Why So Many Internet Customers Call the Wrong Number for Help

Key findings

These figures describe people who reached sales lines that our call center answers. They are not a national rate, and they are not a score for any one provider. The research hub explains how this page sits next to our price research.

How common are support callers on sales lines?

We counted conversations, not every ring. A conversation here means at least 60 seconds on the phone. Short calls, test calls, answering machines, robocalls, wrong numbers, and dead air are left out. The same call is counted once.

An existing customer, in this file, is someone who already had service and called about a bill, an outage, a move, or equipment. The agent then pointed them toward the provider. A new-service shopper was calling to compare or order service they did not already have.

Put together, 226,681 conversations produced a support share of 17.5%. The other 82.5% were new-service shoppers. That split is in the table under “Both logs combined.”

The two logs do not tell the same story. The older call log shows a support share of 8.7% (195,752 conversations). The newer call log shows 73.2% (30,929 conversations). The newer log included more ads that used a provider’s brand name in search. Those ads attract people who already have service and are hunting for help. Because of that mix, the 8.7% share and the 73.2% share are not directly comparable. Neither one is the share of all United States households.

A blank cell would mean the count was too small to publish. None of the rows in this table were blanked. Counts from 1 to 10 are hidden, and a rate is shown only when it rests on at least 30 conversations.

How long people stay on the wrong line

The median time on the phone for a support caller, across both logs, was 3.4 minutes. The median is the middle call: half were shorter and half were longer. That figure covers 39,660 support conversations, which matches the count of existing-customer conversations in the share table.

Each log has its own middle. On the older log the median was 2.3 minutes. On the newer log it was 4.6 minutes. The 25th and 75th percentiles show the spread. A caller near the 75th percentile on the newer log was on the phone for 8.5 minutes before the redirect.

Three or four minutes feels short until you are in an outage. Use the first sentence well. Say that you already have service and that you need the provider’s support team. If the person who answered sells new service, ask them for the support path and then hang up. Do not read a password, a one-time code, or a full card number to a sales line.

Why it happens

The usual path is a search. You type the provider’s name plus “customer service,” “outage,” or “phone number.” The page that ranks first is not always the provider. It can be an ad, a directory, or a site whose name only looks official. The number in that result rings a sales line.

Brand-name ads make this worse. An ad can show the provider’s name because that is what you typed. The click still goes to a company that sells service, not to the team that fixes the line you already pay for. The newer call log in this study carried more of those ads, which is why its support share is so much higher. Read that gap as a change in who found the number, not as proof that support got worse.

Look-alike names are the second trap. A page can add “help,” “support,” or “billing” next to a provider’s name and still be a shopping site. A phone number printed in a large font is not proof. The proof is that the same number appears on your bill or inside the account you opened on the provider’s own site.

A third trap is speed. During an outage you want any human. The first number wins. That is understandable, and it is also how a support need becomes a sales call. If you can reach the provider’s app on cellular data, do that before you dial. Our guide to what to do in an internet outage starts with the provider’s own status tool.

How to find your provider’s official support

This page does not list provider phone numbers. A number printed here would go stale, and a wrong one would repeat the problem. Use a source that already knows your account.

1. Open a paper bill or the PDF in your email. The support number is usually on the first page or in the footer, next to the account number.
2. Open the provider’s app while you are signed in. Use the help or support area inside the app. Outage tools in that app know your address.
3. Sign in on the provider’s own website. Type the address yourself or use a bookmark you saved earlier. Do not tap the first search result if it is labeled as an ad.
4. If you cannot sign in, use the website address printed on the bill, then look for “support,” “contact,” or “outage” on that site.
5. When you do call, have the account number, the service address, and a one-line description ready. Example: “Internet has been down since 7 p.m. at this address. I need support, not a new plan.”

If the line is down and the Wi-Fi is down with it, use cellular data for the app. A phone can also carry a short call or a payment while you wait. Keep cloud backups paused so they do not eat the cellular allowance.

Equipment questions belong on that same support path when the provider owns the gateway. If you own the router, the provider can still test the line to the gateway, and you can test the rooms yourself. Why your Wi-Fi is slow separates a line fault from a room that the router does not cover.

How to avoid look-alike numbers

Save the support number from the bill into your phone contacts, under the provider’s name plus the word “support.” The next outage is then a contact you already checked, not a fresh search.

On a results page, skip rows marked as ads when you need the account you already have. An ad can be a fair place to shop for new service. It is a poor place to report that the modem is dark. Read the site name in the address bar before you dial. If the name is not the provider’s, treat the number as a sales line until the person who answers says otherwise.

Watch for extra words. “Billing services,” “rewards,” and “plan help” next to a famous name are a hint to slow down. So is a number that appears only on social media. Ask a household member to read the number on the bill out loud and compare it to the one on your screen.

If you already called and the person started a new-service pitch, you can stop. You do not owe them an application, a credit check, or a deposit to get a redirect. Say you need the provider who already serves the address. Then use the bill or the app.

Shopping for new service is a different job. When you do want offers, use the address check on this site rather than a number you found beside the word “support.” People who are choosing satellite plans can read what real satellite orders looked like. People worried about a credit check or a deposit can read how often those steps ended the call.

How home internet reaches a house is the plain-language tour of fiber, cable, fixed wireless, and satellite. It will not give you a support number either. It will tell you which kind of line you are trying to get fixed, which helps when you do reach the right team.

Methodology

This study uses multi-year call records from our call center, where agents help shoppers compare and order home internet from many providers. The question here is narrower: how often did an existing customer land on a sales line while looking for help, and how many minutes passed before they were redirected.

Two phone systems produced two logs. The older log runs from January 2018 through July 2020 and holds 195,752 conversations. The newer log runs from August 2023 through April 2025 and holds 30,929 conversations. Together, from January 2018 through April 2025, there are 226,681 conversations. Dates stay in this section so the findings above can be read without a year attached.

The newer log included more brand-name search ads. Those ads pull in existing customers. The support share of 8.7% and the support share of 73.2% are therefore not directly comparable. The combined share of 17.5% describes these logs, not the country.

Before any count, names, phone numbers, email addresses, street addresses, account numbers, and agent identities were removed or replaced with tokens that cannot be reversed. No recordings were used. We publish totals only.

A share is the conversations in a group divided by all conversations in that log, rounded to one decimal. The median minute figure of 3.4 is the middle support call across both logs. The per-log medians and the 25th and 75th percentiles are in the minutes table. Calls about products other than home internet, such as electricity or gas plans, are excluded.

Small numbers stay hidden. Any count from 1 to 10 is blank. A percentage or a median is published only with at least 30 records. If hiding one cell would let a reader solve it by subtraction, a second cell is hidden too. Geography, when we publish it, is state level only. This page does not break the figures out by state.

Agents recorded how each call ended, and people make mistakes. The two phone systems did not use identical lists of endings, so a category that exists on only one system stays with that system’s total. These were sales lines. The share shows how callers found the numbers, often by searching a provider’s name.

InternetService.ai may earn compensation when you order through links or the phone number on this site. That does not change the counts or the shares.

You may reuse the tables with credit to this page. The license we recommend for the download is Creative Commons Attribution 4.0. Read the license.

Download the tables

• Support share by call log (CSV)
• Minutes on support calls (CSV)

Our editorial methodology covers how we date a source and how to send a correction. Editorial standards name who reviews the guides.