What 84.5 million scored conversations show about when to call, and how the best contact centers staff the night.
This article is part of The Conversation Lab, where we publish short data stories from customer conversations. We analyze interaction patterns across our customer base and share what we find with CX leaders who need stronger signals than sampled QA and survey responses alone.
If you need help from a customer service team, the hour you call matters. You probably pick a time based on hold time — early morning to beat the rush, or midday when you assume more agents would be staffed. Most people optimize for speed. Almost nobody optimizes for the quality of the experience they'll get once they're connected.
That's because traditionally, the data hasn't been available.
Quality dashboards inside contact centers show rolling averages for the team, the queue, or perhaps the type of inquiry. Those averages absorb the variation across the day, which means the hours where service quality peaks and the hours where it drops are invisible to the people running the operation, let alone the customers calling in.
If you can call early, you should
This was the most counterintuitive finding in the dataset. Conversations between 6 and 10 AM posted the highest full-resolution rate of the day, at 39.5%. If you're a customer with a complicated issue — a billing dispute that requires research, a claim that needs documentation, anything where you'd rather not call back — early morning gives you the best odds of getting it handled in one shot.
Those conversations also ran about 25% longer than afternoon ones at the median. Agents are spending more time with each customer and closing more issues as a result, because queue pressure hasn't built yet. When agents aren't managing a packed queue, they have room to work a conversation to completion instead of moving to the next one.
For CX leaders, the implication is routing. If you have high-value or complex cases that benefit from longer handle time and a higher probability of first-contact resolution, the early-morning window is where they're most likely to close. For customers, the implication is simpler: call before the lines get busy, and you're more likely to walk away with your issue resolved.
The overnight drop is the biggest time effect in the data, but the experience depends on the company you're calling
1. Avoid calling between 2 and 6 AM
Between 2 and 6 AM, iCSAT falls 10.7% (from 3.57 to 3.23) from its afternoon level. More than one in three conversations in that window ends with the customer's issue unresolved, compared to roughly one in four during the rest of the day.
This overnight finding gets interesting when you look across the 12 companies in the data set and compare their staffing models. Ten of the 12 companies in the dataset score lower overnight. But two actually score higher between 2 and 6 AM than during the day.
These differences come down to how the night shift is designed.
The companies that hold daytime quality overnight share a set of choices: they staff the night with a dedicated team that works only nights rather than rotating core-hours agents into overnight coverage, they keep their stronger channel open instead of shutting it down and shifting volume to a weaker one, and they manage queue load per agent so that overnight staff isn't carrying more work than their daytime counterparts.
The companies with the steepest overnight drops do the opposite — they cover nights with two to eight agents who also work core hours, they close their voice queue and shift everything to chat, and overnight agents carry heavier loads because there are fewer of them.
Model | Night team | Overnight channel mix | Overnight result |
|---|---|---|---|
Dedicated night team | Separate pool of night-only agents | Stable at 31% to 41% voice | Matches or beats daytime |
Same team, lighter load | Mostly core-hours agents, about 8x lighter load | Shifts toward chat, its stronger channel | Holds close to daytime |
Skeleton crew | Two to eight agents who also work core hours | Voice drops to near zero | Steepest drops in the dataset |
Voice shutting down overnight
Channel mix explains another share of the gap. Voice outscores chat at four of the five accounts with channel data, and several of them close most of their voice queue overnight. The overnight score then tilts toward the weaker channel.
This overnight drop is the largest single time effect we measured, larger than the day-of-week variation (which is minimal — the anticipated Friday slump doesn't exist in the data) and larger than the seasonal variation (summer runs slightly worse than the rest of the year, but the effect is modest; if you're reading this in fall, you're statistically getting better service right now than you were in July).
What this means for CX leaders
Each time effect in this report maps to a decision a contact center already makes: how it staffs the night, routes work through the day, and plans for weekends and summer.
Build the night shift with purpose. The accounts that hold daytime quality overnight run a dedicated night team or keep overnight queues light. The accounts with the steepest drops cover nights with two to eight agents who also work core hours.
Read overnight quality by channel. At the home furnishings retailer, overnight chat and voice each score close to daytime, and the blended overnight score still drops 0.24 points. Keep voice open overnight, or compare overnight scores channel by channel before acting on a drop.
Cap overnight load per agent. Two overnight agents at the auto accessories retailer carry 1.27 times the core-hours load. The account posts the lowest overnight resolution in the dataset.
Route complex cases to late morning and afternoon. Quality peaks from late morning through the afternoon. Send high-value and complex cases to those windows. Hand routine overnight requests to self-service or AI agents where they exist.
Every customer conversation carries a timestamp, but are you scoring enough conversations to see what that timestamp reveals?
The hour a customer calls changes the experience they get. The season matters modestly. The day of the week barely registers. But the design decisions behind each of those windows — who's staffed, which channels are open, how much work each agent carries — are the levers that actually move the number. Contact centers that score all of them see which hours, days, and seasons set their quality, and fix those first.
