Where, inside a company's quality, risk and safety systems, does the management of its information belong? I am Dr Sepehr Abrishami, and on Reimagining AEC, in a season on the ISO standards, I come to that question from my research on generative artificial intelligence, because every new way of producing information eventually needs a home in those same systems. This episode is about information management as three standards around the ISO 19650 series give it that home, ISO 9001 for quality, ISO 31000 for risk and ISO 45001 for occupational health and safety.
Arden Construction Ltd is hypothetical, and nothing said about it describes a real company or a real event. Arden is a lead appointed party on Northgate Quarter, a hypothetical regeneration area, for Block A, its eleven storey residential building. A few months into the job, an information exchange comes back from review. The health and safety information register for Block A holds fourteen risk records with their status still unknown and no linked treatment, six weeks after a design risk workshop agreed treatments for all fourteen. In the same fortnight there was a near miss at a riser in Block A, a services clash where the coordinated model showed a clear route. The mechanical and electrical subcontractor wrote it on a paper near miss card. The card reached the health and safety adviser. It never reached the register. The reviewer marks the register as not complete, and the exchange misses its milestone.
On paper, Arden is well run. It is a design and build contractor in England, with around two hundred and forty staff and around ninety regular subcontractors. It holds certificates to ISO 9001 and ISO 45001 through one integrated management system, with one manual, one audit programme and one management review, and its board keeps a corporate risk register that it describes as aligned to ISO 31000. Its appointment carries exchange information requirements written to the ISO 19650 series, and its management system has never mentioned information management. Now Arden has to decide whose failure this is. The design manager running the information function says the records were the adviser's. The adviser says the register belongs to the delivery team. The quality manager says neither appears anywhere in the manual. All three are telling the truth.
So the question for this episode is where information management should sit inside an organisation's quality, risk and safety systems, and what changes when it sits there. By the end you will be able to read any management system standard for the information question in each of its clauses. You will be able to claim the common data environment as evidence against a clause the organisation is already certified to, keep an information risk register that rests on a framework, trace an information failure to its management system cause, and draft clauses an auditor can test.
We will take six steps. First the shared skeleton of the three standards and the editions in force. Then documented information and the common data environment. Then leadership, accountability and competence. Then risk applied to information. Then incidents, corrective action and the bridge to Part 6 of the series. And last, one system or many, the drafting, the audit and how to stay current. Arden stays with us throughout.
The first step is the skeleton, because the three standards look alike and behave very differently. Strip away the subject, and a management system is a set of commitments about how an organisation will run. It says what the system is for, meaning its intended outcomes, its scope, and the context and interested parties that shape them. Top management leads it, and the organisation plans it by addressing risks and opportunities and setting objectives. It is resourced and run, with competent people, controlled documented information and processes working to stated criteria. It is checked, through monitoring, internal audit and management review. And it is improved, by reacting to nonconformity and correcting causes. Nothing in those commitments is about quality in particular. They are a way of running an organisation, and information management is one of the things an organisation runs.
The commitments are written into a shared clause sequence called the harmonised structure. It sits in Annex SL of the directives ISO uses for drafting its standards, and the 2021 edition of those directives renamed it from the high level structure, so you will hear both names. It gives every management system standard the same clauses from four to ten, the same core text and the same common terms. Clause four is context, five is leadership, six is planning, seven is support, eight is operation, nine is performance evaluation and ten is improvement. Planning is the plan step of plan, do, check, act. Support and operation are the do step, performance evaluation is check, and improvement is act. Information work on projects tends to be all do. Teams produce, share and publish, and the plan, check and act steps are left to appointment documents and exchange reviews. Block A is that pattern exactly, a register produced and shared, and nothing in the organisation checking it until a reviewer did. A management system insists on all four at the level of the organisation, not only the project. The working habit to take from this is to ask, of every clause, what its information question is. Clause seven, for example, is where documented information lives, clause nine is where audit lives, and clause ten is where corrective action lives.
Then the distinction people get wrong most often. ISO 9001 and ISO 45001 are requirements standards. They are written with shall, they follow the harmonised structure, and an organisation can be certified against them. ISO 31000 is a guidelines standard, written with should. It is not a management system standard, it does not follow the harmonised structure, and it is not intended for certification, because there is nothing to certify against. Two standards you conform to, one you draw on. Certification itself is a formal audit by a certification body, and in the United Kingdom those bodies are accredited by the United Kingdom Accreditation Service. Nothing in this episode substitutes for that process. What it does is show where information management belongs inside a system that may already be certified, so that the next surveillance audit finds it there.
Five shared terms carry information consequences. Documented information replaced documents, records, procedures and manuals in the 2015 edition of ISO 9001, and every information container in a common data environment is documented information in that sense. An interested party is anyone who can affect or be affected by the organisation, so an appointing party is one, and so is every task team. A nonconformity is the non fulfilment of a requirement, and a superseded drawing revision in use on site is one. Corrective action removes the cause of a nonconformity so that it does not recur, which is different from correcting the instance. And top management, as ISO 45001 notes, remains accountable for the system whatever it delegates.
Editions matter here more than usual, because the cover of one of these standards carries several dates. Take the British edition of the safety standard, BS EN ISO 45001 of 2023 with Amendment 1 of 2024. Its content is the 2018 international standard. The 2023 date is the year the European Committee for Standardization adopted that text unchanged, and the British Standards Institution, or BSI, reissued its edition to match. The 2024 date is the amendment. Read 2023 as a new edition and you have misread it, and so has any manual that cites it that way. The risk standard is BS ISO 31000 of 2018, with no EN, because it was never adopted as a European standard. And the quality standard has just moved. The clause numbers in this episode are those of ISO 9001 of 2015 with its 2024 amendment, the edition Arden is certified to. In September 2026 ISO published a sixth edition, and BSI published it in the UK as BS EN ISO 9001 of 2026. It adopts the latest harmonised structure, and the 2015 edition is to be withdrawn, with certified organisations moving across in a transition period. A revision of ISO 45001 has reached the draft international standard stage and is targeted for publication in 2027. A third edition of ISO 31000 was registered as a draft international standard in November 2026. Neither draft is in force.
The 2024 amendment is small and worth knowing exactly. It was applied to both requirements standards under the title climate action changes. At clause four point one, on the organisation and its context, a sentence now requires the organisation to determine whether climate change is a relevant issue. At clause four point two, on interested parties, a note says those parties can have requirements related to climate change. That is all of it. No new clause and no new named record. But two records now have to show that climate change was considered, the context analysis and the interested party register, and both feed information requirements, because an appointing party with a net zero commitment will write it into its organisational information requirements.
Edition literacy is itself a documented information control. Clause seven point five point three of both requirements standards asks the organisation to identify and control the documented information of external origin it needs. A standard the system claims conformity to is exactly that. So is the appointing party's information standard, the UK National Annex to ISO 19650-2, and the classification system a project cites. Each has an edition, and each edition can change underneath a team. The control that follows is an edition register, with an owner and a review cycle. At Arden, nobody holds one.
So the first step resolves. Three standards share one skeleton, two of them carry requirements and one carries guidance, and every clause of the skeleton has an information question you can ask without reading the standard cover to cover.
The second step picks up that question at clause seven point five, documented information, and puts it to the common data environment. The clause reads the same in both requirements standards, because it is harmonised core text, and it has three parts. The first says the system includes the documented information the standard requires and whatever else the organisation needs, in proportion to its size, its processes and the competence of its people. The second, on creating and updating, asks for identification and description, suitable format and media, and review and approval for suitability and adequacy. The third, on control, asks that documented information is available where and when it is needed and adequately protected, and it covers the handling of documented information from distribution through to disposal, along with documents of external origin and evidence protected from unintended alteration.
Listen to that third part with a common data environment in mind, and you hear the workflow of clause twelve of ISO 19650-1 described by someone who never used the word container. Identification is the unique container identifier built from the fields of the project's information standard, with status, revision and classification as metadata from the UK National Annex to Part 2. Format is fixed by the information standard and the delivery team's schedule of software and hardware. Review and approval is the quality assurance check and the review before a container moves from work in progress to shared. Under control, the shared and published states make current containers available, and the status code says what each one is suitable for. Permissions by party and by state separate who may view from who may change, which is the distinction the note to clause seven point five point three draws. Revision codes control change. The archive carries retention and protects evidence, because it is never edited. And the reference information and shared resources the appointing party supplies are documents of external origin. A delivery team running the series properly already meets almost all of clause seven point five. What is usually missing is the sentence in the manual that says so. The states and codes themselves are taken much further in the episode The Common Data Environment: Process Before Platform.
The mapping breaks in three places, and Arden has all three. The first is the organisation's own records. The common data environment holds project containers. The manual, the procedures, the audit records, the competence records and the management review minutes live somewhere else, and the manual has to say where, or each manager assumes the other holds them. The second is retention after the project. The archive at close-out belongs to the appointing party's environment. Arden's own retention of the containers it authored, for its liabilities, its insurers and its lessons learned, is a decision under clause seven point five point three that nobody has made. The third is work in progress. It sits under the task team's own control until the transition to shared, and the manual should state that boundary deliberately, rather than leave an auditor to find uncontrolled files and write a finding.
Arden's manual section on document and record control was written for the 2008 edition and never revised. Read against the mapping, it shows four faults. It still uses the old terms and the old clause numbers, so the manual and the certificate disagree about which standard they follow. It controls change by deleting superseded copies, which destroys the record and has no idea of a revision code. Its list of external documents names standards, regulations and client specifications, and leaves out the reference information and information standard the appointing party supplies, so the most important external documents on Northgate Quarter are uncontrolled by Arden's own rules. And it retains quality records for six years and says nothing about models, drawings or structured data. None of these would fail a surveillance audit alone. Together they mean the manual and the delivery team run two different systems.
Turn the mapping round and the delivery team gains three things. A conformity argument, because a tender answer on information control can cite a clause the organisation is certified to rather than a promise in a BIM execution plan, BIM being building information modelling. An audit route, because the internal auditor who already audits document control can audit the environment against the same criteria. And a home for the gaps, because retention, the organisation's own records and the work in progress boundary all have a clause to be decided under and a management review to be decided at. So the second step resolves. The common data environment is documented information control under another name, and claiming it costs one section of the manual.
The third step asks who answers for it. Start with where ISO 19650-2 puts the function. At clause five point one point one it sits with the appointing party, and at clause five point three point one with each lead appointed party. In both, the party nominates individuals from its own organisation or appoints another party to undertake all or part of the function, and where it delegates, it considers the tasks handed over, the authority delegated and the competence needed. The guidance from the framework formerly called the UK BIM Framework, now the IMI Framework, named after the Information Management Initiative, describes the function as the collective responsibility and authority for the information management process, embedded in existing roles. The rule underneath is the one ISO 45001 states for top management. Responsibility can be delegated. Accountability stays where the clause puts it. Two committees reached the same rule.
There are two ways an organisation loses the function, and Arden risks both. The first is the silo. Information management is set up as a separate system with its own procedures and owner, the audit programme never touches it, management review never sees it, and it leaves with the specialist who built it. The second is the orphan. The function is assigned to people in existing roles, as the series intends, but no management system carries it. That guidance warns that people treat information management as secondary under pressure. In an orphan, nothing manages that risk. The aim is a third state, where named processes own each task, someone is accountable under clause five, the people are competent under clause seven, the processes run under clause eight, audit covers them under clause nine, and failures go through clause ten.
Arden already runs five things that behave like management systems, each with an owner, and the phrase information management appears in none of them. The largest is the integrated quality and safety system under the head of quality, safety and environment. Beside it sit the board risk process under the finance director and three sets of procedures, for information technology, for procurement and for project delivery. Every one of the five will touch the Northgate Quarter information model, so the mapping is a matter of finding the natural owner. Nominating the function belongs with roles in the integrated system. The delivery team's risk register belongs with the corporate risk process, so that it inherits criteria rather than inventing them. Container control belongs with documented information, with the information technology manager owning the platform. Task team capability belongs with the commercial director's procurement procedures. Lessons learned belong with organisational knowledge. The choices can reasonably differ between organisations. The justification, cited to a clause, is what matters.
Three clauses of ISO 9001 carry the weight. Clause four point four, the process approach, asks the organisation to determine for every process what goes in and what comes out, how it runs and is controlled, who is responsible for it, and how it is evaluated and improved. Hold the ISO 19650 process against it. Its inputs are information requirements, its outputs are information models accepted at milestones, its sequence runs from assessment and need to close-out, and its criteria are the acceptance criteria. It is a process in exactly that sense, and describing it that way is what gets it an owner. Then clause five point one point one lists what top management must do, and one commitment forbids the silo outright. Top management ensures that the system's requirements are integrated into the organisation's business processes. On a design and build appointment, information management is a business process. A separate information system beside the management system fails that clause on its face. Clause five point three then requires responsibilities and authorities to be assigned, communicated and understood, and nominating individuals to the function is exactly such an assignment. No standard here requires a single named information manager, any more than ISO 9001 still requires a management representative, which its 2015 edition dropped. The record of who does what is the responsibility matrix, and one point matters here. The assignment matrix in Annex A of ISO 19650-2 has four party columns, with a third party column that shows delegation without inventing a job title. The episode Roles and the Responsibility Matrix takes that further.
The people side follows from clause seven. Clause seven point two on competence asks the organisation to determine the competence needed, ensure people have it, act where they do not, evaluate the action and keep evidence. Applied to the people nominated to the function, the design manager told to run it alongside the day job is either competent, with evidence, or the organisation has an action to take. It stops being a favour. Clause seven point three on awareness requires people to know the implications of not conforming, which is the control that stops a subcontractor building from a model that was only shared for coordination. Clause seven point four on communication asks who communicates what, to whom and when, and Arden's communication procedure lists toolbox talks and the newsletter and never mentions the common data environment, its largest channel by volume. Clause eight point four treats any outside provider of a process as externally provided, and a subcontracted task team fits exactly. Arden must evaluate it, control its output and communicate requirements it has first checked are adequate, so a requirement the task team cannot meet is a nonconformity Arden created. And clause seven point one point six on organisational knowledge is where lessons learned go. A close-out report filed and forgotten meets clause five point eight point two of ISO 19650-2 and fails seven point one point six, because the knowledge is neither maintained nor available.
So the third step resolves. Accountability for information management attaches to a clause five point three assignment with no invented role, and the argument for its time and people is made from clause seven, where nothing is asked that the organisation is not already certified as doing for everything else.
The fourth step turns to risk, and to ISO 31000. It is one short document in three parts. The principles, at clause four, describe effective risk management, with the purpose of creating and protecting value. The framework, at clause five, is how risk management is built into the organisation, with leadership and commitment at the centre and integration, design, implementation, evaluation and improvement around it. The process, at clause six, is what is done to a particular set of risks. The difference between framework and process is the hinge of this step. The framework belongs to the organisation and there is one of it. The process is applied to a project, an appointment or a decision, and it can be applied many times. A register is an output of the process. It is not the framework.
The principles each carry an information consequence, and three of them do most of the work. Integrated means information risk belongs in the project risk register, not in a separate list, which is the silo again. Structured and comprehensive means the same criteria on every appointment, so that high on Northgate Quarter means what high means everywhere. And best available information, with its limitations stated, is precisely what a status code and a revision code exist to communicate. Arden's board register already shows a framework, with a policy, a scale, owners and a quarterly review. What it lacks is any information risk, although clause five point four point one lists data, information systems and information flows among the internal context a framework should understand.
Now walk the process with Arden. Clause six point three sets scope, context and criteria before any risk is assessed, and skipping it is the commonest fault in an information risk register. The scope is information delivery on the Northgate Quarter appointment. The context includes a two block site, a task team that has never used a common data environment, and an appointing party whose information standard changed during the tender. The criteria make the process Arden's own. The project inherits the corporate five level scale and rewrites only the consequence descriptions for information delivery. At the top, very high, an exchange is rejected at a milestone, or construction proceeds from information that was never published for it, which is the point where an information risk becomes a safety one. Next, high, a milestone slips or a container reaches shared without its review. In the middle, moderate, a task team reworks its deliverables inside the milestone. Below that, low and very low, a correction inside the team, or something noted and absorbed. Likelihood runs from almost certain down to rare, and either factor at very high pushes the combined level to high or above. The scale is inherited for one reason. The finance director reading the project register at the quarterly board review sees levels that mean what they mean everywhere else.
Identification at clause six point four point two finds the risks, and the standard's factors produce them. Sources, such as shared templates from the appointing party that contain errors. Vulnerabilities, such as native files on a project drive outside the revision rule. Changes in context, such as a second lead appointed party on the same site whose information Arden depends on. And the reliability of information, such as an existing survey of unknown provenance among the reference information. The standard also says to identify risks whether or not their sources are under the organisation's control, so the second party's information goes in the register. Analysis at six point four point three then weighs likelihood and consequence and, the factor most registers forget, the effectiveness of existing controls. A control that works is not the same as a control that exists. Evaluation at six point four point four compares the result with the criteria and chooses among five outcomes, from doing nothing further, through treatment or further analysis or keeping existing controls, to reconsidering objectives. That last outcome gives a basis for going back to the appointing party about a milestone that cannot be met.
Treatment at clause six point five offers seven options, and the governing principle is that they change something about the risk, whether it is run at all, its source, its likelihood, its consequence or who carries it. Two examples show the range. To change the likelihood, Arden inducts the new task team and checks its first three containers before review. To retain a risk by informed decision, Arden accepts the old survey for feasibility only, records that, and requires a new survey before design freeze. The treatment plan should carry what registers rarely do, the rationale for the option, who is accountable for it and how the remaining risk is monitored. That is the answer to proportionality, the question every appointing party asks about how much information effort a risk justifies. The criteria say what the organisation will tolerate. The analysis says where the risk sits. The treatment weighs benefit against cost and effort. The record of those three moves is the justification, and it survives being written down. The episode Security-Minded Information Management in ISO 19650 Part 5 uses the same reasoning for sensitivity.
Then the last two parts of the process. Monitoring and review at clause six point six run at every stage, and each information exchange is a monitoring point. A criterion that never triggers over three milestones is one to question. Clause six point seven, on records and reports, asks for the process to be documented and reported so that it supports decisions and governance, and it cautions that records take account of information sensitivity, because a register describing a vulnerability in a built asset is itself sensitive. Now the test. Ask an organisation that says it follows ISO 31000 to show its criteria and its risk owners above project level. If the answer is the register, it has a register and no framework, and every level in it is a judgement with no stated basis. Arden's board register passes. Its project registers, until now, did not exist.
So the fourth step resolves. Information risk belongs in a register that inherits its criteria from the organisation's framework, walks the process in order, and records why each treatment was chosen.
The fifth step comes back to Block A, and to ISO 45001. The safety standard shares the skeleton and adds things an information manager has to know. Workers are a named group of interested parties. Clause five point four requires a process for consulting workers and for their participation, and it asks for timely access to information about the system that is clear, understandable and relevant. Read that as what it is, an information requirement in the standard's own words, close to a level of information need for workers. A health and safety register that only the adviser can read does not meet it. And the note to clause seven point five point three says that access to documented information includes access by workers and their representatives. Nothing in ISO 9001 matches it, and it is the note an information manager has most often not read. Hazard identification also consumes information by requirement. It must take into account past incidents and changes in knowledge about hazards, and the management of change process at clause eight point one point three treats new information about a hazard as a change in its own right. A revised risk record is a change event, not a filing event.
Three definitions decide what happens next. A nonconformity is the non fulfilment of a requirement, including one the organisation set for itself, so a container shared without its review breaches Arden's own procedure. An incident is an occurrence arising out of work that could or does result in injury or ill health, and a near miss is one. Corrective action eliminates the cause and prevents recurrence, and ISO 45001 widens it to cover incidents. The difference that the whole clause turns on is between correction and corrective action. Replacing the wrong revision on site is correction. Finding out why the wrong revision was there, and changing the process so it cannot be again, is corrective action. Most information failures get the first and never the second.
Clause ten point two runs as a sequence of records, each consumed by the next step. Follow Arden through it. First, react and correct. The fourteen records are updated from the workshop notes, the near miss is entered, and the exchange is resubmitted. It passes, and nothing has changed. Then evaluate the need for corrective action with the participation of workers, which means the operatives who found the clash. They say the card takes a week to reach the adviser and the register cannot be read on site, so the worker access the note requires is not met. Then investigate and find causes. The workshop notes were never transferred because no role owns the transfer, and the card route has no step into the register. Two missing process steps and one missing role. The immediate cause is no owner. The underlying cause is that the register was adopted for the project without a management of change review of the procedure it replaced. Then look for similar cases, and the search finds that no project register at Arden has ever been reconciled with the corporate spreadsheet except by hand, late or never. The problem is systemic.
The second half of the clause decides and checks. The riser risk is reassessed, with the near miss as evidence. The corrective action is chosen using the hierarchy of controls and passed through management of change. Here the project register is named as the record the management system relies on, an owner is assigned under clause five point three, the card route becomes electronic and runs straight into the incident records, and the register is made readable on site. Some of that is administrative control and some engineering, and the hierarchy ranks an engineering control, such as a re-coordinated model, above an administrative one, such as a briefing. Then the effectiveness review, which checks the next two exchanges and whether a near miss now reaches the register within a day. And the close, which retains the evidence and communicates it to the workers and their representatives. That communication is a requirement, and it is the step most often missed. The information cause was a management system cause. That is the finding of Block A.
Now the bridge, which is why these standards belong in a season on ISO 19650. Clause two of ISO 19650-6 lists ISO 31000 and ISO 45001 as normative references, cited without a date, and clause three applies their terms. A normative reference makes the cited content a requirement wherever the citing document invokes it, and because the references are undated, the latest edition applies, amendments included. So incident means what ISO 45001 says it means, including near misses. The level of risk field takes its meaning from ISO 31000, and a note records that ISO 45001 requires assessment to be systematic and proactive, so a level of risk is the output of a defined method with criteria, not a judgement. Part 6 also asks each party to release generalised lessons through its management system arrangements, which assumes a management system exists. Seen from that side, every part of the Part 6 schema has a clause that consumes it. Risk records are outputs of the assessment, incident records feed clause ten point two and return to hazard identification, and treatments sit on the hierarchy of controls, with information itself among the administrative controls. And clause nine makes the register an audit object, with ISO 45001 alone requiring relevant audit results to go back to workers. A register nobody monitors, audits or reviews is not part of a management system. It is a spreadsheet. The schema itself belongs to the episode Health and Safety Information in ISO 19650 Part 6.
So the fifth step resolves. At Block A, the Part 6 failure and the ISO 45001 failure are one failure seen from two documents, and corrective action finds it only if the information management function sits inside the system that runs the process.
The sixth step is the decision Arden has been avoiding. There are three options. A separate information management system is the silo. Nobody audits it, nobody reviews it, and it fails the clause five point one commitment to integrate. Parallel documents inside one system, an information management procedure cross referenced from the manual, are cheap and adequate where the function is small, but the rules for retention, competence and audit then live in a document nobody else reads. Integration clause by clause writes information management into the existing clauses wherever the harmonised structure has a place for it. It is the most work and gives the most protection. Arden has been running a fourth option, which is not deciding, and paying the cost of the silo without the benefit of having chosen anything.
The harmonised structure was designed for integration, and both requirements standards say so in their introductions. Annex A of ISO 9001 adds two points that stop integration becoming a rewrite. The standard's clause order is not a model for the organisation's own documents, and the organisation may keep its own terms. Arden's manual keeps its own numbering. And introducing information management is itself a planned change. Clause six point three of ISO 9001 asks for the purpose and consequences of a change, the integrity of the system, the resources and any responsibilities that move to be considered first, and clause eight point one point three of ISO 45001 asks for a process for planned changes. Block A showed the cost of skipping that. Arden's board chooses integration. The head of quality, safety and environment owns the change. The design manager who led the Northgate Quarter function becomes process owner, and no information manager post is created.
Every paragraph drafted into the manual has to pass four tests. It names the clause it satisfies. It names where it goes in the existing manual, in the organisation's own numbering, never a new manual, because a new manual is the silo with a cover. It names the existing role that owns it, never one invented to receive it. And it names the evidence an auditor will sample. The fourth test separates a manual that passes audit from one that decorates a shelf. Here is a paragraph that fails. The organisation shall control documented information in accordance with clause seven point five. It is true of a pump manufacturer and gives an auditor nothing to sample. Here is one that passes. Project information containers carry a revision code and a status code under the project's information standard, superseded revisions are kept in the archive, and native files are not edited once shared. Its owner is the head of quality, safety and environment, and its evidence is container metadata. The test is simple. If a paragraph would work unchanged for a manufacturer, it is not about information. A few drafting habits keep it auditable. Put undecided values, such as a retention period that needs legal advice, in square brackets and list them for the change plan to close. Cross reference rather than repeat. And paraphrase the standard and cite it by clause rather than paste its text.
Then audit, because clauses that are never audited decay. Clause nine point two asks for a planned programme that weighs the importance of each process and the results of earlier audits, so a newly integrated process is audited soon. Auditors are chosen for objectivity, so the process owner does not audit the process. Both standards point to ISO 19011 for guidance on auditing, and that guidance was itself revised in 2026. Each drafted clause becomes a question the auditor can answer yes or no by looking at something. Is there a roles register entry for the function on every current appointment, with the delegated authority stated, and does the nominated person know what it means? Do ten sampled containers per appointment carry revision and status codes, and do their native files match the shared revision? And for Block A, did the last near miss reach the register within a day? An information audit against ISO 19650 checks the series, and it sits beside these criteria, not in their place, because conforming to the series does not mean conforming to ISO 9001, and the reverse is also true. Management review at clause nine point three then takes the results, and it is where the retention period gets decided and a deputy for the function gets funded. Continual improvement at clause ten point three takes what each close-out taught, held as organisational knowledge. The cycle closes, and the information management function goes round it with everything else Arden does.
Staying current is the same discipline. Keep the edition register. At each annual review, check the ISO catalogue record for every standard rather than commentary about it. Treat each publication as a planned change. And re-audit the drafted clauses against a new edition once it is adopted, because clause numbers and terms may move even where requirements do not. Two neighbours share the same skeleton. ISO 55001, in its second edition of 2024, is the asset management system standard, and where an appointing party runs one, its organisational and asset information requirements will come from it. And ISO/IEC 42001, the management system standard for artificial intelligence, poses the same question for any organisation bringing artificial intelligence tools into information production, which is where this episode began. That resolves the sixth step. Integrated clause by clause, drafted to four tests, audited against criteria and kept current, information management becomes part of the system rather than a document beside it.
Let me bring Arden back to the riser in Block A. The question was where information management should sit inside an organisation's quality, risk and safety systems, and what changes when it sits there. It sits inside them, clause by clause, under the harmonised structure they share. Its containers are documented information under clause seven point five. Its function is a clause five point three assignment with no invented role, resourced through clause seven. Its risks sit in a register that inherits the organisation's ISO 31000 criteria. Its failures go through clause ten point two. And it is audited and reviewed like everything else. What changes is that the fourteen unknown records and the lost near miss card stop being nobody's problem. They become a nonconformity with an owner, a cause and a corrective action that someone checks.
Three changes follow for any organisation like Arden. Before you write any information procedure, ask which existing clause it belongs under, and put it there rather than beside the system. Before you accept a risk register, ask to see the criteria and the owners above project level, because a register without a framework is a list of opinions. And when an information failure happens, refuse to stop at correction, and look for the missing role or the change that never went through management of change.
For reading, start with clauses five, seven and ten of ISO 9001, clause six of ISO 31000, and clauses five point four and ten point two of ISO 45001, then clauses two and three of ISO 19650-6 for the normative references, and clauses five point one point one and five point three point one of ISO 19650-2 for the function. The IMI Framework guidance on the information management function is the readable companion. The episodes Health and Safety Information in ISO 19650 Part 6, Roles and the Responsibility Matrix, and The Common Data Environment: Process Before Platform sit closest to this one. The next episode, The UK BIM Framework and the Move from PAS 1192, is about what the UK BIM Framework is and is not, and how a legacy document set maps onto the current standards.
Take the quality manual of a firm you know, from your own practice or from a research case, open its section on documents and records, and see whether it knows a common data environment exists. I am Dr Sepehr Abrishami, and on Reimagining AEC that is the page I would always read first, because it shows whether information management is part of the system or only a guest beside it.