What is automatic call distribution (ACD)?
Automatic call distribution (ACD) is the system that takes an incoming contact and routes it to the right available agent, using rules based on skills, availability, and priority rather than a simple first-come order. It's the piece of contact center infrastructure doing the actual queuing and routing, whether the contact arrived by phone, chat, or another channel.
ACD has been core contact center plumbing since the 1970s, long before "omnichannel" was a word anyone used. What's changed is what feeds it: routing that used to run on a caller's menu selection increasingly runs on account history, topic, and predicted outcome instead.
This page covers how ACD works, the routing methods it typically supports, how it differs from an IVR, what it falls short on, and how AI is changing what "routing" means.
ACD in one sentence
ACD decides which agent a contact reaches once it's clear the contact needs a person.
How ACD works
Identify the caller and the reason. The ACD pulls whatever context is available, caller ID, account lookup, or the selections made in an IVR, to understand who's contacting and why.
Apply routing rules. Based on that context, the system decides which agent, or which group of agents, should handle the contact.
Queue if nobody's available. If no matching agent is free, the contact waits, and the ACD tracks how long it's been waiting for reporting and for any queue-position messaging the customer hears.
Distribute. The moment a matching agent becomes available, the ACD delivers the contact to them and starts the clock on handle time.
Common ACD routing methods
Skills-based routing. Agents are tagged with the skills or topics they can handle, and contacts route to an agent who has the relevant tag. This is the most common method in centers with more than a handful of agents, because it keeps specialized work with the people equipped for it.
Round-robin or uniform distribution. Contacts rotate evenly across a group of agents regardless of specialty. Simple, and reasonable when the team is genuinely interchangeable.
Longest-idle-agent routing. The agent who's been available longest gets the next contact, which spreads volume evenly across a shift instead of letting the fastest agents absorb a disproportionate share.
Priority routing. Certain contacts, a high-value account, a repeat contact on the same issue, jump the queue ahead of contacts that arrived earlier.
Most contact centers run a blend rather than picking one method for every queue: skills-based for specialized topics, longest-idle within a skill group, and priority overrides for the contacts that need to skip the line.
ACD vs. IVR
Interactive voice response (IVR) and ACD are often mentioned together and do different jobs. IVR is the menu that collects information from the caller, by voice or keypad, before a person is involved. ACD is what takes that information and decides where the contact goes.
A caller who presses 2 for billing is interacting with the IVR. The system that finds a billing-qualified agent and delivers the contact the instant one's free, queuing the caller in the meantime if none are available, is the ACD. Most voice deployments run the two together, with the IVR feeding structured input to the ACD's routing rules.
What ACD data feeds
Because every contact passes through it, the ACD is usually the source of record for queue-level metrics: average speed of answer, abandonment rate, and service level are all built from timestamps the ACD generates as a contact enters the queue, waits, and gets answered. A routing rule that's misconfigured doesn't just misdirect contacts. It quietly distorts every metric downstream of it.
Where ACD routing falls short
Skill tags go stale. An agent trained on a new product line six months ago is still tagged for their original skills unless someone remembers to update the routing configuration. Stale tags are one of the most common, least visible causes of a contact landing with the wrong agent.
Rules-based routing can't see the whole picture. A contact tagged "billing" by the IVR might really be a billing dispute tied to a shipping delay that a different queue is already working. Routing on a single selected topic misses context that a human, or a system with access to the full account history, would catch immediately.
More queues mean more seams. Splitting agents into narrow skill groups improves specialization but adds transfer risk: a contact that lands in the wrong queue has to be manually moved, adding a second wait on top of the first.
ACD and AI routing
Traditional ACD routes on a small set of inputs: the menu selection, maybe the phone number, maybe a queue an agent self-selected into. What's changing is how much more it can now draw on before making that decision, account history, the customer's prior contacts, order status, even a prediction of how complex the issue is likely to be, so the first agent a customer reaches is more often the right one, not just an available one.
That shift matters most for the contacts that used to bounce between queues. A returns question that's really an account-standing issue, or a billing question that's really a fraud flag, can route correctly the first time when the system routing it has more to work with than a single IVR selection.
Frequently asked questions
Learn more
- What is AI hallucination?
- What is AI orchestration?
- What is Net Promoter Score (NPS)?
- What is a CRM?
- What is a chatbot?
- What is a helpdesk?
- What is a knowledge base?
- What is a large language model (LLM)?
- What is a service level agreement (SLA)?
- What is a ticketing system?
- What is after-call work (ACW)?
- What is agent utilization?
- What is agentic AI?
- What is agentic commerce?
- What is agentic customer service?
- What is an AI agent?
- What is autonomous resolution rate?
- What is average handle time (AHT)?
- What is average speed of answer (ASA)?
- What is call abandonment rate?
- What is call center occupancy?
- What is call center service level?
- What is call center shrinkage?
- What is contacts per order (CPO)?
- What is conversational AI?
- What is conversion rate optimization?
- What is cost per contact?
- What is customer churn?
- What is customer effort score (CES)?
- What is customer experience?
- What is customer journey mapping?
- What is customer lifetime value (CLV)?
- What is customer loyalty?
- What is customer retention?
- What is customer satisfaction score (CSAT)?
- What is customer service automation?
- What is customer service software?
- What is deflection rate?
- What is first contact resolution (FCR)?
- What is generative AI?
- What is gross merchandise value (GMV)?
- What is grounding in AI?
- What is interactive voice response (IVR)?
- What is natural language processing (NLP)?
- What is omnichannel customer service?
- What is prompt engineering?
- What is retrieval-augmented generation (RAG)?
- What is schedule adherence?
- What is sentiment analysis?
- What is voice AI?
- What is voice of the customer (VoC)?
Going deeper?
See how Gladly customers put this into practice in their day-to-day customer service work.