Insight
The hidden cost of the spreadsheet layer
Parallel spreadsheets are not free workarounds. They are an unpaid operational system with error rates, key-person risk, and leadership blind spots.
Every growing business has them: the pricing sheet only one person understands, the dispatch tracker updated at noon and again at five, the weekly export pasted into Excel for leadership because the ERP report is wrong.
We call this the spreadsheet layer: parallel systems that exist because the official tools never quite fit.
Is the spreadsheet layer already a system?
Short answer: Yes. It has data models, integrations, access control, a release process, and on-call. It just lacks versioning, audit trails, and error handling, so failures stay silent until they are expensive.
The spreadsheet layer has:
- Data models (tabs, pivots, VLOOKUP chains)
- Integrations (CSV exports, copy-paste from email)
- Access control (who has the link, who broke the formula)
- Release process ("don't sort column D or totals break")
- On-call (whoever gets the call when numbers don't match finance)
How do you estimate the hidden cost?
Short answer: Track labour hours for one week, add recent error cost, then add delay cost from decisions waiting on manual refreshes.
Monthly cost ≈ (hours spent × loaded labour rate) + error cost + delay cost
Labour hours
Track one week honestly:
| Activity | Who | Hours/week |
|---|---|---|
| Maintaining master spreadsheets | ||
| Reconciling exports between systems | ||
| Fixing broken formulas / versions | ||
| Building ad-hoc reports for leadership | ||
| Explaining numbers that don't match |
Multiply average hours by loaded rate (salary + overhead, often 1.3–1.5× gross).
A team spending 15 hours/week at €45/hour loaded costs ~€2,900/month, or ~€35,000/year before error costs.
Error cost
Pick one recent incident:
- Wrong price shipped on an order
- Double booking because two sheets diverged
- Inventory mismatch that delayed a customer
If one meaningful error costs €5,000 and happens quarterly, that is €20,000/year in a single category.
Delay cost
Decisions waiting on Monday's spreadsheet refresh have a cost too: slower collections, missed capacity, rep time on status calls instead of sales.
When are spreadsheets still fine?
Short answer: Low volume, low consequence, one maintainer, and data that is reference-only rather than operational truth.
- Low volume, low consequence
- One maintainer, clear rules, rare changes
- Data is reference-only, not operational truth
When should the layer go?
Short answer: When versions conflict, leadership distrusts the numbers, or the same errors recur every quarter.
- Multiple people maintain conflicting versions
- Leadership makes decisions on numbers nobody trusts
- Onboarding new staff requires "learn the sheet"
- Errors recur in the same column every quarter
- Integrations are manual exports run on faith
At that point, a focused build (see current bands on Pricing) can pay back on labour alone within a reasonable horizon, before error reduction.
Why not just govern spreadsheets better?
Short answer: Governance helps briefly. It fails when rules, roles, or integrations outgrow what a grid can hold.
Better governance (single source template, locked ranges) helps briefly. It fails when:
- Rules outgrow formula complexity
- Role-based visibility is needed
- Real-time integration replaces daily exports
- Audit trails matter for disputes or internal control
Custom software replaces the workflow, not just the grid.
What usually unlocks the fix?
Short answer: Integrations. Spreadsheets often exist because systems do not talk.
Integrations remove the paste step that created the sheet. How we design them is on Backend systems.
How do you build the business case internally?
- Measure one week of spreadsheet labour (names and hours, not guesses)
- List recent errors with euro impact where possible
- Name the operational owner who will sign off a replacement
- Hold a EUR range from Pricing. Discovery narrows it.
- Compare to SaaS using build vs buy. Sometimes the sheet exists because SaaS failed.
Present Discovery Sprint as paid insurance against quoting a large build on wrong assumptions.
Next step
If the arithmetic surprises you, the problem is real enough to discuss.
Use the cost calculator for a planning band, book a fit call, or request discovery.