Case studies / Hospitality P&L agent
Hospitality
How a UK hospitality group standardised property P&Ls with a Copilot Studio agent
The finance team knew exactly what the standard layout should be. The hours went on getting every property's spreadsheet into it. That reformatting job now belongs to an agent.
Reformatting that took hours each month now takes seconds per P&L.
One agreed standard layout, applied the same way every month, whoever runs it.
Built in Copilot Studio and grounded in the group's own Microsoft 365 tenant, so figures never leave the boundary the group already governs.
The finance team still reviews and consolidates. The agent only reformats.
Case study
Situation
The group's properties each produced a monthly P&L, and each one arrived in its own shape. Different line names for the same costs, different groupings, different orders, all set by each property's systems and habits.
Before the central finance team could consolidate or compare anything, every P&L had to be reworked into the group's standard layout. That reformatting was done by hand, every month, and it absorbed hours the team should have spent on what the numbers actually said.
Case study
Challenge
The job was exact and repetitive with no judgement in it, which is precisely what made it a poor use of a finance team and a strong candidate for an agent. It also drifted: two people mapping the same property line differently is how comparability quietly breaks.
The figures themselves were the constraint. Property P&Ls are sensitive, and the group was clear that no financial data would be pasted into personal AI tools. Whatever did the reformatting had to run inside the Microsoft 365 environment the group already governed.
Case study
What the group needed to avoid
This was a formatting problem, and the fix had to stay the size of the problem.
- Financial data leaving the tenant for a personal or consumer AI tool.
- A brittle macro or template that broke each time a property relabelled a line.
- A finance-system replacement project to fix what was one repeated task.
- Output the team could not check line by line against the original.
Case study
What changed
The group built a Copilot Studio agent with one job: take a property P&L in whatever layout it arrives and return it in the group's standard layout. The standard layout and the line mapping were agreed once and written into the agent's instructions, so the specification lives in plain English rather than in one person's head.
The agent runs inside the group's own Microsoft 365 tenant, grounded in their environment, so the figures never leave the boundary the group already governs. A reformat that took hours now takes seconds, and the finance team reviews the output against the original before the numbers go anywhere near consolidation.
Case study
Why this is a Copilot Studio story
Nothing about this build needed a development team or a data project. Copilot Studio, inside the licensing the group already had, was enough to take a defined job off the finance team's month.
That is the point worth generalising. Building with advanced tools is no longer reserved for large firms with technical staff. A smaller group with a clearly described job, an agreed output and a review step can have an agent doing real finance work within weeks.
Case study
Outcome
Month end starts with every property's P&L already in the standard layout. The hours that went on reformatting have moved to review and analysis, and the layout no longer depends on who did the mapping.
The review step stayed. The agent reformats and the team still owns the numbers. That split is what made a seconds-fast version something the finance team could trust.
What the agent does
Whatever shape a P&L arrives in, one shape comes out.
Any input layout
Each property's P&L arrives shaped by its own system. The agent maps it to the standard without a template per property.
The group standard, every time
One agreed layout, applied identically every month, whoever on the team runs it.
Line mapping
Property line items are mapped to the group's standard lines using the agreed mapping, not one person's memory.
Seconds per property
A reformat that used to take a chunk of the afternoon now takes seconds per P&L.
Review before consolidation
The team checks the standardised output against the original before the figures go anywhere.
Inside the tenant
Figures stay in the group's governed Microsoft 365 environment from arrival through to consolidation.
Where the build goes next
Standardised numbers make the next agents easier.
Once every property reports in the same shape, commentary drafts, pack assembly and questions over the figures all become well-defined jobs with the same review discipline.
Variance commentary drafts
With every P&L in the same shape, first-draft commentary on the movements becomes a well-defined second job.
Month-end pack assembly
Standardised P&Ls feed a consolidated pack that assembles itself for review rather than being built by hand.
Questions over the numbers
An agent grounded in the standardised figures can show which property lines feed a group line and how the month compares.
Other reformatting jobs
Every finance team has more than one document that arrives in the wrong shape. The pattern transfers.
What the firm learned
Why a boring, exact job is the right first agent.
- The best first agent is a job the team already does exactly, repeatedly and without judgement.
- Grounding in the group's own tenant answered the security question before it was asked.
- A defined output format is what makes an agent checkable. The standard layout was the specification.
- Copilot Studio put this within reach of a group with no development team.
- Keeping the review step is what let the finance team trust the seconds-fast version.
Sensible next moves
Build on the standard, not around it.
- Next Draft variance commentary from the standardised P&Ls, reviewed before it goes in the pack.
- Next Assemble the month-end consolidation from the standardised outputs.
- Next List the other reformatting jobs in the finance calendar and score them the same way.
- Next Agree who owns the standard layout so the agent's specification stays current.
Related routes
Where this example connects to FiveForward services.
Copilot Studio agent builds
The build service behind this kind of work: scoping, grounding, testing, rollout and handover in your own tenant.
Copilot for finance teams
Copilot habits for reporting commentary, variance narratives and month-end admin, tied to the reporting cycle.
Microsoft Copilot agents
Plain-English guidance on what agents are, where they fit and how to govern them.
Which Copilot model should your finance team use?
Where the model picker matters for finance work, and why the P&L should stay in the tenant.
Next step
Give the reformatting to an agent and keep month end for the numbers.
Talk through the documents your finance team reshapes by hand every month, and whether a Copilot Studio agent in your own tenant would fit the job.