is busy, and keeps the order fair.
- Plays announcements or music while the caller waits.
- Sends different kinds of calls to different teams.
- Switches to another plan, such as voicemail or a callback, when waits grow too long.
- Records who answered, how long callers waited and how many hung up.
How it fits with IVR
An IVR menu asks the caller what they need. ACD then decides who, inside the chosen team, takes the call. A caller who presses the option for orders enters the orders queue, and ACD picks an agent from that queue. IVR Menu Design covers the menu side.
The parts of an ACD setup
- QueueA line that callers wait in. Usually one per team or type of enquiry.
- AgentsPeople attached to one or more queues who can take the calls.
- Routing ruleThe logic that picks which available agent gets the next call.
- Agent statusWhether someone is available, on a call, on a break or offline.
- Waiting planWhat callers hear, how long they can wait and what happens at the limit.
- RecordsCounts of calls offered, answered, abandoned and the time callers waited.
Common routing methods
Different methods answer the question "who gets the next call?" in different ways. None is universally right; the choice depends on your team and the kind of calls.
Ring everyone
Every available agent's phone rings and the first to pick up takes the call. It is simple and quick for tiny teams, but it creates noise and unfair habits: the keenest person always wins, and others may stop trying.
Round robin
Calls rotate through the agents in a fixed order. It spreads work evenly in count, though not necessarily in effort, because one call can last two minutes and another forty.
Longest idle first
The call goes to the agent who has waited longest since their last call. This keeps the load fair in practice and is a popular default for general support desks.
Fixed priority
Agents are ranked, and the call goes to the highest-ranked available person. Useful when a specialist should take calls first and others act as backup, but it can overload the specialist.
Least calls handled
The call goes to whoever has taken the fewest calls today. It evens out counts but can send a hard call to someone who has not had a break.
| Method | Works well for | Watch out for |
|---|---|---|
| Ring everyone | Very small teams | Noise, unequal effort |
| Round robin | Similar agents, similar calls | Uneven call length |
| Longest idle first | General support desks | Needs reliable status |
| Fixed priority | Specialist plus backup | Overloading the top agent |
| Least calls handled | Fairness over a whole day | Ignores call difficulty |
Designing your queues
A queue is the unit callers wait in, so it is also where your design choices become visible. Good queues are few, clearly named and tied to real teams.
What to split queues by
- Topic: orders, billing, technical help, new enquiries.
- Language: one queue per language your team handles.
- Customer type: existing customers and prospects often deserve different handling.
- Site or region: useful when teams are in different time zones.
Do not over-split
Every extra queue divides your agents into smaller pools, and small pools mean longer waits for the same number of calls. Start with two to four queues. Only add another when a group of calls has clearly different needs, for example a topic that requires training the others do not have.
Agents in more than one queue
An agent can belong to several queues, perhaps as a primary member of one and a backup in another. This helps absorb peaks without creating extra headcount, as long as the priorities are clear.
Queue design worksheet
- List the reasons people call and how often each happens
- Group them into two to four teams
- Name each queue in words customers and agents both understand
- Decide who is primary and who is backup for each queue
- Choose an overflow queue for when one fills up
- Set the hours each queue is open
Skill-based routing
Skill-based routing adds one condition: only agents with a certain skill are offered the call. A call about a specialist product goes only to those trained on it, even if other agents are idle.
It raises the chance a question is solved on the first call, but there is a trade-off. If only two people hold a skill, calls needing it queue even while others sit idle. Keep skills few and broad, such as language or product line, and train backups so a single absence does not stall a queue.
Managing waiting time
Waiting is where callers form their opinion of you. Three habits make the wait more bearable and keep it from becoming a hang-up.
A sensible waiting plan
- AcknowledgeTell the caller they are in the right place and their call will be answered.
- InformSay roughly how busy you are, only if the figure is reliable.
- Offer a choiceAfter a set time, offer a callback or voicemail instead of holding.
- Set a limitDecide the longest wait you accept before the call is moved to another plan.
- Follow upTreat every voicemail or callback request as a call that must be returned.
Choose a service target you can measure, such as answering most calls within a set number of seconds, and track it. Answer Rate and Abandonment Rate explains the measures, and Using Callbacks and Voicemail in Your IVR covers the alternatives to waiting.
Agent status and wrap-up
ACD can only route correctly if it knows who is free. If someone leaves their desk but stays marked available, the system keeps ringing an empty chair. If someone is ready but marked on break, callers wait for no reason.
- Keep the status list short: available, on a call, on a break, offline.
- Make changing status quick, so people do it honestly.
- Give agents a short wrap-up period after calls for notes, and decide how long it can last.
- Let supervisors correct a status when someone forgets.
Setting ACD up in stages
- List the call types and their approximate volume.
- Create the queues and attach agents.
- Choose a routing method for each queue.
- Set waiting rules: announcements, limits, overflow and out-of-hours behaviour.
- Place test calls from outside and check the route, the wait and what happens at the limit.
- Go live, then review reports weekly for the first month.
Test calls to run
- A call when everyone is free.
- A call when everyone is busy.
- A call outside business hours.
- A call that needs a skill nobody currently available has.
- A caller who hangs up while waiting.
Reviewing after launch
Look at a few numbers regularly: how many calls were offered, how long the longest wait was, how many callers hung up, and how many calls ended up in the overflow queue. A rising abandonment rate during a particular hour tells you where staffing or rules need to change.
Symptom and likely cause
| Symptom | Likely cause | |
|---|---|---|
| Long waits at lunch time | Too few agents scheduled then | Plan breaks in shifts and add backup |
| One agent always busy | Fixed priority or a rare skill | Rebalance with longest idle first and train backups |
| Callers pressing 0 constantly | Menu does not match their needs | Review menu wording and queues |
| Rings with no answer | Agent status out of date | Use a clear status routine and supervisor override |
Pair this with First Call Resolution to check that fast answers are also correct answers, and with Workforce Management to staff for the busy hours you find.
Doing it in Call
In Call, each queue has its own settings for targets, opening hours and routing. You can require a skill, name an overflow queue and choose how long an agent's phone rings before the call moves on. When a queue is closed you can choose between taking a message and sending callers to another queue. Agents set their own status from the menu, and supervisors can change it from the Supervisor screen. A routing simulator lets you walk a pretend call through hours, holidays and skills without dialling anyone. See Setting Up Call Queues for the steps.
Frequently asked questions
What does ACD stand for?
Automatic call distribution. It is the part of a phone system that hands incoming calls to available agents using rules you define, and queues callers when nobody is free.
Is ACD the same as IVR?
No. An IVR asks the caller what they need and chooses a team. ACD then picks the agent within that team and manages the waiting line.
Which routing method should a small team use?
Longest idle first is a safe default because it shares work fairly without extra setup. Ring everyone can be fine for a very small team where noise is not a problem.
How many queues do I need?
Usually two to four. Every extra queue splits your agents into smaller pools, which tends to lengthen waits, so add one only when calls clearly need different handling.
When should I use skill-based routing?
When some calls need knowledge only certain people have, such as a language or a product line. Keep the list of skills short and train backups so one absence does not block a queue.
What should happen when a caller waits too long?
Offer a choice such as a callback or voicemail, or send the call to an overflow queue. Decide the limit in advance rather than letting callers wait without end.