The transformation

From spreadsheets to systems that think.

Few businesses reach a connected operation in a single move. These are the stages we take them through — where we start depends on where you are today, and you do not have to walk them all at once.

  1. 01

    Manual & disconnected

    Spreadsheets, email approvals and re-keyed data. Every department has numbers, and no two sets agree.

  2. 02

    ERP foundation

    Dynamics 365 Finance, procurement and Supply Chain establish one ledger, one agreed set of supplier terms and one stock figure for the whole business.

  3. 03

    Connected business systems

    Sales, Commerce, Marketing, HR & Payroll and the delivery arm join that core instead of standing beside it, so a customer is one customer everywhere.

  4. 04

    Automated processes

    Power Platform takes over the approvals, reconciliations and hand-offs that used to move by hand.

  5. 05

    Intelligence

    Fabric and OneLake put every workload on one copy of the data; Power BI and Copilot Studio turn it into forecasts, recommendations and answers.

Stages 02 and 03 run on Microsoft Dynamics 365; stages 04 and 05 add Power Platform and the Fabric data platform on top. Here is how all eleven modules played out for one business.

The business in question

One company, followed the whole way through. A regional distributor and retailer: six branches, three warehouses, selling through its own shops, online and to trade — and installing and servicing what it sells. Every figure on this page belongs to this one business.

What it has

  • Six branches, each reporting in its own format
  • Three warehouses, counted by hand
  • Retail, online and wholesale, priced separately
  • Around 340 staff under three sets of payroll rules
  • An installation and service arm out on the road

Where it was stuck

  • Three ledgers that disagreed every month-end
  • Stock figures a day old before anyone read them
  • A point-of-sale island, re-keyed each morning
  • Buying from memory, then arguing invoices line by line
  • Pipeline in a spreadsheet, payroll a monthly ritual
  • Installation jobs scheduled on a whiteboard

What it wanted

  • To open new markets without growing the back office
  • One set of numbers everyone trusts
  • The same experience for a customer on any channel

These three are what the closing section is measured against — not the software.

6 branches 3 warehouses 3 sales channels 340 staff 11 modules, in order
A worked example

One business, one module at a time.

Eleven modules, in the order this business took them. Jump to any of them from the rail on the left, and watch the platform beside it assemble itself as each one goes live.

An illustrative composite drawn from typical implementations, not a specific client.

  1. STEP 01

    Dynamics 365 Finance

    The ledger

    BeforeMonth-end takes eleven days. Three spreadsheets disagree on what is owed, and every branch reports in its own format.
    AfterOne chart of accounts and one ledger, with multi-currency and local tax handled inside the system.

    Close drops from weeks to days — and the numbers stop being negotiable between departments.

    Month-end close
    11 days 3 days
    −73%
    Journals keyed by hand
    640/mo 95/mo
    −85%
    Ledgers reconciled
    3 1
    Branches on one chart of accounts
    0 of 6 6 of 6

    What we implemented

    • One chart of accounts across six branches
    • Multi-currency with daily revaluation
    • Local tax and e-invoicing rules
    • Opening balances migrated and tied out

    Connects to

    Receives posted entries from Purchasing, Commerce and HR & Payroll; publishes the closed period to Fabric & OneLake.

    What to watch out for

    The risk. Everything downstream inherits the chart of accounts, so a rushed design is expensive to unpick later.

    How we handled it. We froze the chart with finance and audit before a single transaction was migrated, and ran two parallel closes against the old ledgers before switching.

  2. STEP 02

    Procurement

    What the business buys

    BeforeBuying happens from memory and relationships. Nobody can say what was agreed with a supplier, so invoices are argued line by line.
    AfterRequisition, approval, purchase order and receipt run in one flow, matched against the contract before anything is paid.

    You buy on the terms you negotiated, and you can prove it when the invoice arrives.

    Purchase order cycle
    6 days 1.5 days
    −75%
    Supplier on-time delivery
    71% 93%
    Off-contract spend
    34% 8%
    −26pt
    Three-way match automated
    0% 90%

    What we implemented

    • Requisition and approval thresholds by branch
    • Supplier catalogues and contract pricing
    • Three-way match: order, receipt, invoice
    • Supplier scorecards on delivery and quality

    Connects to

    Raises demand from Supply Chain, posts commitments and invoices to Finance, and routes approvals through Power Platform.

    What to watch out for

    The risk. Tight matching rules stop invoices dead when the receipt is late, and Accounts Payable feels it first.

    How we handled it. We set tolerance bands per category and gave AP a queue with a reason code on every held invoice, rather than a silent block.

  3. STEP 03

    Dynamics 365 Supply Chain

    Stock across three warehouses

    BeforeStock counts live in a spreadsheet that is a day old. Purchasing reorders from memory, and slow movers quietly tie up cash.
    AfterReal-time inventory across all three warehouses, with reorder points and replenishment driven by actual demand.

    Less capital sitting on shelves, and fewer stockouts on the lines that actually sell.

    Stock accuracy
    78% 98%
    +20pt
    Inventory holding
    index 100 index 82
    −18%
    Stockouts on top lines
    14/mo 3/mo
    −79%
    Reorder decisions automated
    0% 85%

    What we implemented

    • Three warehouses on one item master
    • Reorder points from demand history
    • Cycle counting replacing the annual stocktake
    • Batch and serial traceability

    Connects to

    Signals demand to Purchasing, commits stock for Commerce and Project Ops, and values inventory into Finance.

    What to watch out for

    The risk. Reorder points calculated on bad history simply automate the old mistakes.

    How we handled it. We ran the calculated points in advisory mode for two cycles and let planners overrule them, then promoted only the ones that held up.

  4. STEP 04

    Dynamics 365 Sales

    The pipeline

    BeforeThe pipeline lives in a spreadsheet and in people's heads. Quotes are built in Word, and the forecast is a number the sales manager feels.
    AfterAccounts, opportunities and quotes in one place, with pricing pulled from the same catalogue the shop sells from.

    You can see what is really going to close, early enough to do something about the gap.

    Quote turnaround
    4 days same day
    Forecast accuracy
    ±40% ±12%
    Win rate on qualified deals
    22% 31%
    +9pt
    Deals with no next step
    46% 9%
    −37pt

    What we implemented

    • Stage definitions with exit criteria
    • Quotes priced from the Commerce catalogue
    • Forecast rollup by branch and product line
    • Guided next-best-action on every open deal

    Connects to

    Prices from Commerce, checks availability in Supply Chain, hands won deals to Project Ops, and feeds the forecast to Power BI.

    What to watch out for

    The risk. A pipeline nobody updates is worse than a spreadsheet, because it looks authoritative.

    How we handled it. Stage exit criteria are enforced in the form, and the weekly review is run from the live board rather than an exported copy.

  5. STEP 05

    Dynamics 365 Commerce

    The retail arm

    BeforeThe shop's POS is an island. Online orders are printed and re-keyed by hand every morning.
    AfterRetail, online and wholesale all draw on the same catalogue, pricing and stock figure.

    One customer and one price list, whichever channel they buy through.

    Orders re-keyed
    40/day 0
    −100%
    Online order to fulfilment
    26 hrs 3 hrs
    −88%
    Price lists
    3 1
    Channels on one catalogue
    0 of 3 3 of 3

    What we implemented

    • POS on the same product and pricing engine
    • Online storefront on shared stock
    • Wholesale price lists and customer terms
    • Click-and-collect across the three warehouses

    Connects to

    Draws stock from Supply Chain, prices Sales quotes, posts takings to Finance, and sends behaviour to Marketing.

    What to watch out for

    The risk. One catalogue means one mistake reaches every channel at once.

    How we handled it. Price and product changes go through a staged publish with a preview against each channel, rather than straight to live.

  6. STEP 06

    Dynamics 365 Customer Insights

    Demand and the customer record

    BeforeCampaigns go out as one list to everyone. Nobody can say which of them produced a single riyal of revenue.
    AfterOne customer profile built from retail, online and wholesale behaviour, with segments that maintain themselves.

    Marketing spend is aimed at people likely to buy, and the revenue it produced can be traced back.

    Revenue traced to campaign
    0% 64%
    Segments maintained
    3 static 28 dynamic
    Cost per qualified lead
    index 100 index 58
    −42%
    Campaign build
    9 days 2 days
    −78%

    What we implemented

    • Unified customer profile across the three channels
    • Dynamic segments that refresh themselves
    • Journeys triggered by real purchase behaviour
    • Attribution back to won revenue

    Connects to

    Reads behaviour from Commerce, hands qualified leads to Sales, and measures itself against won revenue in Power BI.

    What to watch out for

    The risk. Unifying customer records across channels surfaces every duplicate and consent gap at once.

    How we handled it. Matching ran in review mode first, and consent was re-captured per channel before any journey was switched on.

  7. STEP 07

    HR & Payroll

    The people system

    BeforePayroll is a monthly spreadsheet ritual. Leave balances and end-of-service are worked out by hand, differently each time.
    AfterGCC and Middle East payroll rules, leave and end-of-service run in the system and post straight to Finance.

    Payroll reconciles itself against the ledger, and compliance survives the next regulatory change.

    Payroll run
    3 days 4 hours
    −83%
    Corrections after run
    31/cycle 4/cycle
    −87%
    Payroll rule sets handled
    1 3
    End-of-service
    by hand in system

    What we implemented

    • Three GCC payroll rule sets in one engine
    • Leave accrual and end-of-service automated
    • Self-service for 340 staff
    • Journal posted straight to the ledger

    Connects to

    Posts labour cost to Finance, supplies technician availability to Project Ops, and its journal is reconciled in Power BI.

    What to watch out for

    The risk. Payroll is the one system where a rounding error becomes a trust problem the same afternoon.

    How we handled it. Three parallel runs against the old spreadsheets, reconciled to the fils, before anyone was paid from the new system.

  8. STEP 08

    Project Operations & Field Service

    Delivering the work

    BeforeInstallation jobs are scheduled on a whiteboard. Nobody knows what a job really cost until it is finished and invoiced.
    AfterJobs, technicians, parts and time are planned and captured against the project as work happens.

    You can see margin per job while there is still time to do something about it.

    On-time delivery
    82% 96%
    +14pt
    Technician utilisation
    61% 78%
    +17pt
    Quote to invoice
    21 days 6 days
    −71%
    Job margin
    visible at close live

    What we implemented

    • Scheduling board across branches and skills
    • Mobile time and parts capture on site
    • Cost and revenue accrued per project
    • Warranty and service contracts tracked

    Connects to

    Takes won work from Sales, consumes parts from Supply Chain, draws technicians from HR, and accrues to Finance.

    What to watch out for

    The risk. Field capture only works if it works with one hand, on site, on bad signal.

    How we handled it. The mobile flow was cut to four taps and made offline-first, with technicians in the design from the first week.

  9. STEP 09

    Microsoft Power Platform

    The work between the modules

    BeforeApprovals move by email and stall in inboxes. The same figures are copied between systems every week.
    AfterApprovals, reconciliations and hand-offs run as workflows, with low-code apps filling the gaps between modules.

    The repetitive work between systems stops consuming your team's week.

    Manual approvals
    1,200/mo 90/mo
    −93%
    Approval waiting time
    3.5 days 4 hours
    −95%
    Hours returned to the team
    310/mo
    Hand-offs automated
    0 6

    What we implemented

    • Approval flows with delegation and escalation
    • Five apps closing gaps between modules
    • Scheduled reconciliations replacing weekly copying
    • Every run leaving an audit trail

    Connects to

    Sits between every module — carrying approvals for Purchasing, hand-offs from Sales to Project Ops, and exceptions into Power BI.

    What to watch out for

    The risk. Flows built by whoever needed them become an estate nobody owns.

    How we handled it. Everything runs in managed solutions with named owners and a monthly review of failures and orphans.

  10. STEP 10

    Microsoft Fabric & OneLake

    One copy of the data

    BeforeFive systems each keep their own copy. Fourteen extract jobs run overnight, and by morning the copies already disagree.
    AfterOne copy in OneLake that every workload reads — engineering, warehousing and real-time intelligence over the same rows.

    Arguments about whose number is right stop, because there is only one number.

    Copies of the same data
    5 1
    −80%
    Overnight extract jobs
    14 0
    −100%
    Data freshness
    monthly near real time
    Conflicting measure definitions
    12 0

    What we implemented

    • OneLake as the single store for every workload
    • Pipelines replacing 14 overnight extracts
    • One governed semantic model
    • Lineage and sensitivity labels end to end

    Connects to

    Reads from every operational module and serves Power BI and Copilot Studio — it is the floor the answers stand on.

    What to watch out for

    The risk. Centralising the data makes every existing quality problem everyone's problem at once.

    How we handled it. Quality rules run at ingestion with an owner per domain, so a bad feed is quarantined and named rather than quietly averaged in.

  11. STEP 11

    Power BI & Copilot Studio

    Answers, not reports

    BeforeReporting is a monthly PDF that is already out of date the day it lands. Six people can build a report; everyone else queues.
    AfterLive dashboards over the one governed model, and a Copilot Studio agent answering questions inside the apps people already use.

    Decisions get made on today's numbers, by the people who own them.

    Report prep
    26 hrs/mo 0
    −100%
    People self-serving answers
    6 140
    Time to answer a new question
    5 days minutes
    Questions handled by an agent
    0 400/mo

    What we implemented

    • Dashboards per role, not per department
    • Row-level security following the org
    • A Copilot Studio agent grounded on the governed model
    • Answers surfaced inside Teams and the apps

    Connects to

    Stands entirely on Fabric & OneLake, and closes the loop by feeding decisions back into Sales, Supply Chain and Marketing.

    What to watch out for

    The risk. An agent that sounds confident and is wrong destroys trust faster than no agent at all.

    How we handled it. It answers only from the governed model, shows the measure behind every figure, and says it does not know rather than guessing.

What eleven modules added up to.

Each module is worth something on its own. What matters is what they are worth together — and whether the business got the three things it asked for at the start.

More effective

The work itself got better, because the people doing it can see what is happening while they can still change it.

  • On-time delivery 82% → 96%
  • Job margin visible live, not at close
  • Forecast accuracy ±40% → ±12%
  • Supplier on-time 71% → 93%

More efficient

The same output for far less effort — and the effort that came back went into work that pays.

  • Month-end close 73% faster
  • 1,110 fewer approvals a month
  • 310 hours returned each month
  • 14 overnight extracts gone

More valuable

Capital came off the shelves, spend came under contract, and the company became straightforward to audit and to finance.

  • Inventory holding −18%
  • Off-contract spend 34% → 8%
  • Branches on one chart of accounts 6 of 6
  • Revenue traced to campaign 0% → 64%

Better run

The operating rhythm changed. Questions get answered by the person who has them, on the day they have them.

  • People self-serving answers 6 → 140
  • Time to a new answer 5 days → minutes
  • Copies of the data 5 → 1
  • Data freshness near real time

Open new markets without growing the back office

The seventh branch inherits a chart of accounts, payroll rules, a stock model and a price list that already run. Opening one is now a configuration task, not another finance hire and another spreadsheet.

One set of numbers everyone trusts

Five copies of the data became one in OneLake, and twelve conflicting definitions of the same measure became none. The number in the board pack is the number in the dashboard, because there is only one of them.

The same experience on any channel

One catalogue and one price list behind retail, online and wholesale. A customer gets the same price and the same stock answer whichever door they come through, and the business can see them as one customer.

A business that used to review itself monthly, in arrears, now reviews itself continuously — which is the difference between reporting the past and running the company.

Ready to start your transformation?

Talk to Our ERP Experts
Back to home