In 1985, David Maister published a paper called "The Psychology of Waiting Lines." It's been cited thousands of times, and almost everything useful about wait time in customer service traces back to it.

His core insight: the actual duration of a wait matters less than how the wait feels. A 2-minute wait with no information feels longer than a 5-minute wait with a countdown timer. And that perception gap explains why some companies with 10-minute response times have happy customers while others with 2-minute response times have angry ones.

Maister's Principles (Applied to Support)

Occupied time feels shorter than unoccupied time. If a customer is staring at a "please wait" screen with no indication of progress, every second stretches. If they're reading a status update ("Your message has been received. A specialist is reviewing it now. Typical response time: 4 minutes"), the same wait feels shorter.

This is why hold music exists. Nobody likes hold music, but silence on a phone line makes people think the call dropped. The music says "you're still connected." A terrible solution to a real problem, but better than nothing.

For chat support, the equivalent is typing indicators and status messages. "Agent is typing..." makes a 90-second wait feel manageable. A blank screen for 90 seconds makes people close the tab.

Uncertain waits feel longer than known waits. "We'll get back to you soon" is torture. "We'll respond within 2 hours" is manageable, even if 2 hours is longer than "soon" would imply. The certainty matters more than the duration.

This is why SLA promises work even when the actual time is the same. Telling a customer "you'll hear from us within 4 hours" and then responding in 3 sets an expectation and then beats it. The customer feels like they got fast service, even though 3 hours is objectively slow.

Unexplained waits feel longer than explained waits. "We're experiencing higher than normal volume due to a system update" converts rage into understanding. The customer's mental model shifts from "nobody cares about me" to "there's a reason for this."

Pre-process waits feel longer than in-process waits. The time before someone acknowledges your message feels eternal. The time after someone says "I'm looking into this" feels much shorter, even if the total time is the same.

This is the biggest argument for instant acknowledgment. Even if you can't solve the problem for 2 hours, responding in 30 seconds with "Got your message, looking into it now" cuts the perceived wait dramatically.

How This Applies to AI Support

AI eliminates the pre-process wait entirely. The customer sends a message. Within 200 milliseconds, it's classified, and within a few seconds, they get a response. No queue. No hold. No "please wait."

For the 40 to 60% of queries that AI can resolve outright, the wait time is effectively zero. "Where's my order?" gets an instant answer with tracking info. "What are your hours?" gets an immediate response. The customer never experiences waiting at all.

For the queries that need a human, AI provides the instant acknowledgment that Maister's research says matters most. "I can see you're asking about [specific issue]. I've flagged this for our team. You'll hear back within [SLA]. In the meantime, here's some info that might help: [relevant article]."

That response takes 3 seconds. It eliminates the pre-process wait, explains the remaining wait, and provides occupation during the wait (the article link). Three of Maister's principles addressed in one automated message.

The Anxiety Factor

Maister's most overlooked principle: anxiety makes waits feel longer. A customer who's worried their money is lost perceives every minute of waiting as twice as long as a customer asking a routine question.

This is why billing and payment issues need the fastest response times. The emotional stakes are higher. The anxiety multiplier is real. A 30-minute wait for "how do I change my email?" is annoying but tolerable. A 30-minute wait for "I was charged twice and my rent is due tomorrow" is genuinely distressing.

AI classification helps here because it identifies high-anxiety intents (billing disputes, account access issues, fraud reports) and routes them to faster queues. The classification doesn't just sort by topic. It sorts by emotional urgency.

Supp's priority scoring does this automatically. A message about a double charge gets a higher priority score than a message about feature preferences. The customer who's anxious about their money gets seen first.

What "Fast" Actually Means

The perception research suggests diminishing returns on speed. Going from a 24-hour response to a 4-hour response is a massive improvement in perceived quality. Going from 4 hours to 2 hours is noticeable. Going from 2 hours to 1 hour is barely noticeable. Going from 1 hour to 30 minutes is invisible to most customers.

The exception: chat and in-app messaging. For synchronous channels, every second matters. A 30-second chat response feels slow. A 5-second response feels instant. This is where AI has the biggest impact, because human agents can't respond in 5 seconds, but AI can.

For email, the sweet spot is 1 to 4 hours during business hours. Faster is nice but delivers diminishing returns on satisfaction. The key threshold is whether you respond the same day.

For phone, any hold time over 2 minutes feels unacceptable to most people. Under 1 minute is perceived as "no wait." Between 1 and 2 minutes is "a short wait." Above 2 minutes, satisfaction drops sharply with each additional minute.

Practical Changes

Add expected wait times to your support flow. "Typical response time: 4 hours" on your contact form. "Estimated wait: 3 minutes" in your phone queue. The number doesn't have to be short. It has to be honest.

Send instant acknowledgments for every message. Automate this. The customer should never wonder "did they get my message?" Even a simple "We got your message and will respond within [X]" reduces perceived wait by 30 to 40%.

Prioritize by anxiety, not just by order received. First-in-first-out is fair but not optimal. A billing dispute submitted 5 minutes ago is more urgent than a feature question submitted 10 minutes ago. Route accordingly.

Give customers something to do while waiting. A relevant help article. A status page link. An FAQ answer that might resolve their issue before an agent responds. Occupied waiting always feels shorter.

And measure perceived wait, not just actual wait. Ask customers "how long did you feel you waited?" alongside tracking actual response time. The gap between the two tells you whether your wait experience needs work.