Share of US customers who left their provider because of poor customer service in 2020: cable 25%, financial 25%, retail 24% — customer loss by industry

1 in 4 cable customers leave because of poor customer service

When a customer leaves a company, people usually look at price or competition for an explanation. Yet a large share of departures has a much simpler cause: the way the customer was treated when they needed help.

A survey of 1,100 consumers conducted in the United States in August 2020, included in Statista's dossier on customer service, measures, industry by industry, the share of customers who left for that reason. The gaps are clear, and the highest figures concern industries where customers do have a choice.

The numbers, industry by industry

Two industries tie at the top: cable and financial or credit services, with 25% of customers leaving because of poor service. General retail follows closely at 24%, then online retail (22%) and telecom or wireless (21%).

Travel sits at 18%. At the bottom of the ranking, big-box electronics stores see only 11% of departures linked to customer service.

In other words, in five industries out of seven, at least one customer in five switched provider because the service experience let them down. These figures date from 2020 and cover the US market.

These figures should be read for what they are: customers stating the main reason they left. They do not say price is irrelevant; they say service alone is enough to tip a significant share of customers over the edge.

Customers who left because of poor service (US, 2020)Cable 25%, financial and credit 25%, general retail 24%, online retail 22%, telecom and wireless 21%, travel 18%, big-box electronics 11%. Customers who left because of poor service (US, 2020) 25% Cable 25% Financial & credit 24% General retail 22% Online retail 21% Telecom & wireless 18% Travel 11% Big-box electronics Source: Statista (2025), Customer service in the U.S. — dossier; 1,100 respondents

Why these industries are the most exposed

Cable, banking, telecom and retail have one thing in common: the customer relationship is continuous. You call your provider about an outage, your bank about a disputed charge, an online store about a parcel that never arrived. Every contact is a moment where the relationship can grow stronger or weaker.

In consumer electronics, purchases are more occasional. Customers need after-sales service less often, so they have fewer chances to be disappointed by it.

The more frequent the contacts, the more each answer matters. And the more records of those conversations exist.

Travel sits in between. Trips are occasional, but when something goes wrong, a cancelled flight or a lost booking, the customer depends heavily on how the company responds. That moment of truth weighs on the decision to book again.

The reason for leaving is already known

This is the key point: the reason for leaving is not hidden. It is written in tickets, said on the phone, repeated in follow-up emails. "Third call about the same problem." "Nobody called me back." "I'll go elsewhere."

These sentences exist in the tools of every company concerned. What is missing is the time to read them at scale. A customer service manager sees a handful of difficult cases each week; they do not see the thousands of conversations where frustration slowly builds.

So the departure looks like a surprise, when it had been announced, often several times.

Read every conversation, not a sample

Automatic language analysis can process every ticket, email and call transcript, and classify each conversation by tone, topic and level of frustration.

For a contact center, that means spotting the topics that come up most in tense conversations, identifying customers who have already chased several times, and tracking how the tone evolves week after week. Our call centers page describes these uses.

Teams can then step in before frustration turns into a cancellation.

Reading everything also changes internal discussions. Instead of debating a few memorable stories, teams share one measure: which topics come up, how often and how strongly. Improvement decisions are then based on facts rather than impressions.

It relies on models such as Mistral, BERT, Llama, GPT or Claude depending on the case, and data stays hosted in France, never used to train models.

Where to start

Take one month of tickets or transcribed calls. Look at the share that expresses frustration, and on which topics. Then compare with the list of customers who left over the same period. The overlap is often telling.

To go further on customers leaving, you can also read our article on reducing customer loss.

Then share the findings with the teams who can act on them: product, logistics, billing. Customer service often sees problems first, but rarely owns their fix. A clear weekly summary of what customers complain about most helps the right people take ownership.

Sources

Lire cet article en français →