Ask an IT department for its asset inventory, and you’ll typically get a list of laptops, servers and software licenses. That’s useful. But it’s missing the most important assets of all: the information and the processes the organization actually lives from.
Step 5 is about knowing what you’ve got. In Step 7, you’ll assess your information security risks. But risks to what? The inventory answers that, and the classification you build here gives you the impact levels for the risk methodology in Step 6.
Annex A before the risk assessment
None of the tasks in this step comes from the clauses of ISO 27001. The three controls involved, A.5.9 on the inventory, A.5.12 on classification and A.5.13 on labelling, are Annex A controls, and like every Annex A control, you need them if your risk treatment says so. So strictly speaking, ISO 27001 doesn’t require an asset inventory.

Why do it now, then? Because it makes everything that follows easier. The inventory tells you what you’re protecting, and that’s what your risk assessment looks at. The classification gives you consistent levels for confidentiality, integrity and availability, which you’ll use to rate the impact of a risk. And the labels tell everyone how to handle information, without a separate risk assessment for every document. In Step 8, your Statement of Applicability confirms these controls like any other.
1. Build the inventory (A.5.9)
Control A.5.9 asks for three things in one sentence: the information and the assets around it, an owner for each of them, and an inventory that’s kept up to date, not one created once for the audit.
An inventory pays off well beyond the audit. When a new vulnerability is published, which of your systems are affected? When there’s an incident, which information is at stake, and who decides? When a site goes down, what do you need back first? Without an inventory, each of these questions starts with a search.
Primary and supporting assets
An asset is anything that has value to the organization. In information security, there are two kinds.

Primary assets are the information itself, and the business processes and activities. That’s what really matters to the organization. Supporting assets are everything the primary assets rely on: hardware, software, networks, personnel, sites and the organization’s structure. And that’s exactly what the IT list was missing: the primary assets.
The two are connected. Take order fulfillment. It works with customer order data, runs on an order management system, which runs on an application server in a data center. If the data center fails, the process stops. So you map the supporting assets to the primary assets they support, and a supporting asset needs the protection of the most critical information it holds.
What goes into the inventory
For every asset: a name and a type, an owner, the classification, where it is as far as that’s useful, and for supporting assets, which primary assets they support. Two things save a lot of work. The level of detail should fit your needs: you don’t list every mouse or every short-lived virtual machine. And the inventory doesn’t have to be a single list. If IT already keeps a configuration management database, or HR a list of personnel, you refer to it instead of copying it.
The owner
An owner isn’t the person who legally owns the asset. It’s the person or group responsible for it over its whole life cycle.

That means the asset is inventoried, classified and protected, the classification is reviewed periodically, there are rules for acceptable use, access matches the classification, the asset is disposed of securely at the end, and the owner is involved in managing its risks. An owner can delegate the daily work, for example to IT as a custodian, but stays accountable.
2. Classify your information (A.5.12)
If everything is confidential, nothing is. When every document carries the same label, people stop paying attention to it. A classification scheme avoids that.
Control A.5.12 asks you to classify information according to your security needs, based on confidentiality, integrity, availability and the requirements of relevant interested parties. So it’s not just about confidentiality. And if a customer hands you its data together with its own requirements, those count, too.

A typical scheme has four levels for confidentiality. Public: disclosure causes no harm, like your website. Internal: minor damage, like an internal phone list. Confidential: a significant short-term impact, like customer contracts or personal data. Secret: a serious long-term impact, like the source code of your products or your strategy. The names are up to you. Just keep in mind that another organization’s “confidential” can mean something different, so when you share information, agree on how your levels translate.

For integrity and availability, three levels are typically enough: low, medium and high. The question is always the same: what happens if this information is wrong, or not there when you need it? You rate that across a few categories, like financial loss, disruption of operations, reputation, compliance and, for some organizations, safety. And that’s exactly the kind of impact scale you’ll need in Step 6. The work you do here pays off twice.
The owner classifies, and the classification isn’t fixed forever: once a product is launched, its brochure is no longer confidential. Watch out in both directions. Over-classification leads to controls you don’t need, under-classification leaves information unprotected. And a laptop or a server takes the classification of the information it holds.
3. Make it visible with labels (A.5.13)
A classification nobody can see is just an entry in a spreadsheet. The person holding a printout, or opening an attachment, needs to know right there how to handle it.
Control A.5.13 doesn’t say “put a sticker on everything”. It asks for procedures: how, where and when you label. Typical techniques are headers and footers, metadata, watermarks and physical labels. Metadata deserves a second look: if the label is in the metadata, your systems can act on it, for example by encrypting a confidential attachment automatically.

Your procedures define it medium by medium. Printed documents carry the level in the header or footer. Digital files carry it in their metadata. Emails show it in the subject line or through a label from your email system. Devices and storage media typically aren’t labelled at all, they’re classified in the inventory. And anything unlabelled is treated as a default level, for example internal.
You don’t have to label everything. Labels can be omitted where they add nothing, typically for public information. Where labelling isn’t technically possible, you protect the whole system at the highest level of the information it holds. And keep in mind that a big “secret” on a file tells an attacker exactly where to look.
At Barvello: starting with what Barvello lives from
Barvello Coffee Systems is the fictional company of our ISO 27001 Lead Implementer course: a family-owned maker of professional coffee machines in Munich, with about 500 employees and its own cloud platform, Barvello Connect. When project lead Nadia Berger asked the head of IT for Barvello’s asset inventory, she got an export from the IT service desk tool: more than a thousand entries, from laptops to printers. Everything was there, except what Barvello actually lives from: the firmware, the machine data in Connect, the customer contracts.
So Nadia didn’t ask “what systems do you have?”. She asked each department head: what does your department do, and what information can’t it do without? The processes came out of that: product development, production, Connect operations with pay-per-cup billing, field service and sales. And the information: the firmware source code and its signing key, the product designs, the machine data in Connect, the customer contracts and the personnel files. Each got an owner and a classification. The firmware is secret, because whoever has the signing key can push firmware to thousands of machines.
Then the supporting assets, together with IT: SAP, the production systems, Connect on Azure, Microsoft 365, the code repository, the signing server, the server room, the network and the suppliers. The laptops, tablets and phones are one line, with a reference to the service desk tool. And Nadia carried over what she learned in the gap analysis: the SAP restore that was never tested, and the firmware team as a supporting asset of its own, because only four people know the signing key.

The mapping made one connection visible that matters a lot. Pay-per-cup billing works with the cup counters, which come from the controller boards in the machines, flow into Barvello Connect, and Connect runs on Microsoft Azure. If Azure has a problem in the EU, Barvello can’t invoice some of its biggest customers. That’s exactly what the risk assessment in Step 7 will need.
For the classification, Nadia kept the four levels and filled in Barvello’s own examples, so nobody has to guess. Secret: the firmware and its signing key, and the designs of the next product line. For availability, “medium” means downtime must not exceed 48 hours, the same two days the COO named in the business case. And safety stayed in as an impact category, because Barvello’s machines work with steam and hot water, and manipulated firmware could make that dangerous.
The labels come from Microsoft 365: one click, and the label sets the metadata and adds the level to the footer. Printed work instructions in production carry “Internal”. Anything without a label counts as internal until its owner classifies it. Two weeks after the start, the inventory was complete: twelve primary and fifteen supporting assets, each with an owner and a classification.
Your turn: Step 5 in the project plan
In the project plan of our ISO 27001 Toolkit, Step 5 is ten tasks:
ID | Task | Control |
|---|---|---|
42 | Establish topic-specific policy on asset and configuration management | A.5.9 |
43 | Publish topic-specific policy on asset and configuration management | |
44 | Identify primary assets | A.5.9 |
45 | Identify supporting assets | A.5.9 |
46 | Document and map primary and supporting assets | A.5.9 |
47 | Identify and assign asset owners | A.5.9 |
48 | Establish topic-specific policy on information classification | A.5.12 |
49 | Classify information | A.5.12 |
50 | Establish procedures for information labelling | A.5.13 |
51 | Label information | A.5.13 |
At the end of Step 5, you have four deliverables, and the toolkit has a template for each: the asset inventory (REC-022), the policy on asset and configuration management (POL-016), the information classification policy with the labelling rules (POL-007) and the asset management process (PRO-018).

All four are determined. ISO 27001 doesn’t explicitly require them, but once you’ve decided on these controls, they’re how you show that you’ve implemented them. Start with the primary assets: ask what your organization lives from, not what IT has. You’re done with Step 5 when every primary asset has an owner and a classification.
From the auditor’s chair
An inventory that only lists laptops and servers tells you something: the organization hasn’t asked yet what it actually needs to protect. A useful one starts with information and processes.
And a useful inventory is alive: owners who know they’re owners, classifications that were reviewed, and changes that come through a process, not through a clean-up the week before the audit.
Frequently asked questions
Is an asset inventory mandatory for ISO 27001?
Not by the clauses. A.5.9 is an Annex A control, so you need it if your risk treatment says so. In practice, almost every organization does, because the risk assessment and incident response depend on it.
What’s the difference between primary and supporting assets?
Primary assets are the information and the business processes, what the organization lives from. Supporting assets are what they rely on: hardware, software, networks, people, sites and suppliers.
Do we have to list every laptop?
No. The level of detail should fit your needs. You can list end-user devices as one line and refer to the inventory in your IT service desk tool or configuration management database.
How many classification levels should we use?
Typically four for confidentiality (public, internal, confidential, secret) and three for integrity and availability (low, medium, high). Fewer levels are easier to apply; more levels are rarely used correctly.
Does every document need a label?
No. You can omit labels where they add nothing, typically for public information, and define a default level for everything unlabelled.
Step 5 in the course.
With the templates to build it.
One lecture for each of the three controls, the inventory, classification and labelling, and a case study. Your turn: 10 tasks in the project plan and the templates for the inventory, both policies and the asset management process.
Inventory of Information and Other Associated Assets
A.5.9
Classification of Information
A.5.12
Template: Asset Inventory
REC-022
Template: Information Classification Policy
POL-007
With your assets known and classified, you’re ready for Step 6: the risk management methodology, where you decide how you’ll assess your risks, including the impact levels you’ve just prepared.


