Home · Solutions · Management & planning

Solution · Management & planning

Sixty cost centres, one template, one version, and a forecast that lands on the day it is due

Budget and forecast collection without version chaos

Templates are generated per cost centre with actuals already in them, owners submit in the browser, every submission is validated on the spot, and consolidation happens by itself.

DepartmentalMicrosoft TeamsHuman in the loopDeterministic automation
72analyst-days a year go into generating, chasing, checking and consolidating budget and forecast templates in this illustrative company.

Executive summary

Challenge

Stop spending a quarter of an analyst's year collecting spreadsheets from cost-centre owners.

What changes

The design starts from one governed template instead of sixty copies.

Business value

The cycle finishes on the date it was planned to finish, because chasing and checking stop being manual steps at the end.

Systems involved

SharePoint planning site and archive; the consolidation model in Excel; Power BI semantic model

Business problem

Planning & FP&A

Every company that budgets runs the same contract between finance and the line: the line knows what it intends to spend, finance knows what shape the number must take. Today that contract is an email attachment, and the moment it leaves the analyst's outbox it becomes sixty documents that drift apart.

The drift is not carelessness. A warehouse manager inserts two rows because there is no line for agency staff. A regional director forwards the file to a colleague who returns it with values pasted over the formulas. Someone plans in local currency because the rate cell was blank. Each is reasonable for a busy manager, and each costs an analyst a call, a corrected cell and one more version.

Three groups feel it. The two FP&A analysts organise their quarter around collection rather than analysis. The sixty owners experience planning as an administrative penalty with a deadline and no context. The CFO presents a forecast assembled in the last forty-eight hours and fields questions whose answer sits in somebody's mailbox.

Scale makes it worse in a particular way. Cost centres multiply faster than analysts, a reorganisation renumbers half of them mid-cycle, and the consolidation workbook grows another tab and another set of links until nobody will touch a formula.

How it works today

Sixty workbooks leave the finance team and sixty sets of numbers come back: that is the cycle in most companies that plan in Excel.

  1. PersonAn analyst copies last cycle's workbook, renames the tabs, pastes fresh actuals from an SAP export and saves sixty files to a drive
  2. PersonSixty emails go out with the file attached and the deadline in the subject line
  3. WaitingReturns arrive over two to three weeks: some as attachments, some as links, a few scanned
  4. PersonThe analyst checks each file: totals footing, headcount against the HR list, currency lines at the planning rate
  5. Risk of errorInserted rows, overwritten formulas and pasted values survive undetected until consolidation, sometimes until the board pack
  6. WaitingNumbers are pasted into a workbook with a tab per cost centre; whatever does not tie starts an email and a second and third correction round
  7. PersonVariance commentary is requested separately, arrives as prose in email and is retyped into the board pack the night before
PersonWaitingRisk of error

Why the current process costs more than it appears

The cost grows where nobody is looking.

  • Re-keying is the cheapest part of the bill. Reconstructing what an owner meant, finding which of three files is current and repairing a broken link cost more analyst time than typing ever did, and none of it is recorded.
  • Late information decays. A forecast reaching the board two weeks after the cut-off describes a quarter already under way, so hiring, stock and capital decisions get made on the previous view.
  • Owners learn what the exercise is worth. A template that arrives without their own actuals invites last year plus a percentage, a plan nobody defends and everybody revises.
  • Continuity depends on one workbook and the person who understands it. An audit request for last year's approved version turns into archaeology through mailboxes and undocumented links.

Cost of inaction

One planning year of collection at today's pace≈ €20,160
The same effort carried across a three-year plan horizon≈ €60,480
One planning year with the sixty owners' hours counted in≈ €45,360

FP&A absorbs this six times a year, and the six does not become five. What continues unchanged is the calendar: every quarter the forecast lands after the decisions it was meant to inform, and every budget round ends with a week of reconciliation nobody planned. Each reorganisation renumbers cost centres and the sixty files are rebuilt by hand, so the process slows exactly when the business changes fastest.

The quieter risk is that the plan cannot be defended. When an auditor, a lender or a new CFO asks what a cost centre's approved budget was in March and who approved it, the answer has to be assembled from mailboxes and a workbook whose links nobody wants to open.

Illustrative scenario

A plausible organisation with realistic proportions. The figures are there to be recalculated on your data; they are not a client result.

Organisation

A distributor of hospital supplies and medical consumables: four regional warehouses, a tender office, a field service unit, about 900 employees. SAP holds the ledger, Microsoft 365 E3 is in place, planning runs in Excel. Sixty cost centres, each with a named owner.

Volume

Six collection cycles a year, the annual budget in two rounds and one forecast round each quarter: 360 submissions with roughly forty planning lines each. About half need a correction round.

Current process

Three people in FP&A build the workbooks from last cycle's file, email them out, chase what is missing and paste accepted numbers into a consolidation workbook with sixty tabs.

Bottleneck

Around twelve analyst-days per cycle go into generation, chasing, checking, corrections and consolidation. The forecast reaches the board roughly two weeks after the cut-off, when the first month of the quarter is nearly gone.

Solution

Robots generate one workbook per cost centre with actuals, headcount and planning rates in it. Owners fill it in the browser in their own folder, every submission is validated as it is sent, accepted versions are stamped and archived, and reminders, status and commentary sit in Microsoft Teams.

Potential outcome

In the modelled case collection effort turns into review effort, the consolidated forecast is ready within days of the cut-off instead of weeks, and every line carries a submitter, a version and the sentence that explains it. The figures are a model, not a measurement.

Proposed solution

The design starts from one governed template instead of sixty copies. Finance owns a single master workbook: the line structure, the input ranges, the validation lists, the commentary fields and the locked areas. At the start of each cycle a robot reads the cost-centre master data, pulls actuals by cost centre and account from SAP with the planning rates and the headcount list, and builds one workbook per cost centre, so the owner opens twelve months of their own actuals, run rate and approved budget rather than an empty grid.

The file never travels. It sits in the owner's folder on the SharePoint planning site, with permissions from the Entra ID group that follows the cost-centre master data, and is edited in Excel Online. On Submit, a flow hands the workbook to the rules: totals foot, headcount ties to the HR list, currency lines use the planning rate, the structure is unaltered, and every line past the agreed threshold carries a written comment. Failures return within minutes as a card in Microsoft Teams naming the cell and the rule, so the person who can fix the number sees it.

What passes is stamped with cycle and version, archived read-only and appended to the consolidation model behind the Power BI management report. Reminders, escalation, live status for all sixty cost centres and the controller's sign-off happen in Teams. This is not a planning application: no modelling engine, no allocation ledger. It is the collection, validation and version layer around the model you already run, and if you later buy a planning tool, the rules, the master data mapping and the commentary trail move with you.

Native capabilities used

SharePoint document libraries with metadata, permissions and version history; Excel Online and Office Scripts run from Power Automate; Microsoft Lists as the submission register; UiPath Orchestrator time triggers, queues and audit logs; UiPath Integration Service connectors for Microsoft OneDrive & SharePoint and Microsoft Teams; Adaptive Cards through the Workflows app in Teams; Microsoft Teams Approvals app; Power BI semantic model

What we build

The governed master template and generation routine, the actuals and headcount pre-fill, the validation rule set and its messages, the submission register and status view, reminder and escalation logic, the versioned archive, the consolidation routine, and the variance commentary workflow

Custom integration

SAP actuals, cost-centre master data and planning rates through UiPath SAP connectors (BAPI/OData), or from your existing reporting layer; the HR headcount extract where the payroll system offers no interface

How the automated process works

  1. AutomationOn the cycle start date an Orchestrator trigger pulls actuals, master data, headcount and planning rates, and reconciles the cost-centre list with the owner list
  2. AutomationA robot generates one workbook per cost centre from the master, writes the pre-fill, locks the structure and files it in the owner's folder
  3. SystemEach owner receives an Adaptive Card in Teams with the deadline and a direct link; the workbook opens in Excel Online, with no attachment
  4. PersonThe owner plans the lines, writes the commentary the rules require and presses Submit
  5. AutomationThe validation run checks totals, headcount, rates, structure and commentary, and returns any failure as a card naming the cell and the rule
  6. AutomationAccepted submissions are stamped with cycle and version, archived read-only and appended to the consolidation model; the Power BI page refreshes
  7. PersonThe controller signs off per cost centre in the Approvals app, and the analyst spends the remaining days challenging outliers
AutomationSystemPerson

Human-in-the-loop model

Automation handles

  • Generation of the sixty workbooks, the pre-fill of actuals, headcount and rates, and the structural locks
  • Reminders, escalation and live submission status per cost centre
  • Validation of totals, headcount, rates, structure and commentary at the moment of submission
  • Versioning, the read-only archive and the append into the consolidation model

People decide

  • Every number in the plan: owners plan, the automation only checks completeness and consistency
  • Whether a late or deviant submission is accepted, and on what condition
  • The validation rules, commentary thresholds and approval path, which stay owned by the controller
  • Which variances matter and what the company does about them

Before and after

BeforeAfter
Analyst-days per collection cycle12modelled 3 to 4, spent on review
Working days from cut-off to consolidated forecast10 to 15modelled 3 to 5
Submissions accepted without a correction roundabout halfmodelled 85 to 90%
Versions in circulation per cost centreas many as there are mailboxesone, with its history
Variance commentaryrequested by email after consolidationwritten at submission, attached to the line

Systems and integrations

The stack is deliberately short: one engine, one execution layer, one place where a person decides.

Inputs

  • SAP actuals by cost centre and account
  • cost-centre and owner master data
  • planning rate table
  • HR headcount list
  • the previous accepted plan version

Automation layer

  • UiPath Orchestrator
  • UiPath Robots
  • UiPath Integration Service
  • Power Automate
  • Office Scripts

Target systems

  • SharePoint planning site and archive
  • the consolidation model in Excel
  • Power BI semantic model

Human touchpoints: Excel Online in the browser; Adaptive Cards and the status tab in Microsoft Teams; Approvals app for sign-off

SAP actuals by cost centreUiPath OrchestratorUiPath RobotsSharePoint planning siteExcel Online in the browser

Technologies used

Microsoft SharePoint

a folder per cost centre, submission metadata, version history and the read-only archive

A
Microsoft Excel Online and Office Scripts

the governed template with locked structure; scripts write the pre-fill and read submissions back

A
UiPath Robots and UiPath Orchestrator

scheduled pulls, generation, validation and consolidation runs, retries and audit

A
UiPath Integration Service (Microsoft OneDrive & SharePoint, Microsoft Teams connectors)

files, Excel ranges, list items and channel messages without bespoke code

A
Microsoft Teams (Workflows app, Adaptive Cards, Approvals app)

reminders, validation feedback, the live status tab, controller sign-off

A
Power Automate

the Submit trigger, Office Scripts runs, reminder and escalation timers

A
SAP (BAPI and OData through UiPath SAP connectors)

actuals by cost centre and account, master data, planning rates

A
Power BI

collection status during the cycle, consolidated plan against actuals after it

A
Averified product capability (vendor documentation)

Illustrative economic model

Numbers you can check against your own data.

Illustrative model
6 analyst-days a month × 480 minutes= 48 h / month
48 h × €35 fully loaded hourly cost= €1,680 / month
× 12 months≈ €20,160 / year
Annual FP&A capacity released (illustrative)≈ €20,160

The calendar drives every number here, and the calendar belongs to an illustrative company, not to a measured client. This company runs six collection cycles a year, the annual budget in two rounds plus one forecast round per quarter, and each absorbs about twelve analyst-days of generation, chasing, checking, corrections and consolidation. Six cycles of twelve days is 72 analyst-days a year, or six analyst-days a month averaged over the year, an analyst-day being 480 minutes. €35 is a fully loaded hourly cost for an FP&A analyst in Central Europe. The sixty owners spend their own time too, about two hours each per cycle: 60 by 6 by 2 is 720 hours a year, roughly €25,200, kept outside the calculator and shown separately below. We model released capacity, not headcount reduction.

Run the numbers on your data

hours released per month
of annual capacity released

An illustrative estimate from your own inputs. It models released capacity; it is not a promise of savings.

Business benefits

  • The cycle finishes on the date it was planned to finish, because chasing and checking stop being manual steps at the end
  • Owners plan against their own actuals rather than a blank grid, which lifts the quality of the first submission
  • Errors are caught at submission by the person who can correct them, not three weeks later by an analyst who must interpret them
  • One version of each cost centre's plan exists at any moment, with its history, so the question of which file is current disappears
  • Variance commentary is written next to the number it explains and travels with it into the board pack

The management view

  • Submission status is visible while the cycle runs, per cost centre and owner, instead of being reconstructed after it closes
  • Every figure in the board pack traces to a named submission, a version and the person who approved it
  • Planning capacity scales with the number of cost centres, not the number of analysts

Board-level KPIs

working days from cut-off to consolidated forecaston-time submission ratesubmissions accepted at first passcorrection rounds per cycleanalyst-days per cycle

Security and governance

Trust in automation is built on the audit trail, not on a promise.

  • Owners see their own cost centre and nothing else: access follows Entra ID groups built from the cost-centre master data, so a reorganisation changes group membership, not a hand-kept permission list.
  • Where personnel costs are planned, that block carries its own permissions and a Microsoft Purview sensitivity label; who may see salary detail is decided once, not file by file.
  • The robot reads SAP through a dedicated display-only account and writes nothing back; its credentials stay in the platform's credential store and no analyst handles them.
  • Every accepted submission is kept as a read-only snapshot with submitter, timestamp, rule set version and sign-off, so an approved budget can be reconstructed years later.
  • Robot execution, queues and logs run on your UiPath Automation Cloud tenant, EU region; plan workbooks stay in your Microsoft 365 tenant.

Why now

01

The planning calendar is fixed and the decisions attached to it are not waiting. A forecast two weeks after the cut-off reports the past; three days after it, the same forecast is an input to a hiring, stock or capital decision still open.

02

The finance team you have is the team you will have. The modelled 72 analyst-days a year, €20,160, buys review and challenge instead of collection, without recruiting into a scarce controller market.

03

The collection layer no longer needs a bought tool or a developer: Excel Online with Office Scripts, SharePoint permissions and Teams cards cover generation, validation and status, and robots cover the SAP side.

Relevant executive roles

CFO

The forecast arrives on the date the board expects it, and every number traces to a person, a version and a comment

Head of Controlling

The team's cycle becomes review and challenge rather than file collection, and the rules stop being renegotiated owner by owner

CIO

Planning data leaves the network drive and the attachments, and access follows master data instead of a hand-kept list

Common questions and objections

We are buying a planning tool next year, so why do this now?

A planning tool assumes clean cost-centre master data, agreed validation rules and a disciplined submission process. Building those now shortens that implementation instead of competing with it, and the collection layer retires the day the tool takes over.

Our owners will not work in a browser.

They open one link to their own workbook, in the same Excel they know, with their own actuals in it. Where a division insists on the desktop application, it opens the same file in the same place; what disappears is the attachment and the second copy.

Sixty templates are not that much work.

Generating them is not the work. The chasing, the checking, the third correction round and the reconstruction of what someone meant are, and that is where the twelve analyst-days per cycle go.

When this is not the right solution

  • Fewer than roughly fifteen cost centres, where a controller who knows every owner runs the cycle by phone faster than any workflow
  • Cost-centre structure or the chart of accounts changes mid-cycle, in which case master data comes first and collection second
  • A planning application with owner input, validation and workflow is already in place and used, where the sensible work is integration

A question for the next management meeting

From our quarterly cut-off to the moment the board sees a consolidated forecast: how many working days is that, and how many of them go into collecting files rather than deciding anything?

Implementation approach

A scope without ambiguity, before anything is signed.

We deliver

  • The governed master template with locked structure, named input ranges, validation lists and pre-fill layout
  • The actuals, master data, headcount and rate pull, scheduled per cycle in Orchestrator
  • The validation rule set and the message each owner sees, agreed line by line with the controller
  • The SharePoint planning site, permissions per cost centre from Entra ID groups, and the versioned archive
  • The Teams layer and the Power BI status page: invitations, reminders, escalation, sign-off and live status

We need from you

  • The current template and two completed cycles, including the files that failed
  • The cost-centre list with owners, deputies and the approval path per division
  • A read account for SAP actuals and master data, and a service account for the planning site
  • A named owner in FP&A who can decide the rules and the commentary thresholds

Stages

Discovery

Template, rules, owners, deadlines and the exceptions of the last two cycles

Design

Master template, validation rules, permission model, reminder and escalation logic

Build

Generation and pre-fill robots, submission and validation flows, consolidation and archive

Dry run

A full cycle replayed on historical data, then one pilot division live

Go-live

The next forecast round under supervision, then the budget cycle with hypercare

Departmental. Effort follows the number of cost centres and template variants, how far the validation rules already exist in writing, and whether actuals come from SAP directly or from a reporting layer.