- Security needs continuous attention because systems, staff and threats keep changing
- MFA greatly reduces the risk of account takeover, but its strength depends on the method used
- Exploited vulnerabilities in internet-facing systems are a major entry point; prioritise by exposure and known exploitation
- Standard accounts limit what malware can change, not the files a user can already reach
- Company size alone does not determine risk; exposure, credentials and the value of your data and services matter more
This article describes what commonly goes wrong and how to recognise it. For what to implement, see our cybersecurity baseline for SMEs. The seven mistakes are not ranked by frequency: treat them as recurring weaknesses worth reviewing in your own environment.
1. Treating security as a one-time project
A common pattern in small business cybersecurity is the "project" mindset: a burst of activity, a set of tools installed, and then an assumption that the job is done. A firewall was configured two years ago. Antivirus is running. Passwords were changed after the last scare. These are useful steps at a fixed point in time, but the environment around them keeps changing.
New vulnerabilities are discovered in software every week. New phishing techniques emerge as old ones are filtered by email gateways. The systems in your business change too: new staff join, new cloud services get added, old laptops are replaced without being properly decommissioned. Each of these changes can open a gap that did not exist when you last reviewed your security posture.
A defined review cycle handles this without a large team: a quarterly check of user accounts and access, a monthly confirmation that patches are being applied, and an annual review of your overall controls. Part of that cycle is keeping an up-to-date inventory of devices, cloud services, accounts and internet-facing systems, because you cannot patch, protect or recover what you do not know exists. A modest rhythm that is actually followed is usually worth more than a large one-off investment that is never revisited.
How to recognise it: nobody owns security as an ongoing task, and the date of the last review is unknown or years old.
2. Relying on passwords alone
Passwords can be stolen, guessed, reused and leaked without the account holder knowing. Leaked credential databases from unrelated breaches circulate online, and automated tools test username and password combinations against business email, cloud storage and remote access systems, often soon after a credential database is published.
Multi-factor authentication (MFA) changes the risk of account compromise: a correct password on its own is no longer enough to sign in. A 2023 Microsoft study of commercial Azure Active Directory accounts that showed suspicious activity (Meyer et al., 2023) estimated that MFA reduced the risk of compromise by 99.22% across the population studied. That is a strong result, not an absolute guarantee. Protection depends on the method: SMS codes, authenticator app codes (TOTP) and some push notifications can be captured or approved through real-time phishing, while FIDO2/WebAuthn methods such as passkeys and security keys are bound to the genuine site and are much more resistant. Our article on how attackers get around MFA explains these techniques and the defences against them.
Initial enablement can be quick on most major platforms. A secure rollout takes longer: it has to cover every user (senior staff included), every application (including legacy protocols that bypass MFA), the methods you allow, exceptions, account recovery and emergency access accounts. Start with email, because it is the password reset route for most other services, then remote access, cloud storage and financial or payroll platforms. Our MFA guide covers methods and rollout in more detail.
How to recognise it: some accounts, often administrators, executives or shared mailboxes, still sign in with a password only, or nobody can list the MFA exceptions.
3. Ignoring software and firmware updates
Vulnerabilities in internet-facing systems are an important attack vector, especially those already known to be exploited. In Verizon's 2026 Data Breach Investigations Report, exploitation of vulnerabilities was the most common initial access route in its dataset, at 31% of breaches, ahead of credential abuse and phishing. Once a fix is published, attackers can study it too, and scanning for systems that have not applied it can start quickly.
Many SMEs run software that has been waiting for updates for months or years. Updates take time, sometimes break things, and patching is rarely anyone's explicit job. Prioritise by risk rather than trying to do everything at once: internet-facing systems first (VPNs, firewalls, remote access, email and web servers), vulnerabilities listed in CISA's Known Exploited Vulnerabilities (KEV) catalogue next, then the rest, taking exposure and existing mitigations into account. Not every exploit depends on a missing patch: misconfigurations, default passwords and unsupported products that will never receive a fix need attention too.
Firmware is often overlooked. Routers, switches, firewalls and storage devices receive security updates just as operating systems do. Old firmware, or a device the vendor no longer supports, is a warning sign worth investigating, not proof of compromise. Checking network hardware for updates and support status each quarter, alongside software patching, closes a category of exposure that is easy to forget.
How to recognise it: nobody can say which systems are reachable from the internet, when they were last updated, or whether any of them are out of vendor support.
4. Giving everyone administrator access
If every member of staff has administrator rights, a single compromised account can install software, disable protection and change security settings, and depending on its privileges it may reach other systems too. Removing local admin rights reduces what malicious code can change on the device. It does not stop everything: ransomware running under a standard account can still encrypt the files that user can access, including synced folders and network shares.
Administrator access allows software to be installed, system settings to be changed and security controls to be modified. In a small business where everyone installs software freely, granting admin rights feels like the path of least resistance. In practice, malware inherits the access of the account it runs under, so excessive rights widen the potential reach of a compromise.
The principle of least privilege is the right approach: each account should have only the access it needs for the job it is used for. Standard user accounts for day-to-day work, and a separate administrator account used only when elevated permissions are genuinely required. Most business software runs well under standard permissions. Software that insists on administrator rights should prompt a question about whether that is really needed.
Least privilege works best as one layer among several: tight permissions on shares and cloud folders, network segmentation, application or execution controls, backups that cannot be altered from a user's account, and monitoring for unusual activity. A separate admin account is a useful step. On a handful of machines it can be set up quickly; across a larger fleet, with software that expects admin rights and staff who need exceptions, plan it as a small project.
How to recognise it: most users can install software themselves, or the same account is used for email and for administration.
5. Having backups but never testing them
Many businesses hit by ransomware did have backups. The question that matters is whether anyone had tested that those backups could actually be restored. Backup software runs quietly in the background, logs report completion, and it is easy to assume the data is usable until a real recovery is needed.
Backups fail in ways that logs do not always show. Files backed up while open in another process can be corrupted in the backup. Cloud backup configurations set up once and never reviewed often miss folders, drives or systems added since. Backup media can fail silently because nobody has tried to read from it.
Match restore tests to how critical each system is. For ordinary files, restore a sample to a test location and confirm they open. For critical systems, go further: restore databases and check they are consistent, confirm authentication still works, check integrations with other systems, and confirm the service actually runs for users. Record how long the recovery took and compare it with the business's recovery time objective (RTO) and recovery point objective (RPO): how long you can be without the system, and how much data you can afford to lose. Quarterly is a reasonable illustrative cadence; critical systems may need more frequent tests, low-risk data less. Documented results also help with insurers and, where relevant, regulators.
An untested backup has unverified recoverability. You only know it works once you have restored from it.
How to recognise it: nobody can say when a restore was last tested, how long it took, or whether that time is acceptable to the business.
6. Not training staff to recognise phishing
Phishing remains an important way in, but it is not the only one. In Verizon's 2026 report it accounted for a smaller share of initial access than vulnerability exploitation, yet it remains one of the main routes alongside credential abuse. It works because it targets people rather than systems: a convincing email only needs someone to click a link, open an attachment or enter credentials on a fake login page. Email filtering stops a large proportion of attempts, but not all, and the messages that get through tend to be the most convincing.
Training works better as a habit than as an annual event. Short, regular sessions can be combined with simulated phishing exercises, provided they are proportionate and educational: the aim is to find where people struggle and help them, not to catch them out. Measure more than clicks: how many people report the message, how quickly, and whether staff verify unusual requests (such as changed bank details) through another channel. Individual simulation results are personal data under UK GDPR, so tell staff how results are used, limit who can see them, and do not use them for disciplinary purposes without a clear, communicated policy.
The practical elements to cover in any phishing training are: how to inspect a sender's actual email address rather than just the display name, the signs of urgency and pressure tactics, how to verify a request through a different channel rather than replying to the email, and what to do after clicking something suspicious. That last point is critical. Staff who are afraid to report a mistake will stay silent, and an unreported incident usually does more damage.
How to recognise it: staff are unsure how to report a suspicious message, or reports are rare even though phishing clearly reaches inboxes.
7. Assuming your size makes you an unlikely target
Company size alone does not determine risk. Small businesses face both opportunistic and targeted attacks. Much of what reaches them is automated: tools scan the internet for exposed services and known vulnerabilities, test leaked credentials against login pages, and send phishing campaigns at scale, without checking company size first. Targeted attacks happen too, such as invoice fraud aimed at a finance team.
Risk depends more on exposure (what is reachable from the internet), the state of your credentials, unpatched vulnerabilities, and how attractive your data or services are to an attacker, whether for extortion, fraud or onward access. A small business with weak controls in these areas can be an easier target than a larger one with stronger controls.
Supply chains add another dimension. Small businesses often act as suppliers, contractors or technology partners for larger organisations. Attackers who want access to a larger target sometimes compromise a smaller supplier first and use that foothold to move on. The risk for a small business is therefore not only about protecting its own data, but also about being used as a bridge into a wider breach, with the reputational and contractual consequences that follow.
How to recognise it: risk discussions start with "we are too small to be interesting" rather than with what is exposed and what data you hold.
Turning awareness into action
Reading a list of common mistakes is useful. Acting on them before an incident demands an emergency response makes recovery far easier. All seven areas are worth reviewing regardless of where you think you stand. If you want a structured review rather than a self-assessment, Cyvra's cybersecurity practice works with SMEs across the Netherlands and UK to identify gaps and build practical controls.
| Mistake | Associated risk | First action |
|---|---|---|
| Security as a one-time project | Controls drift as systems, staff and threats change | Name an owner and set a review calendar |
| Passwords alone | Account takeover with leaked or phished credentials | Enforce MFA on email and remote access; phishing-resistant methods for admins |
| Missing updates | Exploitation of known vulnerabilities in exposed systems | List internet-facing systems; fix KEV-listed issues first |
| Admin rights for everyone | Malware can change settings, disable protection and spread | Remove admin from daily accounts; use separate admin accounts |
| Untested backups | Recovery fails or takes longer than the business can tolerate | Restore one critical system; compare the time with the RTO |
| No phishing awareness | Phishing and payment fraud succeed and go unreported | Make reporting easy; train staff to verify requests another way |
| The size assumption | Exposure is underestimated and left unmanaged | Review exposed systems, credentials and supplier dependencies |
Self-assessment: what to verify
For each area, ask whether a practice exists and whether it is followed today, not just on paper. Gaps between policy and reality are where incidents often start, and evidence is what lets you answer with confidence.
| Area | What to verify | Evidence |
|---|---|---|
| Continuous security | An owner and a review cycle exist and are followed | Review calendar, notes from the last review, asset inventory |
| MFA | All users and applications covered; methods match the risk | MFA coverage report, list of exceptions and emergency accounts |
| Updates | Exposed systems and known exploited vulnerabilities fixed within agreed times | Patch status report, exceptions with mitigations |
| Least privilege | Daily accounts have no admin rights; admin accounts are separate and few | List of privileged accounts, date of last access review |
| Backups | Critical systems can be restored within the RTO and RPO | Restore test records with measured recovery times |
| Awareness | Staff report suspicious messages and verify unusual requests | Reporting figures, training records |
| Exposure | Internet-facing assets and supplier dependencies are known | External asset list, supplier list with access granted |
If you find several gaps and are unsure where to start, MFA on email and remote access and fixing known exploited vulnerabilities on exposed systems are usually the first priorities. Our cybersecurity baseline for SMEs sets out how to implement the controls.
None of these steps requires a large budget or specialist infrastructure. They require time and consistency. Working through them can close a significant part of the gaps you identify, although no set of controls removes risk entirely.
The NCSC publishes practical cybersecurity guidance for small and medium businesses, covering the same foundational controls described in this guide. ENISA's cybersecurity best practice resources provide additional implementation detail for organisations building out their security programme.
If you are looking for a structured approach to closing the gaps described here, a cybersecurity roadmap gives you a sequenced programme with business priorities attached to each control decision.