The Phone Step Is the Shortest Part of Signup. Why Do So Many Users Quit There?
A phone number takes five seconds to type. Verifying it can take five minutes. The code arrives late, or the form rejects the country code, or the user switches apps to read the SMS and never comes back.
On paper, the phone step is the shortest part of signup, but in practice it is one of the most fragile. It depends on things your product does not control, starting with a carrier in another country that decides whether your message gets through.
Short answer. Phone number verification confirms that a user controls the number they entered. To reduce drop-off at this step, remove typing and waiting. Detect the country automatically, let the device autofill the code and offer a second channel when SMS fails. Ask for the number only when the user sees why you need it. A phone check proves control of a number, not identity, so higher-risk flows add a document or database check.
Why the shortest step loses users
Phone verification fails in four different places. In your funnel, all four look the same: a user who stopped. Each needs a different fix.
Delivery. International SMS passes through several networks, and carriers filter messages they suspect are spam. A code can arrive late or not at all.
A code that takes 40 seconds instead of five has not technically failed. The user has: they tap "resend", get two codes, enter the wrong one and give up.
Input. The form rejects a valid number because of a missing country code, a leading zero or spaces. Or it accepts an invalid one and sends a code into the void.
Context switching. To read the SMS, the user leaves your app. Every app switch is a chance to get distracted, and on some devices coming back to a web form reloads it.
Timing. The form asks for the phone before the user has seen anything worth giving a number for. That is a motivation problem, and no delivery provider can fix it.
Four ways to verify a phone number, compared
Four methods are in common use, and they differ in what the user has to do, what you pay for and where each one tends to fail.
Method | What the user does | Cost driver | Weak spot |
|---|---|---|---|
SMS one-time code | Types the code, or the device autofills it | Per-message fees, varying by country | Slow delivery, SIM swap, interception, SMS pumping |
Voice call or flash call | Answers a call, or the app reads the missed call | Per-call fees | Flash calls unsupported in some markets; hard for hearing-impaired users |
WhatsApp or RCS code | Opens the message in a messaging app | Per-conversation or per-message fees | Needs the app installed |
Silent network authentication | Nothing: the carrier matches the number to the SIM | Per-check fees from carriers or aggregators | Mobile data and supporting carriers only |
A flash call is a missed call: the caller's number carries the code, and the app reads it. RCS is the newer rich-messaging standard for phones. SIM swap, interception and SMS pumping are attacks, explained in the security section below.Silent network authentication removes the code entirely, so it deserves a closer look. Mobile operators and industry initiatives now offer it as a standard API. It lets your app ask the mobile network one question: does the number entered match the SIM in this device? The answer is yes or no.
Carriers in some markets have started offering it together. In September 2025, five Malaysian operators announced shared access for banks and online retailers. Coverage is still patchy, so most teams treat it as a first attempt with SMS as the fallback.
Mixing channels also helps. According to Twilio, Duolingo saw a 12% better verification success rate in a two-week test across 12 countries. It sent codes over WhatsApp first and fell back to SMS.
Twilio ran the test and sells the product, so read the figure as a direction only.
Ten fixes that cut drop-off at the phone step
Most of these take a day of work or less, and none requires changing your verification provider.
1.Detect the country automatically from the device locale or IP address, and show the flag and dialing code before the user types.
2.Validate the number before you send a code. Google's open-source libphonenumber library catches typos and landlines. Store the result in E.164, the international format: a plus sign, the country code and the number, up to 15 digits.
3.Let the device fill in the code. On the web, add autocomplete="one-time-code" to the input, and Safari on iOS 14+ offers the code from the SMS. Chrome on Android does the same through the WebOTP API, but only if the message ends with a line like @yourdomain.com #123456.
4.Do the same in native apps. Android's SMS Retriever API reads the code without SMS permission if the message includes your app's 11-character hash. iOS uses the .oneTimeCode content type.
5.Hide the resend button behind a countdown. A visible 30–60 second timer stops users from requesting three codes and entering the wrong one.
6.Offer a second channel after the first resend: a voice call, WhatsApp, or email where your risk policy allows.
7.Say what went wrong in plain words. "This number looks incomplete" helps. "Invalid input" does not.
8.Keep the session alive while the user is in their messages app, so they never return to an empty form.
9.Ask at the moment of value. Request the number when the user is about to do something that needs it, such as a first deposit, where your regulator allows.
10.Let users fix the number on the code screen without starting over.
How to measure the phone step
A single "phone step conversion" number hides which of the four failures you have. Track five numbers separately:
•Delivery rate: codes delivered as a share of codes sent, by country and carrier where your provider reports it.
•Code entry rate: users who entered a code as a share of users who received one.
•Time to verify: median and 90th percentile, from "send" to "verified".
•Resend rate: a high rate usually points to slow delivery.
•Fraud rate on verified numbers: accounts later flagged for fraud. This is the number that keeps the fraud team on your side.
When delivery rate is high and code entry rate is low, the problem is UX or motivation. When delivery rate is low in specific countries, the problem is routing, and a second channel will help more than a redesign.
How attackers get around an SMS code
An SMS code proves that someone could read a message sent to a number. It does not prove they hold the SIM, or that they are the person who owns the account.
SIM swap. A fraudster convinces a carrier to move the victim's number to a SIM they control, then receives the victim's codes. This fraud grew fast. In 2021 alone, the FBI's Internet Crime Complaint Center recorded 1,611 SIM swap complaints, with adjusted losses over $68 million.
For 2018–2020 combined, it recorded 320 complaints and about $12 million in losses. In November 2023 the FCC, the U.S. telecom regulator, adopted rules that make carriers verify customers before a SIM change. Carriers must also notify customers of the request and offer a free SIM-change lock.
Our glossary also covers SIM cloning, a related attack.
Interception. SS7, the decades-old signaling system between telecom networks, has weaknesses that let attackers redirect messages. In 2017, O2-Telefónica in Germany confirmed that attackers had used SS7 access to intercept banking SMS codes and drain customers' accounts. Read more about OTP interception.
SMS pumping. Fraudsters use your phone number field to trigger large volumes of codes to premium-rate number ranges. A complicit operator shares the revenue with them, and you pay for every message.
Enea, a telecom security vendor, and the research firm Mobilesquared estimated that this artificially inflated traffic cost brands about $1.16 billion in 2023. Treat that as a rough vendor estimate, because the market has no agreed definition of this fraud. The problem reaches large platforms too: in February 2023, Elon Musk publicly endorsed a claim that Twitter was losing about $60 million a year to fake two-factor SMS.
Disposable and virtual numbers. Many signup abusers use numbers that exist only to receive one code. Line-type checks, which tell mobile, landline and virtual numbers apart, catch some of them, but none catch all.
How NIST rates SMS codes
NIST, the U.S. National Institute of Standards and Technology, puts SMS and voice codes in a lower tier it calls "restricted." Its current digital identity guidelines, finalized in July 2025, list the public telephone network as the only restricted authenticator (SP 800-63B-4). Organizations may still use it, subject to extra requirements in the guidelines (section 3.2.9).
In practice, that tier is usually acceptable for signup. For account recovery or approving a large withdrawal, it often is not.
When a phone number is not enough
The right check depends on what the account can do. A phone number is a reasonable gate for a free trial and a weak one for a wallet that can send money abroad.
Risk level | Example | Typical checks |
|---|---|---|
Low | Newsletter, free trial | Phone or email only |
Medium | Marketplace seller, betting account with deposit limits | Phone + ID document + face match |
High | Payment account, crypto withdrawals, high-limit betting | Phone + document + liveness check[/liveness-check/] + AML screening |
A liveness check confirms that a live person is in front of the camera. AML (anti-money laundering) screening checks the user against sanctions and watch lists. For medium and high risk, phone verification becomes one step inside a full identity verification flow.
What a phone check cannot tell you
A verified number tells you that someone had access to a SIM or a messaging account at one moment. It does not tell you who that person is, whether they are old enough or whether they are on a sanctions list.
It also cannot show whether they will still control the number next month. Treat it as a contact detail you have confirmed, and nothing more.
How KYCAID handles phone verification
KYCAID's mobile number verification sends a one-time SMS code inside the same flow as the document and face checks. The user sees one process instead of three. The check is typically completed within a minute.
In countries with registry checks, you can combine the phone step with a national ID lookup.
Phone verification is only one part of customer due diligence. It confirms a contact channel, not who the customer is or how much risk they bring. In a full KYC process, KYCAID adds document verification, a liveness check and AML screening against watch lists and adverse media. For higher-risk customers, the flow can also include proof of address, source of funds checks and ongoing monitoring after onboarding.
FAQ
🔹What is the best way to reduce signup drop-off with phone verification?Remove typing and waiting from the phone step. Detect the country, validate the number before sending and let the device autofill the code. When SMS is slow, offer a second channel.
🔹Is SMS OTP secure enough for KYC?
SMS OTP, a one-time passcode sent by text, is secure enough to confirm a contact number at signup, but it cannot prove identity. NIST rates it a restricted authenticator because of SIM swap and interception risks. Regulated KYC (know your customer) onboarding therefore adds document, face or registry checks.
🔹What is OTP interception?
OTP interception is when an attacker receives a one-time code meant for someone else. They do it through SIM swap, malware on the phone, or weaknesses in the SS7 telecom signalling network.
🔹Why don't SMS verification codes arrive?
SMS codes usually fail to arrive because of carrier filtering, international routing delays, wrong number formats or a phone without signal. A delivery report from your provider shows which one it is.
🔹Should you verify the phone number before or after signup?
Ask for the phone number when the user reaches the feature it protects. Some regulators set an earlier point. Where they do not, asking before the user sees value increases abandonment without improving security.