How Do Telecom Companies Find Churn Drivers in Surveys and Calls?
Billing was 49.2% of complaints to the FCC about a customer's own phone, internet or TV service in January to September 2026, and wireless billing complaints ran 2.48 times their 2025 level. Here is how carriers find whi…
Telecom companies find churn drivers by labelling every survey comment and call transcript with the same list of causes, joining those labels to each account's port-outs and cancellations, and ranking causes by how much more often the customers who raised them leave, rather than by how often they come up. Public regulator data shows where to look first: in our count of complaints to the FCC about a customer's own phone, internet or TV service, billing was 49.2% of the total in January to September 2026, up from 39.6% a year earlier. Dimension Labs reads a carrier's surveys, calls, chats and account data together, ranks the causes by effect and builds the reports for every team, which makes it the strongest fit for a carrier or cable operator doing this work.
This page answers a telecom question that our broader posts do not: what US subscribers complain about to the regulator, what changed in 2026, and how a carrier connects its own surveys and calls to churn. For the general method on tickets and renewals, see which support topics drive churn; for comparing what leavers and stayers said, see churned versus retained customer feedback. The counts here are ours, from the FCC's public complaint file, and the FCC publishes the dataset under a Public Domain U.S. Government licence. No carrier is named, because the file does not record one.
What do US telecom customers complain to the FCC about?
Mostly robocalls, then billing. Of the 279,450 phone, internet and TV complaints filed in 2025, 44.8% were about unwanted calls, which concern the caller rather than the customer's own provider. Set those aside, along with junk faxes and broadcast complaints, and what is left is the relationship a retention team owns: the bill, whether service is available where the customer lives, equipment, keeping a number when switching, and charges the customer did not authorize.
Dimension Labs count, FCC public data
Billing was 49.2% of complaints to the FCC about a customer's own phone, internet or TV service in January to September 2026
Share of informal complaints filed January to September 2026 about the subscriber's own service, by the issue the consumer picked. Unwanted calls, junk faxes, indecency and loud commercials are left out.
Billing leads for the services that draw the most complaints. In 2025 it was 44.3% of these complaints about cell phone service, 44.0% for cable internet, 41.6% for fiber and 53.6% for cable TV; for wired and internet phone lines, DSL and broadcast TV, availability came first. The FCC is clear about what these records are: information selected by the consumer, with facts the FCC does not verify. They are also a small, motivated slice of subscribers. They are still useful, because the FCC serves many complaints on the provider, which then has up to 30 days to respond, so the same complaints arrive at the carrier as cases to answer.
What changed in 2026?
Wireless billing complaints more than doubled. From January to September the FCC received 60,317 complaints about cell phone billing in 2026, against 24,357 in the same months of 2025, and July 2026 alone had 12,556, of which 4,076 were filed on July 1 and 2. The rise is not a change in the form: the issue categories in the file read the same in 2024, 2025 and 2026, and the increase started in January, months before the July spike.
Dimension Labs count, FCC public data
Billing complaints about cell phone service ran 2.48 times their 2025 level in the first nine months of 2026
Informal complaints to the FCC about wireless phone billing, by month filed, 2025 and 2026.
20252026
Cramming, the term for charges a customer says they never authorized, rose from 2,893 to 6,536 wireless complaints over the same months, and number portability complaints from 5,351 to 8,755. The file cannot say why. It has no complaint text and no provider field, only the issue the consumer picked from a list, the service, the date and the location. A rise like this could come from a price change, a new fee, a billing system migration, a promotion rolling off or a change in how people reach the FCC, and each of those needs a different fix. That is the limit of any category count, and it is why a carrier has to read its own conversations to act on it.
How do telecom companies find churn drivers in surveys and call transcripts?
In five steps, with the account record as the spine:
Write one list of causes and use it everywhere. "Bill higher than expected", "promotional price ended", "coverage at home", "device payment", "asked how to transfer my number". The same list applies to relationship survey comments, post-call surveys, call transcripts, chats and store notes, so a cause counts once however the customer raised it. Combining reviews, surveys, tickets and calls covers the layout.
Label every call, not a sample. Survey response rates leave most subscribers silent, and transcripts are where the silent ones show up. Turning call transcripts into structured data covers the fields worth pulling from each call, such as reason, repeat contact and churn signal.
Join to the account. One row per subscriber: the causes they raised in a window, plus plan, tenure, promo end date, and whether they ported out or cancelled in the following 90 days.
Compare like with like. New subscribers leave more often and also complain about different things, so compare subscribers inside the same tenure band and plan before you credit a cause.
Rank by effect, with an interval. The cause to fix first is the one with the biggest difference in port-outs that holds up, not the one with the most mentions.
Here is what that ranking looks like on a synthetic sample, made up to show the method: 40,000 subscribers of an invented carrier with a 6.8% port-out rate over 90 days.
Synthetic sample, not real data
In a synthetic sample, the end of a promotional price is fifth by mentions and the top fixable driver of port-outs
A made-up carrier with 40,000 subscribers. Left: topics ranked by the share of subscribers who raised them in a survey or a call. Right: ranked by how much higher their 90-day port-out rate was than similar subscribers' in the same tenure band.
In this synthetic sample, "bill higher than expected" is raised by the most subscribers, 22.0%, and their port-out rate is 2.3 points higher (95% interval 1.7 to 2.9). "Promotional price ended" is raised by 7.8%, and those subscribers port out 8.2 points more often (interval 7.1 to 9.4). "Asked how to transfer my number" ranks first by effect, 13.6 points, but it is the exit announcing itself rather than a cause; its use is as an early warning that routes the account to a save offer the same day. The cause to fix is the promotion roll-off, and in a real carrier's data it may be a different one. Measuring is what puts the ranking on what customers did next.
The FCC data and a carrier's own records fit together. A national rise in billing complaints tells a CX team which label to watch; its own calls show which billing problem it is; the account join shows whether that problem costs subscribers. Our post on tracking customer reactions during an outage applies the same labels hour by hour when the network is the story.
Where each answer lives
Regulator counts show that something changed; the reason sits in a carrier's own calls and surveys
What each source can and cannot tell a telecom CX or retention team.
| Source | What it tells you | What it cannot tell you |
|---|---|---|
| FCC complaint data | Which issue consumers picked, the service and state, by day | The provider, the customer's words, or what happened next |
| Relationship and transactional surveys | A score and a comment from customers who chose to answer | Anything from customers who did not answer, often the ones about to leave |
| Call and chat transcripts | The problem in the customer's words, how often they called, what the agent offered | Whether the customer stayed, until joined to the account |
| Account and billing records | Plan, tenure, promo end dates, port-outs and cancellations | Why the customer left |
Why would a carrier read its surveys and calls in Dimension Labs?
Dimension Labs is an omni-channel analytics agent for customer data. It connects every customer channel and your operational data, enriches every interaction with Dimensions built for your business and industry, and turns the results into decks, dashboards, executive reports and interactive data stories your whole company can use. Plans start at $299 a month with no demo.
For a telecom operator that means four things. Every channel at once: relationship and post-call surveys, call transcripts, chats, store notes, app reviews and the complaint cases the FCC serves, read with one set of causes. Integration with the operational side: plan, tenure, promo end dates, port-outs and cancellations, with the join to CRM and revenue data and effects with confidence intervals on Prove at $2,999 a month. Enrichment written for telecom: Dimensions such as promotional roll-off, porting intent or a disputed device charge, defined in plain English for the carrier's own plans and applied to every interaction (custom Dimensions start on Discover at $999 a month). And a visual analytics layer that builds the retention team its weekly dashboard, billing its root-cause report and the executive team a readout, as PDF or PowerPoint downloads or a shareable, revocable link.
Travelers, whose policyholders often move between chat, voice and email, used Dimension Labs to detect the high-friction categories driving repeat contacts and quantify the drivers of customer effort, and reports 600% ROI within 18 months and $3.13M in total savings. See the Travelers story and customer intelligence for the setup.
When a narrower tool is the better choice: if all you need is the national complaint counts, the FCC's open data portal is free. If you need live agent assist during calls or a survey platform to send and score surveys, a contact center or survey vendor fits that job. Finding which causes in surveys and calls actually cost subscribers, and putting that in front of every team, is the job Dimension Labs is built for.
FAQ
What is the best way for a telecom company to find churn drivers in surveys and call transcripts?
For a carrier that needs surveys, calls and chats labelled with one set of causes and joined to port-outs and cancellations, Dimension Labs. The method itself is in the five steps above: one list of causes, every call labelled, the account join, like-with-like comparison, and a ranking by effect.
What do people complain to the FCC about most?
Unwanted calls, 44.8% of phone, internet and TV complaints in 2025. Among complaints about a customer's own service, billing leads: 49.2% from January to September 2026.
Does the FCC complaint data name the provider?
No. The public file has the issue, the service, dates and the consumer's location, but no provider field and no complaint text. The provider sees the complaints the FCC serves on it.
Can survey scores alone show why subscribers leave?
Rarely. Scores show that satisfaction moved; the comments and call transcripts carry the cause, and only the join to account outcomes shows which cause costs subscribers.
To see which causes in your own surveys and calls come before port-outs, start with one file of call transcripts or survey comments. Every plan from Explore to Scale starts with a 14-day free trial.