CISA's #StopRansomware Guide names three routes a small accounting firm should know: phishing emails that reach staff, macros that can deliver ransomware, and managed service providers whose own systems can carry an attack into a client's office. Each route has a direct response, and the first is training staff to identify and report suspicious activity.
> Key takeaways > - Phishing emails are the first route. Train staff to recognize suspicious activity and tell someone about it. > - Macros in documents can deliver ransomware, so treat macro-enabled files from outside the firm as a risk to manage. > - Filter known malicious indicators at the email gateway and block suspicious IP addresses at the firewall. > - Outside IT support is a route too, so the firm's provider deserves the same scrutiny as its own staff. > - Limit shared local administrator passwords, and use LAPS where your systems allow it.
Route one: phishing emails that reach your staff
Phishing is the first route, and it reaches a firm through the people who read the mail. An accounting office receives client documents, tax forms and payment requests by email all the time, so a fake message that looks like a real client request is a realistic threat. The guide's answer is training: the firm should implement a cybersecurity user awareness and training program that includes guidance on how to identify and report suspicious activity (phishing) or incidents.
The guidance asks for two skills, not one. Staff must recognize suspicious activity, and they must know how to report it. A person who spots a strange message but does not know who to tell, or who fears looking foolish for raising a false alarm, leaves the firm exposed just as much as someone who never noticed. For a small firm, a reporting path can be simple: one named contact, one written instruction, and one clear rule that nobody is blamed for reporting in good faith.
The training also covers incidents, not only phishing messages. A staff member who clicks a link and realizes the mistake a moment later should report it, because early reporting gives the firm a chance to act before the problem spreads. Operations leads should make sure the training reaches everyone who handles client files, including part-time and seasonal workers, since they receive the same messages as full-time staff.
Training works best when it uses the firm's real mail. Build examples from requests the firm actually gets, such as a changed payment detail on an invoice or an urgent request for a client's login. Repeat the reporting instruction at onboarding and at the start of each busy season, when message volume and pressure are highest. Phishing rarely depends on technical skill. It depends on a person acting quickly, and good training gives that person a pause and a path.
Route two: macros hidden in documents
The second route sits inside files that look ordinary. The guide warns that these macros can be used to deliver ransomware. A macro is a set of instructions embedded in a document or spreadsheet to automate work. The same capability that saves time on a payroll sheet can also run code the firm never intended.
For an accounting firm, the exposure is practical. Clients send workbooks, examples and forms, and staff open them to finish their work. A file from a familiar contact, or one that appears to come from a trusted vendor, carries the trust of the sender even when the file itself is the problem. Because the file looks routine, the person opening it has little reason to pause, which is why this route can slip past staff who are otherwise careful with mail.
Treat macros as a risk to be managed, not as an ordinary feature. Before opening a file from outside the firm, ask whether it needs macros at all, and be especially careful with macro-enabled files. If a client's workflow depends on a macro, the firm should understand what it does before running it. That question belongs with whoever supports the firm's computers, not with the person who happened to receive the file.
The macro route connects to the training route. Staff who know what a suspicious document looks like are better placed to stop one. Training should say which kinds of files are higher risk, what to do with an unexpected document, and who decides whether to enable content in a file. Writing those rules down removes the guesswork a busy employee faces at the end of a long day.
The three routes compared side by side
The three routes are stated briefly, which makes them easy to compare. The table sets each route beside a short paraphrase of CISA's statement, so the differences are visible at a glance.
| Route | What CISA states | Quote reference |
|---|---|---|
| Phishing emails | Staff should identify and report suspicious activity such as phishing | CPG 2.I |
| Malicious macros | Macros can be used to deliver ransomware | CPG 2.N |
| Managed service providers | Infection reaches numerous client organizations through MSPs | CPG 1.I |
The first two routes depend on what a person does with one message or one file. The phishing route depends on whether staff report what they see, and the macro route depends on whether a file gets opened and its content enabled. The third route works differently. It reaches the firm through a vendor, and MSPs have been an infection vector for ransomware impacting numerous client organizations. A firm's exposure can run through someone else's systems, so its own training cannot close that gap alone.
The routes also differ in who owns the response. Partners should own the training and the file-handling rules, because those shape daily behavior. Operations leads should own the question of what outside support can reach and how that support is monitored. Each route starts with an ordinary action, reading a message, opening a file, or relying on a support provider, which is why each one deserves a named owner.
Stopping bad mail at the email gateway
Two technical filters complement staff training. The first is to implement filters at the email gateway to filter out emails with known malicious indicators, such as known malicious subject lines, and block suspicious Internet Protocol (IP) addresses at the firewall. That sentence describes two places where filtering happens, and each one looks for a different kind of signal.
The table shows what each filter examines.
| Control | Where it runs | What it targets |
|---|---|---|
| Email gateway filter | Incoming email | Known malicious indicators |
| IP address block | Firewall | Suspicious IP addresses |
The gateway sits between the internet and staff inboxes, so it can stop some messages before anyone sees them. These filters depend on indicators that have already been identified as malicious. A new message that matches no known indicator still reaches a person, which is why the filters and the reporting habit work as two layers rather than substitutes.
The firewall rule is a separate layer. Blocking suspicious IP addresses stops connections to addresses marked as suspicious, so a message or download trying to reach one of those addresses can be stopped at the network edge. For an accounting firm, the practical step is to ask the firm's IT support how the gateway and firewall are configured, and to confirm that the filters are actually running on the mail the firm receives.
Blocking malicious executables with Sysmon
As of Sysmon 14, the FileBlockExecutable option can be used to block the creation of malicious executables, Dynamic Link Library (DLL) files, and system files that match specific hash values. The setting covers three kinds of file creation, and the blocking is tied to hash values.
| Blocked item | Action | Version named |
|---|---|---|
| Malicious executables | Creation blocked | Sysmon 14 |
| DLL files | Creation blocked | Sysmon 14 |
| System files | Creation blocked on matching specific hash values | Sysmon 14 |
A hash value identifies one exact file. The rule therefore works against files already identified as malicious, the same limit that applies to the email filters above. Its value is that it stops known bad files from being created on machines, instead of relying on a person to notice them.
Sysmon configuration belongs to the firm's IT support, not to a partner adjusting settings on a work machine. The useful question for an operations lead is simple: does the firm's support provider run this setting, and can it explain which machines have it? Sysmon 14 is the version where this option appears, so an older monitoring install may not offer it at all.
Removing shared local admin passwords
The recommendation is to implement Local Administrator Password Solution (LAPS) where possible if your OS is older than Windows Server 2019 and Windows 10 as these versions do not have LAPS built in. Two points sit in that sentence. LAPS is a recommended control, and machines on older systems need it added because those versions lack it built in.
The reason it matters is that a local administrator account has elevated rights on one machine. If the same local administrator password is reused across many computers, someone who learns that password can move from one machine to the next.
The firm can check its exposure with a simple inventory. List every computer and server, and note which operating system each runs. Any machine on a version older than Windows Server 2019 or Windows 10 should have LAPS added separately where possible, since those versions do not include it. Where a machine cannot take LAPS, the same inventory shows which machines need a different plan. Operations leads should ask the outside IT provider to confirm the list, since the provider usually manages these machines.
Using lower-permissioned accounts for daily work
Every staff member who signs in to a computer, mail system or client portal has some set of rights. Some accounts can change system settings, install software or manage other users. Others can only read and edit the files their job requires. Microsoft's guidance on multifactor authentication makes a point that applies to any firm: using lower permissioned accounts helps improve security for your organization.
For an accounting firm, the practical habit is to separate two kinds of work. Administrative tasks, such as changing user settings or adding a new mailbox, should happen in an account used only for that purpose. Everyday tasks, such as reading client email, preparing returns and saving spreadsheets, should happen in an account with fewer rights. If the everyday account is compromised, the attacker can only do what that account was allowed to do.
Operations leads can start with an inventory. List each person and each shared account, then note what that account can change. Accounts that hold broad rights but are used for routine work are the first applicants for narrowing. Partners often hold the broadest rights because they set up the firm's systems years ago, and those rights tend to remain long after they are needed.
The change also takes some adjustment. Staff may resist when a task that used to be quick now requires a separate sign-in. Explain the reason in plain terms: the separate account exists so that one mistake or one stolen password does less harm. A narrower account will not stop every incident, but it is a right the firm can reduce on purpose, and that is worth doing.
Limiting what a stolen credential can do
Accounts are not the only credentials in use. Many tools issue tokens, which are strings that let a program or service act on someone's behalf. GitHub's documentation on personal access tokens sets out the core rule: a personal access credential has the same capabilities to access resources and perform actions on those resources that the owner of it has, and is further limited by any scopes or permissions granted to it.
That sentence contains two limits. First, a personal access credential starts with the owner's reach. If the owner can reach a set of files or actions, a personal access credential issued under that owner can reach the same set. Second, the credential is cut down by its scopes or permissions. The owner's reach is the ceiling, and the grant given to the credential is what actually applies.
For a small firm, the lesson is to ask what each personal access credential can reach before it is issued. GitHub's rule states that a personal access credential carries the same reach as the owner who created it, so a broad account means a broad credential in any script, integration or support tool that holds it. If the credential is lost, copied or exposed, the damage follows the permissions it was given. A credential granted only the access one task needs gives an attacker much less to exploit.
The steps are straightforward. Keep a list of the tokens the firm uses, who created each one, and what it can do. Remove tokens that no longer serve a purpose. When outside support asks for access, ask whether it needs a personal access credential at all and, if so, what scope it requires. The documentation does not describe how any particular tool stores tokens, so confirm those details with the tool's own documentation and with the firm's IT provider.
Which one should you pick?
The firm does not need to choose only one, but it does need a starting point. The two controls address different parts of the same problem. Lower-permissioned accounts limit what a person's sign-in can do on the firm's systems. Access scope limits control what a credential can do when a program or service uses it. Both rest on one principle: rights should be narrow enough for the job.
If the firm has to pick one first, start with accounts. Almost every staff member signs in every day, and nearly every machine depends on accounts. The change affects the whole office and removes unnecessary rights where most work happens. Microsoft's guidance supports this basic step, as noted above.
Tokens deserve attention once the account work is underway, and they may matter more for firms that connect many tools. A firm that links accounting software, document tools or a support provider through tokens should list those connections and their rights. Where a connection is no longer used, remove it. Where it has more access than its task needs, narrow it.
A simple rule ties the two together: if a credential exists for one job, give it only the rights that job requires, and review it when the job changes. That rule costs little to apply, which makes it a sensible first step for a small firm with limited time.
Account and credential limits at a glance
The table below sets the two controls side by side, using only what the sources state. It helps an operations lead decide where to start and what to ask the IT provider.
| Control | Applies to | What the source says | Source |
|---|---|---|---|
| Lower-permissioned accounts | People signing in to systems | Narrower rights for everyday sign-ins | Microsoft Learn |
| Credential capabilities | Programs and services acting for an owner | Personal access credentials start from the owner's reach | GitHub Docs |
| Scopes and permissions | Each individual credential | Further limit what the credential can do | GitHub Docs |
Read the table as a sequence. The owner's rights set the outer edge, and any personal access credential issued under that owner can only operate inside it. The scopes then narrow the credential further. The sources state no numbers or deadlines for these controls, so the table contains none.
Why outside IT accounts need limited rights
Outside IT support often holds the most powerful credentials a firm has. A managed service provider may sign in to servers, change settings and reach client data from its own systems. That reach is the reason to limit outside accounts rather than trust them with full access, as the MSP route above makes clear.
Apply the same principle as before. An outside technician's account should have the rights the support agreement requires, and no more. If a technician needs temporary access to repair a server, the firm should grant it for that job and remove it afterward. Broad rights left in place after a project ends are the standing access an attacker would look for.
Tokens and integrations linked to a provider need the same review. A credential issued to a support tool should be limited by its owner's reach and by whatever scopes are granted, and the firm should confirm those limits with the provider. The firm should know who issued each credential and what it can reach, because a provider's tool may hold more reach than any single technician.
Operations leads should ask concrete questions. Which accounts does the provider use? Which of them are shared? Can the firm see when they sign in? The answers belong in writing. The guidance does not set a standard for them, so the firm should judge the answers against its own needs and the provider's written commitments.
Related guides: Use 7 CP49 notice checks to brief small firm clients · Account deletion for firm records: 3 checks before removal · Audit log retention: 4 settings to check for your firm.
Reader questions
Does a lower-permissioned account stop ransomware? No source says that. Narrower rights are one layer, alongside staff training, email filtering and careful file handling.
Should partners stop using administrator rights altogether? The sources do not say to eliminate them. The Microsoft guidance supports using lower-permissioned accounts for everyday work.
If a personal access credential is stolen, what can it do? A credential's reach is bounded by its owner's rights and by the scopes or permissions granted to it. A stolen credential can therefore do no more than that grant allows.
Can a personal access credential ever have more access than the person who created it? No. For GitHub personal access credentials, one starts from its owner's reach and can only be narrowed from there.
Should we treat our IT provider as a risk? Yes, as something to review. The MSP route above describes how a provider can carry an infection into client offices. Give the provider limited rights and insist on written answers about its access, rather than assuming it is safe.
Do older machines need LAPS? Where possible, yes. Machines on a version older than Windows Server 2019 and Windows 10 do not have LAPS built in, so check which machines fall into that group.
Will email filters catch every phishing message? No. The filters work against known malicious indicators and suspicious IP addresses. Messages that match no known indicator still reach staff, which is why reporting suspicious activity remains necessary.