The reason to sort out the work before buying ERP or groupware is simple: the software does not decide how your company works. If work is assigned over chat and decisions live in phone calls and meetings, a new system does not change that pattern. It adds one more job: typing the results in afterwards. So the thing to have in hand before comparing products is not a requirements list but a set of audit deliverables that record how the company actually runs today. This article lists those seven deliverables, taken from a work design report we delivered to a client.
What this article covers
- Why the work has to be sorted out before buying ERP — a conclusion repeated since 1990
- What actually happens when you add one more tool
- The seven audit deliverables — what each records, and which purchasing question it answers
- Four criteria for choosing ERP or groupware using those deliverables
- How to get the audit, and what it costs
Why sort out the work before buying ERP — a 35-year-old conclusion
That layering technology on untouched work fails is not a new finding. In the July–August 1990 issue of Harvard Business Review, Michael Hammer published “Reengineering Work: Don’t Automate, Obliterate”. The title is the argument: using information technology to run an existing procedure faster only speeds up its inefficiencies along with it.
In the July–August 1998 issue, Thomas H. Davenport’s “Putting the Enterprise into the Enterprise System” pointed out that enterprise systems such as ERP impose their own logic on a company’s strategy, culture and organization, and cautioned against handing responsibility for adoption to technologists because the business implications are so large. The company changes to fit the flow the package has decided on, and whether that flow fits the company is a management question, not a technical one.
Section 3.6 of our published work design report sample writes the same conclusion as a working order. Before deciding whether to automate anything, every task is examined in the sequence Eliminate → Simplify → Standardise → Integrate → Automate. Skipping that order and starting with automation only means repeating removable work faster. The report’s conclusion (chapter 9) says the same thing: if, after the system is in, work is still assigned by chat, progress is still reported separately and only results are typed into the system, the system becomes a new data-entry job.
What actually happens when you add one more tool
Adding a tool to unsorted work adds a channel. According to Microsoft’s Work Trend Index special report “Breaking down the infinite workday” (June 2025), the average Microsoft 365 user receives 117 emails and 153 Teams messages a day and is interrupted every two minutes during core working hours — 275 times a day — by meetings, emails or notifications. Those numbers are not the result of having too few tools.
Section 3.5 of the report sample reaches the same diagnosis. In the law firm observed, the important technology problem was not a lack of new technology but that work and information were split across messenger, email, an AI tool, the court system and the work management system, with people doing the connecting in between. Sorting out the work before buying ERP means finding and writing down exactly those “people-connected” gaps. Only once they are written down can you compare whether a new system closes a gap or opens one more.
Seven audit deliverables to have before buying ERP
The seven items below are tables that actually appear in chapters 2 to 4 of the report sample. The industry is a law firm, but the frame of each table is industry-neutral. For each one we note what it records and which purchasing question it answers.
| # | Deliverable | What it records | In the sample |
|---|---|---|---|
| 1 | Evidence grading table | Whether each finding was directly observed (Observed), or comes from documents (Documented), interviews (Reported), interpretation (Analysis) or is unconfirmed (To Verify) | Section 2.2 |
| 2 | Value structure table | For each task, whether it is core professional work (keep) or a candidate for simplification, standardisation or reduction | Section 3.1 |
| 3 | Workflow diagnosis table | Current state and direction for the unit of management, work requests, review and progress sharing | Section 3.2 |
| 4 | Decision classification table | Six types of request reaching the owner — three information dependency, three decision dependency | Section 3.3 |
| 5 | Information structure table | Where client, matter, task, consultation, email, file, decision and billing information should attach | Sections 3.4–3.5 |
| 6 | Waste and automation table | 13 tasks classified from Eliminate to Automate, including a “not possible” ruling on an integration | Section 3.6 |
| 7 | Operating problem priority table | Findings from seven areas grouped by cause into five operating problems, ranked P1/P2 | Chapter 4 |
Deliverable 1 — Evidence grading table: how much was actually seen
Not every sentence in an audit carries the same weight. Section 2.2 of the sample labels each item as Observed (confirmed on site), Documented (from client-provided material), Reported (explained by the owner or staff), Analysis (a judgement derived from observation and material) or To Verify (cannot be confirmed on current evidence). This table is what lets a reader tell which cells in the next six tables are fact and which are interpretation. In section 2.3 of the sample, eight items — matter management, work communication, owner dependency, administrative work and others — are marked Observed, and two — document review and consultation automation — are marked Documented.
Deliverable 2 — Value structure table: mark what must not be cut first
If a task is judged inefficient simply because it takes a long time, professional work ends up on the automation list. Section 3.1 of the sample sorts seven tasks into “core professional work → keep”, “necessary but approach can improve → simplify” and “repeated reprocessing of existing information → reduce”. The owner’s client consultations and legal judgement are kept; reorganising matter progress information and the owner’s simple status checks are reduced. Without this table, an ERP requirements list becomes “put every task into the system”, and that requirement becomes data-entry load as written.
Deliverable 3 — Workflow diagnosis table: where the work breaks
Section 3.2 of the sample simplifies the observed flow to “client or matter arises → individual task → assignee performs it → confirmed or shared by messenger and meetings → owner checks and decides → revision → next task”, then records the current state and direction for six diagnostic items. The basic unit of management goes from task-centred to matter-centred, work requests from external messenger to Task-centred, and progress sharing from separate communication to reading from the system. The “direction” column of this table is the structure to demand from a system.
Deliverable 4 — Decision classification table: leave the owner only real decisions
This table (section 3.3) splits requests reaching the owner into Information, Status and Context (information dependency) and Approval, Professional Decision and Exception (decision dependency). The first three are routed to system lookups, task screens and task history; only the last three are judged by a person. How the table is used is covered separately in “Staff keep asking the owner”. Switch on a groupware approval workflow without this table and information questions climb the approval chain too, which only lengthens the owner’s approval queue.
Deliverable 5 — Information structure table: where information should attach
Section 3.4 of the sample lists ten kinds of information — client, matter, task, consultation, email, file, decision, schedule, billing and court matter information — and where each should attach: Customer, Project or Task. The three axes are Customer (whose work is it) / Project (which matter) / Task (what has to be done now). Section 3.5 then records the current use and direction for five tools: messenger, email, AI tool, court system and work management system. This is the table used most in product comparison. It is where you find out whether a candidate product holds those three axes as one structure, or whether “projects” and “clients” live apart.
Deliverable 6 — Waste and automation table: what to automate, and what cannot be
Section 3.6 of the sample classifies 13 tasks, from client phone consultations to repeated document drafting, as Keep / Eliminate / Simplify / Standardise / Integrate / Automate. Of the 13, only three carry Automate — creating follow-up work after consultation, re-summarising email and work content, and repeated document drafting — while “simple information questions to the owner” is Eliminate. The table also records what is not possible. Court matter information was confirmed to have no official API integration and is excluded from custom automation: the Korean court Judicial Information Sharing Portal provides basic matter information and progress history only through its linked API and states that an open API is “to be provided later”. If a vendor proposal advertises “automated court integration”, one line of this table settles it.
Deliverable 7 — Operating problem priority table: what comes first
Group the findings from the seven diagnostic areas by shared cause and they collapse into a handful of operating problems. Chapter 4 of the sample has five: OP-01 lack of a matter-centred operating system, OP-02 work, information and communication scattered, OP-03 information dependency and decision dependency on the owner mixed together, OP-04 repeated reconstruction of context before work, OP-05 hand-offs between consultation, external information and work. The first four are P1, the last is P2. Whether a P1 is solved by installing a system or by changing how work is done is the criterion that carries into the next chapter (solution design criteria, section 5.1).
After these seven come solution design (chapter 5), the target operating system (chapter 6), the roadmap (chapter 7) and measurement (chapter 8). The measurement chapter contains no “30% saving” figure. The report’s principle is that no savings rate is stated before quantitative data exists; in its place are 13 KPIs for which baseline values are captured first.
Four criteria for choosing ERP or groupware with the seven deliverables
- Does the product hold the three axes of the information structure table (deliverable 5) as they are? Check that client → matter (project) → task runs as one line and that consultations, files, emails and decisions attach to the matter. The criterion is the connecting structure, not the feature count.
- Does the product cover the tasks classified Integrate in the waste table (deliverable 6)? Check that work requests by messenger, confirmation and revision by messenger, and billing work are handled inside tasks and the matter record. A product that adds features to tasks classified Eliminate earns no credit.
- Exclude features that presuppose an integration ruled not possible. A proposal that designs around something deliverable 6 has recorded as “not possible” collapses after go-live.
- Can the system hold the review and approval criteria from the decision table (deliverable 4)? Check that each approval step can carry a written criterion and that normal progress flows without approval. A structure that demands sign-off at every step has only moved owner dependency into the system.
After go-live, use the 30-day review in section 8.4 of the sample as it stands: are active matters managed as projects, are work requests and confirmations still happening in messenger, are things that can be read from the system still being asked of the owner. If any of the three is “yes”, the first suspect is not the product but a skipped Phase 1 (working-method reset, section 7.1).
How to get these deliverables
The seven tables above are part of the operating system audit report produced by the operating system audit. The audit has a fixed price. Basic is KRW 800,000 (excluding VAT): one work flow or one department, three days of on-site observation and five days for the report. Advanced is KRW 1,200,000 (excluding VAT): two or more work flows or the whole company, five days of observation and ten days for the report, with a draft Workflow Blueprint and an implementation roadmap added to the audit report. A full refund applies before observation begins. The last chapter of the report carries a written quote for design and build with scope, schedule and price, but the report is yours whether or not you build with us. The full report is on the sample page.
References
- Harvard Business Review — Reengineering Work: Don’t Automate, Obliterate (Michael Hammer, 1990)
- Harvard Business Review — Putting the Enterprise into the Enterprise System (Thomas H. Davenport, 1998)
- Microsoft WorkLab — Work Trend Index Special Report: Breaking down the infinite workday (2025)
- Korean Court Judicial Information Sharing Portal — linked API guide (open API to be provided later)
Focus keyword: before buying ERP
SEO keywords: before buying ERP, groupware adoption, work design, operating system audit, audit deliverables, business process review, ERP readiness
자주 묻는 질문
What should be ready before choosing ERP or groupware?
Audit deliverables that record how work actually runs today, rather than a requirements list. Three are used directly in product comparison: the information structure table (where client, matter and task information attach), the waste and automation table, and the decision classification table.
What happens if we adopt a system without an audit?
As the report's conclusion puts it, chat instructions and separate progress reports remain and only results get typed into the system, so the system becomes a new data-entry job. Check 30 days after go-live whether work requests and confirmations are still happening in messenger.
What does the audit cost and how long does it take?
Basic is KRW 800,000 (one work flow or one department, three days of observation and five days for the report); Advanced is KRW 1,200,000 (two or more work flows or the whole company, five days of observation and ten days for the report), excluding VAT. A full refund applies before observation begins.