Security testing for websites and apps: what to check and why
A plain guide to security testing: the common weaknesses in web and mobile apps, the checks every product should pass, and when you need a penetration test.

Most security problems are not clever attacks. They are ordinary mistakes: a page that forgets to check who is asking, a password reset link that never expires, an error message that says too much. Security testing is the work of finding those before someone else does.
This guide covers what security testing is, the weaknesses that turn up most often, and what a small team can check for itself.
What security testing is
Functional testing asks "does the product do what it should?" Security testing asks the opposite question: "can the product be made to do what it should not?"
The tester stops behaving like a customer and starts behaving like someone trying to misuse the product: reading data that belongs to someone else, skipping a payment, or getting in without an account.
The weaknesses that turn up most often
The security community keeps a well-known list of the most common risks in web applications, the OWASP Top 10. The details change from edition to edition, but the same themes appear every time.
Broken access control
The product does not properly check what each user is allowed to do. This is the most common serious weakness, and often the easiest to exploit.
- Changing the number in an address such as
/orders/1042to/orders/1043shows someone else's order. - A normal user can open an admin page by typing its address.
- A button is hidden from the screen, but the action behind it still works if called directly.
Weak sign-in and session handling
- No limit on password guesses.
- Password reset links that never expire, or can be used twice.
- A session that stays valid after the user signs out or changes their password.
- Session details placed in the page address, where they end up in browser history and logs.
Injection
The product takes something a user typed and treats it as a command. The classic examples are SQL injection, where input changes a database query, and cross-site scripting, where input placed on a page runs as script in another user's browser.
A quick sign of trouble: typing a single quote or a <script> tag into a field causes an error, or shows up unescaped on the page.
Exposed data
- Sensitive information sent without HTTPS.
- Passwords stored in a form that can be read back.
- An API that returns more than the screen shows: the page displays a name, but the response behind it also contains the email, phone number and address.
- Card details or personal data written into log files.
Misconfiguration
- Default passwords left on admin panels.
- Detailed error pages that reveal file paths, software versions or database structure.
- Test pages, backup files or an open admin tool left on the live server.
- Storage or databases reachable from the internet without a password.
Out-of-date components
Every product is built on libraries and frameworks written by others. When a weakness is found in one of them, a fix is published, and every site still running the old version becomes a target.
Checks every team can do
You do not need to be a security specialist to catch the most common problems. Go through this list on your own product.
Access
- Sign in as user A, copy the address of a private page, sign in as user B, and open it.
- Open pages meant for signed-in users while signed out.
- Open admin pages as a normal user.
Sign-in
- Enter the wrong password twenty times. Does anything slow you down?
- Use a password reset link twice, and use one a day later.
- Sign out, then press Back.
Input
- Put a single quote, a
<script>tag and a very long string into every field, including search boxes. - Upload a file of the wrong type where an image is expected.
Data
- Open the browser's network tab and look at what the responses really contain.
- Confirm every page loads over HTTPS and that
http://redirects.
Setup
- Trigger an error on purpose and read what the page reveals.
- Check that test accounts and default passwords are gone from the live site.
- Check when your libraries and plugins were last updated.
Mobile apps have their own list
- What is stored on the phone, and whether it is readable by other apps or in a backup
- Whether secret keys are built into the app, where anyone can extract them
- Whether the app refuses to talk to a server with an invalid certificate
- What appears in the app switcher screenshot and in notifications on a locked screen
When you need a penetration test
The checks above find the common, visible mistakes. A penetration test goes further: a security specialist spends days attacking the product the way a real attacker would, chaining small weaknesses together, with tools and experience a general tester does not have.
You should arrange one when:
- You handle payments, health records, or large amounts of personal data
- A customer, regulator or standard requires it
- You are about to launch something that would be costly to get wrong
- It has been more than a year since the last one, or the product has changed a lot
Automated scanners are a useful first pass, but they do not replace a person. A scanner can spot an out-of-date library. It cannot work out that your refund process can be used to take money.
Where functional testing and security meet
A good deal of security is found during ordinary testing by a tester who is paying attention. Checking that one user cannot see another's data is part of testing the feature properly. So is checking what the API returns, and what an error message reveals.
We test for the weaknesses in this guide as a service of its own: see security testing. Permission and data checks are also part of API testing, and misuse cases come up naturally in website testing and mobile app testing.
Security is one of the reasons testing has become more important in general. For the others, read why software testing matters more now than it used to.