AWRA vs Spreadsheets: The True Cost of Manual Operations
Excel is free the way a leaking pipe is free — the true cost of running operations on spreadsheets, computed honestly: reconciliation hours, error rates, the one-person dependency, and what they buy you compared to a governed system.
Nobody chooses spreadsheets for operations; everybody arrives at them. The stock tracker was a temporary fix in 2021. The payment log started as one tab and is now eleven. The master workbook has formulas one person understands and a filename ending in FINAL_v7_USE_THIS. It all works — in the sense that a leaking pipe still delivers water. The question is not whether spreadsheets function; it is what they actually cost, and the honest arithmetic surprises almost everyone who runs it.
The five costs, computed
1. Reconciliation labor: the salary nobody itemizes
The defining spreadsheet activity is making numbers agree: the stock tab against the sales log, the payment sheet against the bank, the branch workbook against head office. A typical SME finance or admin person spends 30–50% of their time reconciling records a connected system would never let diverge. At a KES 80,000 salary, that is KES 24,000–40,000 a month buying agreement between documents — roughly the cost of the software that makes the disagreement impossible.
2. Error rates: the research is not kind
Peer-reviewed audits of real-world spreadsheets consistently find errors in the overwhelming majority of workbooks — formula errors, broken ranges after a row insert, copy-paste drift. In operations spreadsheets the errors compound silently: a VLOOKUP that stopped matching in March misprices stock until someone notices at year-end. Systems have bugs too, but a bug is fixed once for everyone; a spreadsheet error is private, unaudited, and yours.
3. The one-person dependency
Every serious operations workbook has an author, and the author is a single point of failure. When they resign, travel, or fall sick, the organization discovers that the workbook was not a system but a performance — one that only they could give. Institutions have lost entire operating histories to a corrupted file and a departed employee in the same quarter. A governed system's knowledge lives in the system; a spreadsheet's lives in its maker.
4. No controls, no trail
A spreadsheet cannot enforce anything: approval thresholds are a convention, past dates are editable, and there is no record of who changed what. Every control an auditor asks about — approvals before commitment, three-way matching, attributable changes — must be performed by humans remembering to behave. That is not a control environment; it is a courtesy. And the gap is exactly where shrinkage, quiet overspends, and procurement findings live.
5. Decisions made on stale fiction
A spreadsheet is right at most once — the moment it was updated. Every decision between updates runs on fiction: stock that was sold, budgets already committed, prices that moved. The owner checking the dashboard on Sunday night is reading last Tuesday. Real-time is not a luxury feature; it is the difference between managing a business and reviewing its history.
The honest comparison
| Dimension | Spreadsheets | Governed system (AWRA) |
|---|---|---|
| Upfront cost | Zero — the hook | Subscription in KES, published on the plans page |
| Monthly true cost | Reconciliation labor + error losses + stale-data decisions | The subscription, plus onboarding effort up front |
| Data entry | Everything typed, often twice (cashbook + tracker) | Captured once at the transaction, flows everywhere |
| Controls | Conventions, hopefully followed | Enforced: approvals, thresholds, matching, audit trails |
| Multi-user reality | Version chaos or one-at-a-time locking | Concurrent, role-based, attributed |
| When the expert leaves | The system leaves with them | The system stays; access is revoked in one click |
| Audit week | Reconstruction | Retrieval |
| Scale (2nd branch, 5th grant, 20th staff) | Complexity grows quadratically | One more location/dimension in the same structure |
What spreadsheets are still for
This is not spreadsheet abolition. Modeling, one-off analysis, planning scenarios, the odd import template — spreadsheets remain unbeatable where the work is exploratory and personal. The line is simple: the moment a spreadsheet becomes a shared, permanent record that money moves against, it is doing a system's job without a system's guarantees.
The migration is smaller than the mythology
The spreadsheet's last defense is switching-cost fear — years of tabs, custom logic, the way we do things. In practice, migration takes master data (items, customers, suppliers) and opening balances; history stays archived in the workbooks for reference. The implementation checklist walks the whole path, and the pricing guide prices it honestly. Most teams are shocked less by the effort than by what the first connected month reveals — the variances the spreadsheets had been quietly absorbing for years.
Run the arithmetic on your own operation
Count the reconciliation hours, price the last error, and compare it to a published KES plan — then bring one messy workbook to a demo and watch it become a system.
Compare AWRA to your spreadsheetsFrequently asked questions
Our spreadsheets work fine. Why change what isn't broken?
"Working fine" in spreadsheet operations usually means the failures are invisible: unexplained stock variance, margins computed on stale costs, and reconciliation labor priced at zero because it is someone's normal day. Run one honest week of counting those three and the "fine" recalculates itself.
Can we keep some spreadsheets after moving to a system?
Yes — analysis, modeling, and planning stay in spreadsheets, fed by exports of trustworthy system data. What ends is the spreadsheet as the record: no more typing transactions into tabs, no more version reconciliation, no more FINAL_v7.
What is genuinely worse about a system compared to Excel?
Honesty deserves an answer: a system is less flexible by design — you cannot invent a column mid-morning or fudge a date, and that rigidity is the control. There is also real onboarding effort and a subscription. The trade is flexibility-for-one against reliability-for-everyone; businesses past a handful of people rarely regret it.
How long does it take to get off the spreadsheets?
For a typical SME: master data and opening balances in the first week or two, one module live within a month, and the workbook retired module by module rather than in one leap. The pattern that fails is running both forever "to be safe" — set a retirement date per spreadsheet and honor it.
