Comcast Counted 30 Million Conversations That Didn't Happen. Verizon Counted Who Stayed. AT&T Put a Clock on the Phone.
· CX Pulse
Three carriers reported one quarter inside six days: Comcast counted the calls that didn't happen, Verizon counted who stayed, AT&T put a clock on the phone.
Comcast Counted 30 Million Conversations That Didn't Happen. Verizon Counted Who Stayed. AT&T Put a Clock on the Phone.
Three carriers reported the same quarter inside six days, and each one put a different number on the customer relationship. On Wednesday, October 21, AT&T reported 379,000 postpaid phone additions, 0.92% churn, and a guarantee it had expanded the week before: reach a person in five minutes or get a callback. On Thursday, Comcast's CFO Jason Armstrong told analysts the company had "reduced customer service contact rates by roughly 20% and eliminated more than 30 million service interactions year-to-date versus the prior year." On Monday, Verizon reported its best third quarter for phone additions in five years and consumer churn of 0.87%, and said its promotional spend on keeping customers had fallen 30%.
Fewer calls. Fewer departures. A faster answer. Read together, they're three different theories of what a service number is for.
What each number counts
Comcast's number counts conversations that didn't take place. The company credits a simpler product lineup, fewer promotional price expirations, and AI partnerships with Sierra and Cresta for the drop. Mike Cavanagh said NPS is "improving both sequentially and year-over-year," without a figure. The same quarter, Comcast lost 144,000 broadband customers and 274,000 video customers, which the release attributes to "higher churn from pricing transitions and competition from fiber and fixed wireless." Armstrong also said the quarter had "a higher than typical volume of customers facing pricing transitions across our legacy broadband customer base."
Verizon's number counts the customers who stayed. Dan Schulman's line in the release: "By prioritizing customer-first offerings like Simplicity and our expanded loyalty program over heavy subsidies, we are winning loyal, high-value customers while expanding margins, reducing churn, and generating strong sustained cash flow." The retention figure is the cost of keeping someone, down 30% from a year ago, alongside 137,000 net new accounts, the best in over a decade.
AT&T's number is a clock. Its October 15 release promises an answer within five minutes or a callback at a time the customer picks, one technical expert from start to finish for customers who pair fiber or Internet Air with wireless, and a benefit if AT&T misses. The release cites a survey in which 60% named transfers as a top frustration and more than 75% said they won't wait past 15 minutes. John Stankey's earnings quote names the guarantee as the thing the network is "backed by."
A contact that didn't happen has three explanations
The problem never occurred. The customer gave up before reaching anyone. Or the customer left. A contact rate counts all three the same way, because all three produce the same absence. Comcast's explanation is the first one, and it may well be right. The number can't show it, because the number is a count of silence.
Verizon's figure can at least rule out the third explanation: a churn rate is a customer who stayed or didn't, and nothing in between. AT&T's figure is the only one of the three a customer can check from the outside. Dial the number, start a timer. Either a person picks up inside five minutes or the guarantee owes you something. Nobody outside Comcast can verify a 20% contact-rate drop, and nobody inside it can tell from the rate alone which of the three explanations earned it.
At scale
Contact rate is a cost number, and it's a fine one. It becomes a problem when it stands in for a service number on a slide. The pairing that keeps it honest is cheap: report contacts avoided next to churn among the customers who had an open issue in the prior 90 days. If deflection is working, that churn holds or falls. If the queue got quiet because people stopped trying, it rises, and the two lines on one chart say so.
Then sample the silence. Pull 100 sessions that ended in self-service or a bot without a human, and find out what happened next: did the issue recur, did the customer call back on another channel within a week, did the account close within 90 days? That audit takes an analyst a day, and it answers the question the rate can't.
A clock is worth more than a satisfaction score, and it has a price. AT&T has promised something it will miss on bad days, and it has written down what the miss costs. That's the whole discipline: a public threshold, a remedy that fires without a claim form, and one team that owns the number.
For smaller operations
A quiet inbox is the same ambiguity at a smaller scale. Fewer emails this month means fewer problems, or fewer customers, or customers who stopped bothering. The check takes three lines: how many customers contacted you, how many cancelled or didn't come back, and how long the last ten waited for a reply. Any month the first line falls, look at the other two before calling it good news.
The five-minute idea sizes down too. Most small businesses can't staff a five-minute phone promise, and shouldn't claim one. Most can promise a reply by the end of the day and name what happens when they miss it. That's a smaller clock, and a customer can still read it.
A customer who stopped calling and a customer who stopped having problems produce the same contact rate. What separates them is whether they're still a customer, and whether a person answered when they tried.
Sources: Comcast Q3 2026 earnings call transcript (October 22, 2026) · Comcast Q3 2026 earnings release (SEC) · Verizon Q3 2026 earnings release (SEC, October 26, 2026) · AT&T Q3 2026 earnings release (SEC, October 21, 2026) · AT&T, "AT&T Expands America's Best Guarantee" (October 15, 2026)