LO
Get told when a verified record lands on this occupation → · Mark which of these tasks are yours (VOLO Pro, free) → · For a team or a graduating class →
Business systems owner
The person who owns the system the business runs inside — the CRM or order system: how the process is shaped in it, whether its records still mean anything, and what the numbers coming out of it are allowed to claim.
Write down the three numbers your system is supposed to move and where each one currently stands. If you cannot fill in the second column without asking someone, you are administering an instance rather than owning an outcome — and that gap is the whole conversation.
This is not a probability of losing your job. It combines how much of the role's task load is exposed to automation with how far adoption has actually gone — useful for comparing occupations on one consistent basis, and for nothing else.
The internal owner of a company's CRM, order or workflow system — often called sales ops, revenue operations or business systems, and in smaller firms simply the CRM team. Not the vendor's support desk, and not a software engineer building the product the company sells: this role configures, governs and explains a system somebody else built.
What is actually changing#
The unit of analysis is the task, not the job title. A role is not replaced — its task mix shifts.
Each tile is one task. Its size is how much of the job it is; its colour is where the task is heading. Click a tile to see what the judgement does not establish.
Faster configuration means more configuration, not less work. Every rule added is a rule that will later be wrong and have to be found — so the role's hours move from building to untangling, and untangling is invisible on a roadmap.
Cleaner data makes everything downstream more trustworthy, which is the point — and it also removes the only visible artefact of this job. Nobody notices records that were never wrong, so the work reads as an absence, which is a poor position when the budget is set.
Setting the boundary is not the same as being allowed to hold it. This role usually has the knowledge and not the authority: when sales wants the automation to send more, the person who can say no is further up, and being overruled repeatedly is how the boundary erodes without any decision ever being recorded.
Self-service reporting does not remove the person who knows the number is wrong. It multiplies the number of people confidently quoting a figure whose caveat they never saw — so the work shifts from producing reports to correcting them in meetings, which is slower and harder to defend as a headcount.
This is the first task cut when the role is under pressure, because it produces nothing shippable. Cutting it does not show up as a problem for two quarters, and then shows up as data nobody trusts — by which time the cause is no longer attributable.
Is this your job? Say so and this page narrows to your share of it.
A job title is a bundle of tasks bought together, and no two people hold the same bundle. Nothing is sent anywhere — it stays in this browser.
Read all 5 tasks in full — direction, reasoning and limits →
Recent changes#
Mainland China. Under the interim measures for generative AI services, a service is filed with the national regulator and an application that calls a filed model through an API is registered with the provincial one; a live application must show, prominently or on its product detail page, the model name and the filing or listing number. For whoever owns a business system, that makes switching on a generative feature a step with a named, checkable precondition rather than a configuration change. The counts are of filings, not of usage: they say how many services cleared the gate, not how many companies deployed them, how well they work, or whether anyone checked the disclosure. It says nothing about systems outside China.
Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.
All US firms, nationally representative; reference period Nov 2025 - Jan 2026. The 16% is of AI-using firms, not of all firms. The survey asks the firm what it replaced; it does not identify which system, which vendor, or who administered it, so nothing here says a particular CRM or order system was swapped. Read beside the same survey's labour figures rather than instead of them: among AI-using firms 95% report no AI-related employment change at all, with increases 2.3% and decreases 2.0% (firm-weighted), while among firms that substitute tasks with AI the share reporting a large number of replaced tasks rose from 2.5% to 7% since the first supplement - a rise the authors caution may partly reflect changes to the surrounding survey questions, though the wording of the task-replacement question itself stayed the same.
An employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.
The department's own transparency record for the matching model its data linking team built. It says the model compares records of individuals within and across case management systems in prisons, probation and the courts and scores the probability that two records are the same person; that deterministic, rules-based linkage was considered and judged unsuitable for its inability to capture nuance, high false negatives and the difficulty of managing many rules; that it runs continuously on millions of records per system; and that where it offers a person a list of possible matches, the person makes the final decision, with training for staff who merge and unmerge records. It publishes no accuracy figures and says clerical labels cannot be treated as ground truth. A government justice system, not a company's CRM.
An employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.
The department's own monitoring and evaluation team reporting on 1,000 Microsoft 365 Copilot licences given to UK-based staff from October to December 2024, about 70% volunteers. It found that colleagues' attitudes ranged from encouraging to discouraging and that the attitudes of an individual's team and line manager appeared to influence how much and how the individual used the tool; that training was a significant factor in satisfaction, with self-led training more effective than formal departmental sessions; that colleagues outside the pilot lacked knowledge of its capabilities, limitations and security; and that in observed tasks users did Excel data analysis more slowly and less accurately than non-users. It is an office AI assistant, not a CRM or order system, and the findings are not a controlled measurement of adoption.
Small-scale trial in a real setting. Tells us the deployment conditions are being tested, not that they hold — so one pilot is never enough on its own; two independent ones are.
LinkedIn's engineers describing an internal tool, SQL Bot, which the company does not sell. They say data experts spend a significant amount of their time helping colleagues find data they need, delaying business partners; that SQL Bot is now used by hundreds of employees across LinkedIn's business verticals, with sustained adoption after it was integrated into the company's data-science notebook platform; that it only became usable after domain experts identified key tables in their areas and wrote mandatory table descriptions; and that in a survey about 95% rated its query accuracy "Passes" or above and about 40% "Very Good" or "Excellent". Adoption and accuracy are self-reported; users include analysts as well as business staff.
An employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.
The Treasury's in-house evaluation unit reporting on a trial of Microsoft 365 Copilot by 218 staff from 20 May to 23 August 2024, part of an Australian Government trial. It says the product was used less than originally expected, with about a quarter using it frequently by the end; that lower use likely came from initial high expectations that diminished in practice, technology limitations, and a lack of education and training on how to use the product; and that 42 participants self-nominated as trial champions tasked with supporting their teams. Findings rest on surveys (136 post-trial responses) and focus groups. It is an office AI assistant in a policy department, not a CRM or order system.
Small-scale trial in a real setting. Tells us the deployment conditions are being tested, not that they hold — so one pilot is never enough on its own; two independent ones are.
The EU, and only for deployers of high-risk AI systems. What it establishes is that on the input side of these systems, clean data moves from an internal good habit to a legal obligation — and in most companies the party controlling that input data is the person who owns the system. Article 26(1) additionally requires deployers to use such systems in accordance with their instructions for use. Regulation (EU) 2026/1744 (the Digital Omnibus on AI, published 24 July 2026) moved the application of these obligations from 2 August 2026 to 2 December 2027 for systems that are high-risk under Annex III, and to 2 August 2028 for those that are high-risk under Article 6(1) and Annex I; so this is a rule on the books that does not yet apply. The law names no job title and does not require a dedicated data owner, so attaching the duty to this occupation is our inference; it covers only systems classified as high-risk.
Regulation, subsidy or public procurement is requiring or funding adoption — the mirror of a constraint. It shows adoption is being required, not that it has happened, so one mandate is never enough on its own; two independent ones are.
The agency's entry in the federal AI inventory, reported as deployed with an operational date of 1 February 2023. It says the model identifies which case management records belong to the same individual with a high degree of confidence, drawing on seven form-processing systems; that machine learning is more resilient to fuzzy matches and varying data fill rates and improves entity resolution compared with a rules-based system; that a record is attached only above a likelihood score of 0.98; and that false positives or negatives are sent to a manual resolution queue processed by authorised and trained personnel. It gives no accuracy or volume figures, and it is a government benefits system, not a company's CRM.
An employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.
What this means for you#
This has been one of the reliable ways into a commercial career without a technical degree: learn the system, become the person who knows where everything is, move into operations or analysis. The learn-the-system part is exactly what assistants now do well, so the rung has thinned. What is still scarce, and still learned only by being there, is knowing why the process is shaped the way it is — including the parts that exist because of an argument three years ago.
You are the only person who knows which numbers are load-bearing and which are decoration, and that knowledge is not in the system you administer. The risk is that your value looks like maintenance from outside: when the data is clean and the reports run, the work is invisible, and invisible work is what gets outsourced to the vendor's support plan.
Your options#
Four directions, each with its real constraints and one thing you can test this week. Continuing as you are is a legitimate choice — it just has to be a chosen one.
Own an outcome, not an instance
Administering a system is a cost line; moving a business number is not. The systems owners who stay are the ones whose remit is stated as the number the system exists to move, with the configuration as the means rather than the job.
It means being accountable for a number other people also affect, and giving up the defensible position of having done exactly what was asked.
Write down the three numbers your system is supposed to move and where each one currently stands. If you cannot fill in the second column without asking someone, you are administering an instance rather than owning an outcome — and that gap is the whole conversation.
Write down what the automation may do
Now that the system acts rather than records, someone has to set and hold that boundary. You are the only person who can see both the configuration and the business consequence — and in most companies this has been settled by default rather than decided.
You will be overruled by whoever owns the revenue number, repeatedly. A boundary without a written rule and a named approver erodes quietly, and the erosion is only visible after an incident.
List every automated action your system takes today without a person: assignments, scores, messages, status changes. Mark each one fine, should be reviewed, or should never have been automatic. Bring three counts, not an opinion.
Move to the function you have been serving
You already know a function's process in more detail than most people working in it, because you encoded it. Sales, finance and operations all now need someone who understands both the work and the system it runs in.
You trade a position where correctness is provable for one where judgement is argued, and you will be junior in domain terms for a year even though you know the plumbing better than anyone.
Take the last ten change requests you received from one team and write what each was really trying to achieve, underneath what it asked for. If your version is consistently better than theirs, you already understand their job — which is the whole case.
Cross into data governance
Whose record is authoritative, what a field is allowed to mean, who may change it and what gets logged — those questions are becoming a function of their own as more decisions get made automatically from that data, and you have been answering them informally for years.
It is a policy job with a compliance flavour, measured over years rather than sprints, and it exists as a funded role mainly in regulated industries or companies large enough to have been burned.
Pick the one field in your system that most decisions depend on and try to write its definition in two sentences — what it means, who may set it, when it is wrong. Then show it to two people who use it. If they disagree with each other, you have found the job.
Common questions#
The parts that are fetching and cleaning — deduplicating records, writing the query behind a report — are automating, and configuring the system is being assisted rather than removed. The part that is not is knowing why the process is shaped the way it is and which numbers are load-bearing. Expect fewer of these roles, held by people closer to the business, and expect the first cut to be the person who only administers.
Count the requests. Take the change and report requests your team received last quarter and compare them with the same quarter a year ago. If the count has fallen while the business has not shrunk, the self-service tooling is absorbing your queue — and that shows up before anyone reconsiders the headcount. What it does not tell you is whether the requests stopped or just stopped reaching you, so check where they went before you read the drop either way.
It removes the queue and keeps the liability. More people now quote figures whose caveats they have never seen, and you remain the only person who knows which ones are held together by an assumption. That is more valuable work and less visible work at the same time — so the practical move is to make the caveats part of the report rather than part of your memory.
Yes for ranking, carefully for anything that closes a record or sends something out, and never without writing down what happens when it is unsure. The expensive failures in this job are not wrong scores, they are silent state changes: a record closed, an owner reassigned, a message sent, each of which looks like a normal row afterwards. Decide the boundary in writing while nothing has gone wrong, because after an incident it gets decided for you.
Studying towards this?
These majors lead here. Their pages break down which of their competencies transfer and what graduates typically lack.
What these judgements rest on#
5 of 5 task judgements on this page are backed by a verified event and 0 are platform inference, each labelled where it appears. Behind them sit 2 technology dimensions, a reconstructed trajectory since language models reached the public, and 8 verified events.
See which technologies, how it got here, and the method →
Where it sits in the official classification: skills, knowledge, related jobs →
Other roles in the same function#
A company divides its work into functions before it divides it into jobs. These sit in Technology & data alongside this one — a fact about org charts, not a judgement that they are similar or that they are changing in the same direction.
Junior software developer · Experienced software engineer · Frontend developer · Backend developer · Data engineer · Machine learning engineer · AI researcher · Software tester / QA engineer · Data analyst · Data scientist · Statistician · Economist · Business analyst · IT support specialist / helpdesk · Security analyst (SOC) · DevOps engineer / platform engineer / SRE · Network engineer · Technical writer / documentation engineer · Grant writer / grants officer