Call Queue Announcements: What to Say and When

Most contact centers spend significant effort on agent scripting and IVR design, then deploy queue announcements as an afterthought — a default "please hold" message and whatever the platform generated at setup. That's a mistake. Queue announcements are the primary communication channel between the business and callers who are already waiting. What those announcements say, how often they play, and in what order determines whether callers stay on hold or hang up before an agent ever answers.

This guide covers each type of call queue announcement, what each should say, when it should play, and the specific mistakes — particularly around estimated wait time — that erode trust rather than build it.

Type 1 — Entrance and Welcome Announcements

The entrance announcement plays as soon as a caller enters the queue, before hold music begins. It is the first thing a waiting caller hears, and it sets the tone for the entire hold experience.

A well-constructed entrance announcement does three things in roughly 10–20 seconds: it confirms the caller reached the right place, acknowledges the wait, and tells them what information they will receive next. That structure immediately reduces uncertainty — callers know they are in the right queue, they know there is a wait, and they know they will be told how long.

A good entrance announcement sounds like:

"Thank you for calling EaseDial Support. All agents are currently assisting other callers. We'll give you an estimated wait time in a moment — please stay on the line."

What not to include in the entrance announcement:

• Generic phrases like "your call is very important to us" — callers have heard this thousands of times. It communicates nothing and frequently reads as condescending when wait times are long.
• Long marketing messages — a caller in a support queue does not want to hear a product promotion. Save promotional content for hold messages, where callers have already settled in and are more receptive.
• Alternatives that should have been handled in the IVR — if a caller has made it past the IVR into a queue, presenting "visit our website" or "press 0 to leave a voicemail" in the entrance announcement means your IVR design has already failed. Route those offers earlier.
• Excessive legal disclosures — a brief "this call may be recorded" is appropriate; paragraph-length legal text in the entrance message is not.

Keep the entrance announcement under 20 seconds. Callers are impatient at the moment they enter a queue — they have already been through your IVR and want to know when they will reach a person. Every extra second of entrance message is a small frustration added before the wait even begins.

One important exception: if call volume is unusually high due to a known event such as a system outage or billing cycle, the entrance announcement is the right place to acknowledge it explicitly. See the section on outage notices below.

Type 2 — Position Announcements

Position announcements tell callers where they stand in line: "You are caller number 4" or "There are approximately 3 callers ahead of you." Their primary value is enabling callers to make an informed decision about whether to stay on hold.

A caller who hears "there are 2 callers ahead of you" is likely to stay. A caller who hears "there are 14 callers ahead of you" may decide their time is better spent on another channel. This is not a failure — it is the announcement doing its job. Callers who would have abandoned after 8 minutes of waiting get the information they need at the 60-second mark, and you avoid a long hold followed by a frustrated abandonment.

The Position Framing Problem

How position is communicated matters significantly. Stating an ordinal position — "you are 4th in the queue" — sounds precise but creates a specific failure mode: when a VIP caller is inserted ahead of standard callers via priority queuing, a caller can hear "you are 3rd in the queue," then 90 seconds later hear "you are 4th in the queue." The queue appears to have moved backward. This is technically correct but caller-hostile — it sounds like the system is broken.

Framing position as "approximately 4 callers ahead of you" rather than "you are 4th" softens the precision and makes backward-movement less jarring. The word "approximately" sets the right expectation: this is an estimate, not a reservation number.

Update Frequency

Position announcements should play every 60–120 seconds. More frequent than 60 seconds is intrusive — callers are interrupted from whatever they are doing while they wait. Less frequent than 2 minutes leaves callers uncertain whether the call is still active. Many platforms use a fixed interval; the optimal setting is 90 seconds for most queues.

Position announcements are most useful in short-to-medium queues of under 10 callers. In a deeply backlogged queue with 20 or more callers waiting, a position announcement provides little comfort and may cause callers to abandon immediately. For deep queues, estimated wait time is more informative — a position of 18th in line is abstract, but "approximately 12 minutes" is a decision-able number.

Type 3 — Estimated Wait Time Announcements

Estimated wait time (EWT) announcements tell callers how long they can expect to wait before an agent answers: "Your estimated wait time is approximately 8 minutes." They are the most decision-relevant announcement type — but also the most dangerous to get wrong.

How EWT Is Calculated

The ACD calculates EWT from three inputs: current queue depth (how many callers are waiting), current average handle time (how long calls are taking), and current agent availability (how many agents are live and how many are expected to become available). The formula is roughly:

EWT = (calls ahead of you × average handle time) ÷ agents available

The problem is that "agents available" is a snapshot at a single moment. Agents go on break, encounter unexpectedly long calls, or log out before the system updates its model. A calculation that assumed three agents would be available may find itself applied to a queue where only two agents are actively taking calls. This causes systematic underestimation.

Accuracy Is Not Optional

Research on caller tolerance shows a consistent pattern: an accurate EWT that turns out to be correct produces far better outcomes than an underestimate, even when the actual wait is longer. A caller told "approximately 8 minutes" who waits 8 minutes is satisfied. A caller told "approximately 3 minutes" who waits 8 minutes is frustrated, even though the second caller waited less time in absolute terms. The underestimate creates a broken promise.

Underestimates are significantly worse than overestimates. A caller told 8 minutes who reaches an agent in 5 minutes feels pleasantly surprised. A caller told 3 minutes who waits 8 minutes has their patience tested not once but continuously as each minute passes without the promised connection.

For this reason, best practice is to buffer calculated EWT upward by a factor of 1.1–1.2. If the system calculates a 6-minute wait, announce approximately 7 minutes. This accounts for agent availability fluctuation and delivers a more accurate caller experience in practice, even if it occasionally results in a caller waiting slightly less than announced.

When Not to Give EWT

If your system's EWT calculation is unreliable — for example, because your call volume is erratic, agent schedule adherence is low, or your AHT varies widely by call type — it is better to omit the EWT entirely than to state a number that is frequently wrong. A wrong EWT is worse than no EWT. Position announcements or a simple "we're experiencing high volume" message is preferable to an EWT that consistently underdelivers.

Type 4 — Periodic Hold Reminders

Periodic hold reminders are the brief interjections between music segments that reassure callers the call is still active and they have not been silently disconnected. They are the simplest announcement type and the one most easily over-deployed.

An effective periodic reminder is 5–10 seconds: "Thank you for holding. Your call will be answered shortly." Alternatively, use the reminder to deliver something useful: "While you hold, many common account questions can be answered at easedial.com/support" or "You can also reach us via live chat on our website."

Reminder Frequency Should Decrease Over Time

Most platforms play periodic reminders at a fixed interval — every 60 or 90 seconds throughout the entire hold period. This is suboptimal. A caller who has been waiting for 10 minutes does not need to hear "thank you for holding" every 60 seconds; at that frequency it becomes irritating rather than reassuring. The fact that the call is still live is by that point not in question.

If your platform supports variable reminder scheduling, increase the interval as wait time accumulates. Reminders every 60 seconds for the first 3 minutes, every 90 seconds for the next 3 minutes, then every 2–3 minutes thereafter is a reasonable progression. The goal is to reassure without annoying — a caller who is 12 minutes into a hold experience has already invested and does not need constant confirmation that you remember they exist.

Type 5 — Callback Offer Announcements

The callback offer gives callers the option to hang up and receive a call back when an agent is available, preserving their position in the queue. Done correctly, it is one of the highest-value announcements in the queue experience because it converts a likely abandonment into a scheduled callback.

An effective callback offer announcement covers three things: how to opt in, what happens after they opt in, and approximately when they will be called back. For example:

"Would you like to receive a callback instead of waiting? Press 1 now and we'll call you back at approximately [EWT from now]. You'll keep your current position in the queue and won't need to call us back."

Several deployment decisions matter for callback offers:

• Offer timing: Present the callback option after 60–90 seconds of wait — early enough that callers haven't already invested significant time and become reluctant to hang up. If you wait until a caller has been on hold for 10 minutes, they are much less likely to accept because they've already committed.
• Second offer: If the caller did not accept at 60 seconds, offer again at the halfway point of the estimated wait. A caller who initially said no may reconsider after sitting through several more minutes.
• Accurate callback timing: Only quote an approximate callback time if your EWT is reliable. Telling a caller "we'll call you back in about 8 minutes" and calling them 25 minutes later is worse than offering the callback with no specific time.
• Number confirmation: The announcement should confirm the caller's number — typically the ANI — before they hang up, or prompt them to enter a different number. Callers should not have to wonder whether the system captured their number correctly.

The dedicated guide on queue callback examines how it works technically and operationally, including caller experience and agent-side workflow.

Type 6 — Outage and High-Volume Notices

When call volume is elevated due to a specific, known event — a system outage affecting customers, a billing event generating widespread inquiries, a product launch, or any other situation your team is aware of — the entrance announcement should acknowledge it explicitly.

A specific high-volume notice outperforms a generic one. "We are currently experiencing an issue affecting login for some customers and are working to resolve it" sets accurate expectations and signals awareness. "We are experiencing higher than normal call volume" is the minimum — it's generic but still better than a standard entrance message that gives callers no context for an unexpectedly long wait.

Why this matters: callers who understand why the wait is long are significantly more tolerant than callers who do not. An unexplained 15-minute wait generates frustration. A 15-minute wait following an announcement that acknowledges a known issue and indicates the team is working on it generates patience. The information itself reduces the perceived severity of the wait.

High-volume notices should be updated as conditions change. If a system outage is resolved while callers are still in the queue, update the announcement rather than continuing to reference an issue that has been fixed. Most platforms allow announcement changes without disrupting calls currently in queue.

Announcement Timing and Pacing

The sequencing of announcements across the full hold experience matters as much as the content of individual announcements. The worst queue experience plays an entrance message and then, 30 seconds later, fires a periodic reminder — giving the caller two announcements within the first minute before any music has played for more than a few seconds. It sounds uncoordinated and amplifies the sense that waiting is actively unpleasant.

A well-paced queue experience follows a structure like this:

• 0:00 — Entrance announcement (10–20 seconds)
• 0:20 – 1:40 — Hold music without interruption (80 seconds)
• 1:40 — First EWT or position update (10–15 seconds)
• 1:55 – 3:25 — Hold music (90 seconds)
• 3:25 — Callback offer, if wait exceeds threshold (15–20 seconds)
• 3:45 onward — Music with periodic reminders every 90–120 seconds; second position or EWT update at the 5-minute mark

Each announcement interruption should be under 15 seconds. Longer interruptions break the hold experience without providing proportional informational value. Callers have established a rhythm — they have accepted the wait and settled into it. Long announcements disrupt that rhythm unnecessarily.

One practical note: coordinate the announcement schedule with music segment length. If your music segments are 3 minutes long and your announcement fires every 90 seconds, announcements will interrupt music mid-track. Configure your platform so announcements fire at natural break points between music segments where possible. This produces a noticeably smoother experience even if callers cannot articulate why.

Localization, Accessibility, and Recording vs TTS

Localization

For contact centers serving callers in multiple languages, queue announcements should be available in each supported language, triggered by the language selection made in the IVR. A caller who navigated the IVR in Spanish should hear queue announcements in Spanish — not a language switch mid-call that signals the system has lost track of their preference.

Dynamic content in localized announcements — position numbers, EWT figures — should be handled carefully. Numeric grammar differs across languages (Spanish "cuatro llamadas" versus "una llamada"), and TTS engines vary in their handling of gendered numbers. Test localized dynamic announcements fully before deploying them to a live queue.

Accessibility and Speech Rate

Text-to-speech engines default to a speaking rate optimized for average listeners, which is often 160–180 words per minute. For callers with cognitive disabilities, older callers, or callers with limited fluency in the announcement language, this rate is too fast to reliably parse numeric information like wait times or position counts. Target a TTS speaking rate of 130–150 words per minute for queue announcements to ensure comprehension across a broader caller population.

Regulatory note: several EU jurisdictions require that "this call may be recorded" disclosure be delivered in queue announcements rather than relying solely on website disclosure. If you serve callers in those jurisdictions, include the disclosure in the entrance announcement regardless of whether it also appears elsewhere in your call flow.

Human Recording vs Text-to-Speech

Professional human voice recordings produce higher-quality audio, carry a warmer tone, and signal that the organization invested in its caller experience. Callers perceive human-recorded announcements as more trustworthy than TTS, particularly in enterprise or customer support contexts where tone of voice affects perception of the brand.

The limitation of human recordings is inflexibility. Static phrases — "Thank you for calling EaseDial Support" — record once and remain stable. But dynamic content — estimated wait times, position numbers, day-specific notices — changes constantly and cannot be recorded in advance.

The optimal deployment for most contact centers is a hybrid approach: professional human recordings for all static content, TTS for dynamic values only. The entrance announcement can be fully human-recorded. The EWT announcement uses a human-recorded frame — "Your estimated wait time is approximately" — with a TTS-generated number inserted into the gap. This preserves recording quality for the majority of the announcement while enabling dynamic content where it is needed.

When selecting a TTS voice for the dynamic segments, choose a voice and engine that closely matches the tone and pace of your human recordings. A mismatch between a warm professional female voice in the static segment and a robotic synthesized voice reading the number produces an audible seam that undermines the quality of the overall announcement.

Frequently Asked Questions

How often should estimated wait time be updated while a caller is in the queue?

Most platforms recalculate EWT continuously but play the updated figure only at scheduled announcement intervals, typically every 60–120 seconds. Playing an updated EWT on every recalculation cycle would mean interrupting hold music every 10–15 seconds — disruptive and unhelpful. A position or EWT update every 90–120 seconds strikes the right balance: callers receive current information at predictable intervals without feeling constantly interrupted. Avoid playing an EWT update that is identical to the previous one; if the figure hasn't changed materially, play a brief hold reminder instead.

Is it better to give position or estimated wait time?

It depends on queue depth and handle time predictability. For short queues (under 8–10 callers) with consistent handle times, position announcements are useful because the wait is short enough that a position count is meaningful. For deeper queues or queues with variable handle times, EWT is more actionable — a caller who hears "14 callers ahead of you" cannot translate that into a decision, but a caller who hears "approximately 9 minutes" can. If your EWT calculation is unreliable, position announcements are preferable to a frequently wrong EWT. Some platforms play both, which is fine if the interval is appropriate.

What happens if the EWT announcement is consistently wrong?

Consistently wrong EWT announcements — particularly underestimates — erode caller trust and increase abandonment over time. Callers who have been told "3 minutes" and waited 12 do not remember the phone call positively, even if the agent interaction that followed was good. The fix is to either improve EWT accuracy by correcting the inputs the ACD uses (current AHT, agent availability, not a stale average), apply the 1.1–1.2× buffer to calculated EWT, or disable EWT announcements entirely until the calculation is reliable. If your platform cannot produce reliable EWT figures, position announcements combined with an explicit high-volume acknowledgment in the entrance message is a safer configuration.

Can queue announcements differ between queues in the same contact center?

Yes, and in most cases they should. A billing queue caller has different context needs than a technical support queue caller. Enterprise queues may use more formal announcement language and skip generic callback offers if those callers have dedicated account managers. A high-volume overnight queue might include a self-service web alternative in its periodic reminders because staffing is low and the wait genuinely is long. Modern contact center platforms support per-queue announcement configuration — take advantage of it rather than using a single set of announcements across all queues.

Should the entrance announcement mention the website or self-service options?

Only briefly, and only if those alternatives are genuinely useful for the callers in that queue. A caller in a billing dispute queue cannot resolve their issue on a self-service portal — mentioning the website is noise. A caller in a general support queue who wants to check an order status probably can self-serve, and a brief mention in the entrance announcement or first periodic reminder is appropriate. The entrance announcement itself should remain under 20 seconds; if you include a self-service mention, ensure it does not push the total past that threshold. More detailed self-service direction belongs in periodic hold reminders where callers have time to write down a URL or act on the alternative.