In this lesson
After this lesson you can:
- Decide whether to self-implement on the real constraint, which is owned time rather than expertise.
- Follow the build order that produces a defensible system instead of a document set.
- Recognise the four failure points early enough that they cost an afternoon rather than a Stage 2.
Read first: The implementation roadmap and the mandatory documents
Nobody requires one.
Worth stating plainly, because the market does not. ISO 27001 does not mention consultants. Accreditation rules do not require you to have used one. A certification body does not ask, and could not treat you differently if it knew, because what it examines is the management system and the system does not record who wrote it.
So the question is purely practical: can your organisation produce a conformant management system with the people and time it has? For a lot of organisations the answer is yes, particularly technical ones where the underlying controls largely exist and what is missing is the management scaffolding around them.
Where the answer tends to be no, it is almost never about intelligence or capability. It is about whether one identified person has this as a real part of their job, with hours in the calendar, for several consecutive months. Programmes that stall stall for that reason far more than any other.
What you could never outsource anyway.
This reframes the decision usefully, because a surprising amount of the work is yours whatever you spend.
Nobody external can decide your risk appetite. Nobody external can be your top management for the purposes of leadership and commitment. Nobody external can own a risk on your behalf, attend your management review as your decision maker, operate your day-to-day controls, or be the person an auditor interviews about how your access requests work. A consultant can design, draft, challenge, rehearse and coordinate. They cannot be the organisation.
The practical implication is that the gap between a well-run self-implementation and a well-run consulting engagement is narrower than the price difference suggests, and the gap between either of those and an unowned programme is enormous.
What the certification body will not do.
Plan for this before it bites, because the instinct to ask your auditor is strong and the answer will be no.
ISO/IEC 17021-1 requires the certification body to be impartial. It may not consult on the management system it certifies. In practice that means it will tell you that something is a nonconformity, cite the requirement, and stop. It will not tell you what good looks like, review your draft documents in advance, or suggest a remedy.
Many bodies will sell you a pre-assessment or gap analysis, which is a defined and legitimate service. Read what it includes, because it is usually a report of findings rather than advice on fixing them, for the same impartiality reason.
If you self-implement, you are therefore your own adviser. That is manageable as long as you have planned for it, and genuinely difficult if you were assuming the audit would come with guidance.
The do-it-yourself path.
The order matters more than anything else on this page. Most self-implementation damage comes from starting in the wrong place, which is almost always Annex A.
Phase 1
Learn the standard properly, once
Buy the standard from ISO and read clauses 4 to 10 end to end before writing anything. Most self-implementation failures trace back to starting with Annex A controls, which are the shopping list, rather than with the clauses, which are the system. Consider sending the ISMS owner on a Lead Implementer course at this point rather than later.
ISO/IEC 27001:2022, clauses 4 to 10
Phase 2
Decide scope, and write it as it will appear on the certificate
Scope is the decision that governs cost, duration and maintenance for the next three years. Decide it from what your buyers are actually asking about, name the entity, services, sites and systems, and resist the instinct to include everything to look impressive.
Clause 4.3, determining the scope of the ISMS
Phase 3
Write the risk method before you assess any risk
Define how you identify risk, how you evaluate it, what your acceptance criteria are and who owns the decisions. Then apply it. Doing this in the other order produces a register you cannot defend, which is the single most common serious finding in self-implemented systems.
Clause 6.1, actions to address risks and opportunities
Phase 4
Treat the risks, and let the SoA fall out of that
For each significant risk, decide the treatment and which Annex A controls it needs. Your Statement of Applicability is the output of that reasoning, including the controls you exclude and why. Built this way round, every justification already has an answer behind it.
ISO/IEC 27001:2022, Annex A and the SoA requirement
Phase 5
Operate it long enough to produce records
This is the phase people try to skip and cannot. The audit samples evidence, so the system needs to have been running long enough to have generated some: access reviews performed, training completed, changes approved, incidents handled. A system switched on last week has nothing to sample.
Clauses 7 and 8, support and operation
Phase 6
Audit yourself for real, then review it at the top
Run an internal audit that finds things, by someone independent of the work being audited. Then hold a management review covering every input Clause 9.3 lists, and minute it. Certification bodies expect both completed with corrective actions in motion before Stage 2, and their absence is close to an automatic finding.
Clauses 9.2 and 9.3, internal audit and management review
Phase 5 is the one that cannot be compressed, and it is where optimistic timelines die. The audit samples evidence from a period of operation. A system that has not operated has no evidence, and no quantity of documentation substitutes for it. Build the timeline backwards from the amount of operating history you want to be able to show.
The four places it reliably breaks.
These are not hypothetical difficulties. They are the specific findings self-implemented systems collect, and all four are cheap to avoid and expensive to discover.
| Failure | What the auditor sees | The fix |
|---|---|---|
| Risk assessment with no real method | A spreadsheet of worries, scored by vibe. | Write the method down first, then apply it. Criteria before scores. |
| Statement of Applicability copied from a template | Justifications that fail one follow-up question. | Derive each line from your own risk treatment, in your own words. |
| Internal audit done by the person who built it | Impartiality failure, and a near-certain finding. | Cross-function auditing, or buy internal audit as a service. |
| Management review as an undocumented chat | Mandated inputs never discussed or recorded. | Run the agenda from Clause 9.3 and keep real minutes. |
| Scope decided by what was easy | Either unauditable or needlessly expensive. | Decide scope from what buyers ask about, then price it. |
| Documents written and never operated | Policy says quarterly, practice is annual. | Write only what you will do, then do it and keep the record. |
One, the risk method. People open a spreadsheet and start listing threats, then score them. At Stage 1 the auditor asks why this risk is a four and that one a two, and there is no answer because no criteria were ever written. Clause 6.1 lets you choose your own method and requires you to have one. Write it down first. It takes an afternoon.
Two, the Statement of Applicability. Built from a template, it is a list of 93 controls with plausible-sounding justifications attached. Built correctly, it is the output of your risk treatment decisions and every line has reasoning behind it. An auditor finds the difference with one question, and usually starts with your exclusions.
Three, internal audit independence. Clause 9.2 requires objectivity, and nobody audits their own work. In a self-implementing team the person who built everything is often the only person who understands it, which makes them the obvious and impermissible auditor. Solve it with cross-functional auditing or by buying internal audit as a service, and solve it early, because it is the one failure that cannot be fixed in the week before the audit.
Four, the management review. Clause 9.3 lists what has to be considered. A conversation happened, no agenda followed the list, and the minutes are three lines. This is a documentary finding that good practice elsewhere does not repair. Run the clause as your agenda and keep real minutes.
The honest cost, which is time.
We are not going to give you an hour count, because we have seen the range vary by more than it would be useful to average. What we can tell you is where the time goes, which is more actionable.
The largest single block is usually learning, and it is front-loaded. Reading the standard properly, understanding what a management system is as distinct from a control set, and establishing the risk method consumes a disproportionate share before any visible output appears. This is the phase where unowned programmes die, because it produces nothing to show at a weekly update.
The second block is drafting, which is faster than people fear once the thinking is done. The third is operating the system for long enough to generate records, which is calendar time rather than effort. The fourth is internal audit and management review, which is modest effort and high consequence.
What a consultancy actually compresses is the first block, by removing the learning from the critical path. It cannot compress the third at all. That is worth knowing before you compare a fee to a timeline.
When paid help genuinely earns its fee.
Narrowly, and these are the cases where we would say so to someone who had already decided to do it themselves.
- A deal has a date on it. If certification is gating revenue in a specific quarter, buying the learning curve off the critical path is a straightforward commercial trade.
- Nobody can own it. If the honest answer is that this will be everyone’s third priority, external capacity is the difference between a programme and an intention.
- Your scope is genuinely complicated. Group structures, multiple entities, mixed regulatory obligations and a certificate that has to satisfy several different buyers are where scope mistakes get expensive for three years.
- You need internal audit independence you do not have. This one is structural rather than a matter of skill, and buying the audit function is often cheaper than reorganising to create independence.
- You are already certified and something went wrong. A major nonconformity with a closure deadline is a poor time to be learning the standard.
Notice what is not on that list: company size, technical sophistication, or whether your buyers are enterprises. Small technical teams self-implement successfully all the time.
The hybrid most people end up at.
In practice the choice is rarely all or nothing, and the shape self-implementing organisations converge on looks like this.
- Send the ISMS owner on a Lead Implementer course. It removes most of the learning block and is cheap relative to a full engagement.
- Build the system yourselves, in the order above.
- Buy internal audit from outside, which solves the independence problem cleanly and gives you a competent first pass.
- Buy a readiness assessment before Stage 2, which is a few days of someone telling you what an auditor will raise.
- Appoint the certification body yourselves and run the audit yourselves.
That gets you the two things external help is genuinely good at, which are independence and a trained read on whether you are ready, without paying for a programme you were capable of running. Which course to take covers step one, and what an auditor actually asks is a reasonable self-service substitute for part of step four.
Doing it yourself and want a second read?
A readiness pass is a few days of us running your own records against what a certification body will ask, and telling you what they would raise. It is the smallest useful thing to buy if you have built the system yourselves, and it does not commit you to anything further.
Key takeaways.
- Nothing in the standard requires a consultant and no certification body asks whether you used one. Self-implementation is normal.
- The real constraint is owned time, not expertise. One person with protected hours for several months is what separates a programme from an intention.
- Your certification body cannot advise you, because impartiality prevents it consulting on the system it certifies. Plan to be your own adviser.
- Build in order: scope, then risk method, then risk assessment, then risk treatment, and let the Statement of Applicability fall out of that. Starting at Annex A is the classic error.
- Operating history cannot be compressed. The audit samples evidence from a period, and documents without records give an auditor nothing.
- Write the risk method before scoring any risk. It costs an afternoon and prevents the most common serious finding.
- Internal audit independence is structural rather than a skill problem. Solve it early with cross-function auditing or an external internal audit.
- Run Clause 9.3 as your management review agenda and keep real minutes. It is a documentary finding that nothing else repairs.
Quick check.
Your certification body raises a major nonconformity at Stage 2. Can you ask them how to fix it?
- Yes, that is part of the service
- No, impartiality prevents them consulting on the system they certify
- Yes, for an additional fee
- Only in writing
Check the answer
B. ISO/IEC 17021-1 requires the certification body to be impartial, so it identifies nonconformities and does not design the remedy. Expecting advice from your auditor is the most common planning error in self-implementation. ISO/IEC 17021-1
In what order should a self-implementing team build these?
- Annex A controls, then risk assessment, then scope
- Scope, then risk method, then risk assessment, then Statement of Applicability
- Statement of Applicability, then policies, then scope
- Policies first, since the auditor reads them first
Check the answer
B. The Statement of Applicability is an output of risk treatment, and risk treatment is an output of a stated method applied inside a decided scope. Starting with controls or documents produces justifications with no reasoning behind them. Clause 6.1, actions to address risks and opportunities
Spot the mistake. A team buys a toolkit, completes every template in a fortnight, and books Stage 2 for the following month.
Check the answer
The system has not operated, so there are no records to sample.. Stage 2 tests whether controls operate, by sampling evidence from a period. A complete document set with no operating history gives the auditor nothing to examine, and the mismatch between polished documents and absent records is immediately visible. Clause 9.2, internal audit
Sources.
- ISO/IEC 17021-1, requirements for bodies providing audit and certification of management systemsISO catalogue entry 61651Used here: The impartiality rule, which is why a certification body cannot advise you on fixing what it finds
- ISO/IEC 27006-1:2024, requirements for bodies auditing and certifying an ISMSISO catalogue entry 82908, published March 2024Used here: That audit duration follows a published method and is unaffected by who prepared the organisation
- ISO/IEC 27001:2022, Clause 6.1, actions to address risks and opportunitiesOur clause reference page, which carries the requirementUsed here: That you set your own method and acceptance criteria, which is why the method has to be written first
- ISO/IEC 27001:2022, Clauses 9.2 and 9.3, internal audit and management reviewOur clause reference pages, which carry the requirements and their documentsUsed here: The impartiality constraint on internal audit, and the mandated management review inputs
Changes since publication
This page states no hour counts, month counts or prices for self-implementation. The ranges we have observed vary too widely to average usefully, so the page describes where the time goes instead, which is the part you can plan around.
Last checked against ISO’s catalogue entries for the certification-body standards, and our own clause reference pages on 9 November 2026.