The Roster That Lives in One Person’s Head
Every operation that schedules people has one person who knows the roster. They know who is on leave, who is qualified for what, who does not work Fridays, who cannot be sent to a particular site, and which two people should not be put on the same job.
That knowledge is genuinely valuable and it is also the problem. It means the roster is a spreadsheet maintained by one person, published by email or printed on a wall, and correct until it changes. When it changes, some people find out and some do not. When that person is on leave, scheduling becomes noticeably worse for a week.
The symptoms are familiar. Somebody turns up for a shift that was reassigned. A job goes out with two people when one would have done, because nobody could see who else was free. A manager asks whether the team can take on more work and the honest answer is a shrug.
This article covers where Planning sits relative to Field Service (a distinction worth getting right before you start), what resource scheduling needs to know, how to design roles and skills without over-engineering them, and how to handle the absence and change that makes rosters go stale. It is written for operations and service managers.
Planning or Field Service? Know Which Problem You Have
These two get confused regularly, and the confusion leads to implementing the wrong thing.
Field Service dispatches jobs to customer sites. The unit of work is a customer job: a visit, a repair, an installation. It carries the customer, the location, the task, parts, time recording against that job, and usually an invoice at the end. The question it answers is “who goes to this customer, when.”
Planning schedules internal people and resources against time. The unit of work is a shift or an allocation: someone is working, in this role, between these hours. It is not necessarily tied to a customer at all. The question it answers is “who is working when, and on what.”
Many businesses need both, and they answer different halves of the same day. A service business dispatches jobs to customers and needs to know who is rostered on Tuesday, who is qualified, and who is already committed.
How to tell which you need first. If your pain is that a customer visit went to the wrong engineer or nobody can see job status, that is Field Service. If your pain is that you do not know who is working, who is free, or whether the team has capacity for more work, that is Planning. If both, start with whichever one is costing you more this month.
Odoo capability varies by version and edition, so confirm what is available to you before designing around either, rather than assuming a particular combination.
The Business Problem: Capacity You Cannot See
- The roster is a document, not a system. Correct when published, stale thereafter, and only one person can update it.
- Availability is not visible. Leave is in one place, existing assignments in another, and nobody holds the combined view that would say who is actually free on Thursday.
- Skills and qualifications are informal. Who can do what lives in a manager’s knowledge, so assignment depends on that person being available to ask.
- Double-booking happens routinely. Two assignments for the same person at the same time, discovered on the day.
- Changes do not reach everyone. A reassignment is communicated to those the scheduler thinks of, which is not always everyone affected.
- Utilisation is unknown. Whether the team is at eighty per cent or sixty per cent is a question with no answer, which makes hiring and work-acceptance decisions guesswork.
- Overtime is reactive. Approved in the moment to cover a gap that a visible forward view would have shown a week earlier.
What Bad Scheduling Costs
- Paid time not deployed. The largest and least visible cost. People are paid whether or not they are assigned, and an operation with no utilisation view carries unused capacity without knowing it.
- Overtime that was avoidable. Gaps covered at premium rates that forward visibility would have let you fill with existing capacity.
- Work turned away unnecessarily. The manager who cannot see capacity declines work to be safe, and the business never learns what it could have taken.
- Service failures from double-booking. A customer or a site left uncovered because two commitments collided.
- Compliance exposure on hours and rest. Where working time rules apply, an informal roster cannot evidence compliance even when it is compliant.
- Staff goodwill. Late notice of shifts, reassignments people learn about by arriving, and a perception that allocation is arbitrary. This is the cost that shows up later as turnover.
- Key-person dependency. The scheduler’s absence is disproportionately disruptive, which is the clearest signal the roster is not a system.
What Resource Scheduling Needs to Know
Four things, and most operations are weak on at least two.
- Who exists and when they work. Contracted hours, working patterns, part-time arrangements, and the fixed constraints, such as someone who does not work Fridays, someone who starts late on Tuesdays. These are the rules the schedule must respect rather than discover.
- Who is unavailable, and when. Leave, training, sickness and any other commitment. Crucially this must be the same source the schedule reads, not a separate leave sheet. A roster that does not see approved leave will schedule somebody who is away.
- What each person can do. Roles, skills, certifications and their expiry dates where relevant. This is what allows the schedule to propose someone qualified rather than someone available.
- What work needs covering. Shifts, allocations or jobs, with their duration and the role required. Without this the schedule is a staff list rather than a plan.
A fifth, which is a decision rather than data: who is allowed to change the schedule, and what happens when they do. Without that, the schedule drifts back into being one person’s document with extra steps.
Designing Roles and Skills Without Over-Engineering
This is where implementations get bogged down, and the failure is almost always excessive precision.
- Define roles at the level you actually schedule. If you assign “an electrician” to a shift, electrician is a role. If you need to distinguish between three grades of electrician because they can do materially different work, that is three roles. If the distinction never changes an assignment decision, it is not a role; it is information that belongs on the employee record.
- Keep the list short enough to maintain. A role model with forty entries will be out of date within a year because nobody will keep it current. Ten to fifteen is a workable range for most operations.
- Track certifications only where they gate work. A qualification that legally or contractually determines who may do something belongs in the system with its expiry date. A general competency does not.
- Model the constraints that actually bite. Someone who cannot be sent to a particular site, two people who should not work the same shift, a person who must be paired with a supervisor. Each of these is a rule the scheduler currently holds in their head, and capturing them is most of the value of the exercise.
- Accept that some judgement stays human. The goal is a schedule that proposes sensibly and flags conflicts, not one that removes the scheduler. An operation that tries to automate every nuance builds a model nobody can maintain.
Publishing, Visibility and the Change Nobody Saw
The scheduling failure people remember is almost never a planning failure. It is a communication failure.
- Decide what “published” means. A schedule exists in two states: being worked on, and committed. People need to know which they are looking at, or they will plan their week around a draft.
- Give people access to their own schedule. Self-service visibility removes most of the inbound questions a scheduler handles, and it removes the excuse that someone was not told.
- Notify on change, not just on publication. The original roster reaching everyone is easy. The Tuesday reassignment reaching the two people affected is where it fails.
- Set a notice standard and hold to it. How far ahead the schedule is published, and what constitutes acceptable notice for a change. This is partly an employment matter and partly a goodwill one, and either way it should be a stated commitment rather than whatever happens.
- Keep a record of what was published and when. For disputes about whether someone was told, and for any working-time evidence you need to produce.
Handling Absence, Overtime and the Unplanned
Every schedule meets reality within days. What distinguishes a workable system is how it handles that.
- Short-notice absence needs a defined response. Who finds cover, from where, and what they are authorised to approve. In most operations this is the scheduler making calls and it works; the improvement is that a visible availability view makes the calls targeted rather than speculative.
- Forward visibility is what converts overtime from reactive to planned. A gap visible a week out can often be covered within normal hours by moving an assignment. The same gap discovered on the day costs a premium.
- Watch for the pattern, not just the instance. Repeated short-notice gaps at the same point in the week, or on the same team, are telling you something about the underlying roster rather than about bad luck.
- Record why cover was needed. Sickness, leave, demand spike, no-show. Over a quarter this becomes an operational picture rather than a sequence of individual problems.
- Have a position on under-use as well as over-use. Most operations manage gaps attentively and ignore the opposite. Unassigned capacity on a Thursday is a cost too, and it is the one that never prompts a phone call.
Approaches Compared
| Consideration | Spreadsheet roster | Shared calendar | Odoo Planning | Dedicated workforce management |
|---|---|---|---|---|
| Who can update | Usually one person | Several, without control | Controlled by permission | Controlled |
| Sees approved leave | No, separate sheet | Sometimes | Yes, where leave is in the same system | Yes |
| Skill and role matching | In the scheduler’s head | No | Roles and skills where modelled | Native, usually deeper |
| Conflict detection | Visual, if noticed | Limited | Flagged | Native, with rules |
| Staff self-service view | No | Partial | Yes | Yes |
| Change notification | Manual | Calendar update | On change | Native, often with acknowledgement |
| Utilisation reporting | Manual | None | Available | Extensive |
| Working-time compliance evidence | None | None | Partial, depends on configuration | Native |
| Best fit | Very small, stable teams | Teams with simple, infrequent change | Most mid-market operations, especially on Odoo already | Large, complex or heavily regulated workforces |
Best Practices
- Settle the Planning-versus-Field-Service question first. They solve different problems and implementing the wrong one is expensive.
- Make leave and the schedule read from the same source, or the roster will assign people who are away.
- Define roles at the level you actually schedule, and keep the list short enough to stay current.
- Capture the constraints the scheduler holds in their head: site restrictions, pairings, individual arrangements. This is most of the value.
- Track certifications only where they gate work, with expiry dates.
- Distinguish draft from published, and make sure people know which they are seeing.
- Notify on change, not only on publication. This is where scheduling fails in practice.
- Set and honour a notice standard for publication and for changes.
- Report utilisation, including under-use. Gaps get attention; idle capacity does not, and it costs the same.
- Record the reason cover was needed, so patterns become visible over a quarter.
Common Mistakes
- Implementing Field Service when the problem is rostering, or the reverse. Different units of work entirely.
- A schedule that cannot see approved leave. Guarantees assignments to people who are away, and destroys trust immediately.
- Over-engineering the role model. Forty roles will be stale within a year; nobody maintains a model that detailed.
- Modelling skills that never change an assignment decision. Effort with no output.
- Leaving the scheduler’s informal constraints uncaptured. The system then produces assignments that are technically valid and practically wrong.
- No distinction between draft and published. People plan their week around provisional work.
- Notifying on publication but not on change. The reassignment nobody saw is the failure people remember.
- Managing gaps and ignoring idle capacity. Half the utilisation picture.
- Trying to automate every nuance of judgement. Produces an unmaintainable model and an unhappy scheduler.
- Keeping the spreadsheet in parallel. It wins, because it is faster for the person who knows it.
An Illustrative Scenario: A Facilities Maintenance Team
The following is a composite illustration drawn from patterns common to service operations. It is not an account of a specific Techvaria client, and no figures below come from any real engagement.
Consider a facilities maintenance operation with around thirty engineers across several sites, a mix of planned maintenance visits and reactive callouts, and one coordinator who builds the weekly roster.
The Trigger
Two incidents in a month. An engineer arrived at a site that had been reassigned three days earlier and nobody had told him. And the operations manager declined a sizeable planned-maintenance contract because he could not establish whether the team had capacity, and later suspected it did.
What the Review Found
The roster was a workbook, published on Thursday for the following week, printed and emailed. Leave was approved in a separate HR process and transferred to the roster manually, which was where the reassignment failure originated, since the coordinator had updated the workbook but the printed copy was the one the engineer had.
Qualifications were entirely informal. The coordinator knew which engineers held which certifications and which sites required them. Nobody else did, which made her absence during one week each quarter genuinely disruptive.
Utilisation was unmeasured. The operations manager’s instinct that there was spare capacity turned out to be correct, but he had no way to evidence it at the time he needed to.
What Was Implemented
The Planning-versus-Field-Service question was settled first. Callouts to customer sites were a Field Service problem; knowing who was rostered, qualified and free was a Planning problem. They had both, and started with Planning because the capacity question was the one costing them work.
Leave was connected so the schedule read the same source rather than a transcription, which removed the specific failure that had prompted the review.
The role model was deliberately kept small: around a dozen roles reflecting the distinctions that actually changed an assignment. Certifications were captured only where a site or a task legally required them, with expiry dates, which surfaced two that were closer to expiry than anyone had realised.
The coordinator’s informal constraints took the longest to capture and produced the most value: three sites with access restrictions, one engineer who could not be sent to a particular client, and a pairing rule for a specific type of work.
Self-service visibility was given to engineers, with notification on change rather than only on publication.
After Two Quarters
The capacity question became answerable, and the operations manager took on work he would previously have declined. Overtime moved from reactive to largely planned, because gaps were visible far enough ahead to be covered within normal hours.
The outcome the coordinator mentioned was not on anyone’s business case: her week off was, for the first time, uneventful.
Industry Use Cases
- Facilities and maintenance services. The core case: dispersed teams, qualification requirements and a mix of planned and reactive work. See professional services.
- Manufacturing. Shift rostering across lines and cells, where the constraint is often a specific skill rather than headcount. See manufacturing solutions.
- Healthcare. Qualification and certification gating is strict and auditable, which makes expiry tracking a requirement rather than a convenience. See healthcare solutions.
- Logistics and transport. Driver scheduling with working-time constraints, where compliance evidence matters as much as the roster itself. See logistics solutions.
- Construction. Crews allocated across sites with site-specific access and ticket requirements. See construction.
- IT and professional services. Consultant allocation across engagements, where utilisation is the primary commercial measure. See IT services and ERP.
Frequently Asked Questions
Field Service dispatches jobs to customer sites: the unit of work is a customer visit with a location, task, parts and usually an invoice. Planning schedules internal people and resources against time: the unit is a shift or allocation, not necessarily tied to a customer. Many businesses need both; they answer different halves of the same day.
Whichever is costing more now. If visits go to the wrong engineer or job status is invisible, that is Field Service. If you cannot say who is working, who is free, or whether you have capacity for more work, that is Planning.
Only if leave and the schedule read the same source. A roster fed by manual transcription from a separate leave process will eventually assign someone who is away, and that single failure destroys confidence in the whole system. Settle this at design time.
Define roles at the level you actually schedule: if a distinction never changes an assignment decision, it is not a role. Ten to fifteen roles is workable for most operations. Forty will be out of date within a year because nobody maintains that much detail, and an out-of-date model is worse than a coarse one.
Compliance support depends on configuration and on your jurisdiction’s rules, and this article does not attempt to state those rules. Where working time is regulated, treat evidence requirements as a design input and confirm what the system can demonstrate with your partner and your own HR or legal advice.
Notify on change rather than only on publication, give people self-service visibility of their own schedule, and distinguish clearly between draft and published. The failure people remember is almost always a communication failure rather than a planning one.
Capability varies by Odoo version and edition. Confirm what is available in your instance with your partner before designing an approach around it.
Conclusion
The roster in one person’s head works, right up until that person is away or until somebody asks whether the team has capacity for more work.
What replaces it is not complicated. The schedule needs to know who exists and when they work, who is unavailable (from the same source as leave, not a transcription), what each person is qualified to do, and what work needs covering. The constraints the scheduler currently carries informally, which are the site restrictions, the pairings and the individual arrangements, are where most of the value sits and they are the part most often skipped.
Keep the role model small enough that someone will maintain it. Track certifications only where they gate work. Distinguish draft from published, notify on change rather than only on publication, and give people visibility of their own schedule, because the scheduling failure people remember is nearly always a communication failure.
Then measure utilisation in both directions. Operations manage gaps attentively and ignore idle capacity, and the second one costs exactly the same as the first while never prompting a phone call.
And settle the Planning-versus-Field-Service question before anything else. They solve genuinely different problems, and implementing the wrong one is an expensive way to discover which you had.
Find Out Where Your Capacity Is Going
Techvaria is an official Odoo Silver Partner with more than 350 implementations delivered since 2016, from offices in Bangalore, Ahmedabad and Dubai.
A resource planning review establishes which problem you actually have (rostering or dispatch, or both), then produces a role model sized to the decisions you really make, the informal constraints your scheduler carries written down, a leave connection design so the roster cannot assign someone who is away, and a utilisation view covering idle capacity as well as gaps.
If you have recently turned down work because you could not evidence capacity, that is the fastest way to scope this.
Talk to our Odoo consultants, or see how we approach implementation.
Find Out Where Your Capacity Is Going

Mustufa Rahi is an Odoo Certified Functional Consultant and ERP expert at Techvaria with 15+ years of experience in implementation, automation, and business process optimization, helping organizations scale efficiently.
