Do CAPTCHAs Still Stop Fake Signups? What They Catch, What They Miss and What to Add
If fake accounts are creeping into your signup list, a CAPTCHA is usually the first fix people reach for. It is quick to add and familiar. So why do so many sites with a CAPTCHA still end up with throwaway trial accounts, promo-code farmers and spam profiles?
The short answer is that a CAPTCHA answers one narrow question: is this probably a script? A lot of modern signup fraud is about something else. This guide explains what CAPTCHAs are good at, where they fall short, and how to build a layered defence that stops more fraud while asking less of genuine customers.
What a CAPTCHA actually checks
A CAPTCHA is a challenge designed to be easy for people and hard for automated scripts. Older versions asked you to type distorted letters. Newer ones ask you to pick out images, or quietly watch how you interact with the page and only challenge you if something looks off.
Either way, the test is about how the form is being submitted, not who is submitting it. A CAPTCHA has no view on whether the email address is disposable, whether the phone number is an internet-based line bought in bulk, or whether someone claiming to live in Leeds is connecting through a Tor exit node.
What CAPTCHAs are good at
- High-volume scripted abuse. Bots hammering a form thousands of times are exactly what CAPTCHAs were built for.
- Cheap, simple bots. Basic scripts that cannot run a full browser are often stopped outright.
- Raising the cost of an attack. Even when a challenge can be beaten, it adds time and effort for the attacker.
That is real value. For many sites it is worth keeping a CAPTCHA in place. The mistake is assuming it covers everything.
Where CAPTCHAs fall short
They add friction for real people
Every puzzle is a moment where a genuine customer might give up. Image challenges can be awkward on a small phone screen, frustrating for people using assistive technology, and simply annoying when they repeat. You pay that cost on every signup, including all the honest ones.
Determined fraudsters get past them
Challenges can be passed on to paid human solvers or tackled by increasingly capable automated tools. And anyone abusing a free trial by hand simply solves the puzzle themselves, because they are a human.
They say nothing about identity
The fake signups that cost the most are often human-looking: a real person with a disposable inbox, a VoIP number, a datacentre or proxy IP and a password lifted from a breach list. A CAPTCHA waves them straight through, because they behaved like a person. Catching them means looking at what was typed and where it came from, not just how the mouse moved.
A layered approach works better
Rather than relying on a single gate, combine light checks that each catch something different. Most of them are invisible to legitimate users.
| Layer | What it catches | Cost to real users |
|---|---|---|
| CAPTCHA or invisible challenge | Bulk scripted submissions | Low to high, depending on type |
| Honeypot field | Simple bots that fill in every field, including hidden ones | None |
| Rate and velocity limits | Bursts of signups from one IP range or device | None, unless limits are set too tight |
| Email confirmation | Addresses that cannot receive mail | A small extra step |
| Risk scoring at signup | Human-looking fraud: disposable emails, VoIP numbers, proxies, breached passwords, mismatched countries | None when run in the background |
| Human review | Borderline cases that need judgement | None for the user |
The idea is simple: let cheap checks handle the noise, and use smarter checks for the signups that look real but are not.
Matching the defence to your site
Not every site needs every layer. Think about what fake accounts actually cost you.
- Forums and communities. Spam posting is the main risk. A light challenge plus a honeypot handles the bots; risk scoring catches the hand-made spam accounts.
- SaaS free trials. Repeat trial abuse is usually done by people, not scripts, so identity signals such as email age, disposable domains and repeat-signup detection matter far more than a puzzle.
- E-commerce promo codes. Welcome discounts attract people creating account after account. Velocity checks and cross-field consistency are your friends here.
- Regulated services. If you run expensive ID checks later in onboarding, screening the raw signup first can stop you spending money verifying obvious fakes.
Keeping it fair to genuine customers
- Use three outcomes, not two. Allow, review and block gives you a middle ground for mixed signals, so you are not forced to choose between letting fraud in and turning real people away.
- Tell blocked users what to do next. A polite message and a way to contact you turns a false positive into a recoverable moment.
- Decide what happens if a check is unavailable. Failing open (letting signups through) protects conversion; failing closed protects against abuse. Choose deliberately for each form.
- Look at your mistakes. Regularly review the accounts you blocked or flagged. They are the best evidence you have of what fraud on your site really looks like.
Where Veritus fits
Veritus is built for the layer a CAPTCHA cannot cover. It is not designed to replace a CAPTCHA on its own; it complements one, focusing on the human-looking but fraudulent signups that challenges let straight through.
Each signup is checked against seven enrichment signals: email validity and age, phone validity and line type, address validity, IP reputation (including VPN, proxy, Tor and datacentre detection), a password breach check using k-anonymity, cross-consistency between countries, and behavioural signals. A LightGBM model weighs them together and returns a 0–100 score, an allow, review or block verdict and human-readable reasons, in under 100 milliseconds. You can see the full list on the features page or follow a signup step by step in how it works.
A few details that line up with the advice above:
- The drop-in widget collects behavioural signals such as time-to-fill, paste versus type and hidden honeypot interaction. These are widget-only and not available through the API.
- Review and block thresholds are yours to set (by default 50 and 85), and signups in between can go to a review queue.
- The widget fails open by default if Veritus is unreachable, and you can choose fail-closed or fail-to-review per form instead.
- Scoring and storage are UK-hosted.
Try it alongside your CAPTCHA
The easiest way to find out what your CAPTCHA is missing is to score your own signups. The Free plan includes 1,000 checks a month with all seven signals and no card required. Read the Veritus product overview to see whether it suits your signup form.