Publicité
ERP IMPLEMENTATION
🇫🇷 Lire en français →

Ghost Spreadsheets After ERP Go-Live: Why Excel Comes Back and How to Wind It Down

Excel creeping back after ERP go-live? 6 root causes, a 4-step audit method, internal control risks and a 90-day plan to retire shadow spreadsheets.

Ghost Spreadsheets After ERP Go-Live: Why Excel Comes Back and How to Wind It Down

Three months after go-live, the finance director’s dashboard shows figures nobody can trace back to the ERP. Accounts payable keeps a due-date tracker “just to be safe”. Management accounting rebuilds margin by job in a ten-tab workbook. The project is officially closed, but the real numbers live elsewhere: in ghost spreadsheets.

This is not a rare anomaly. It is one of the most common forms of silent rejection, and the one that never shows up in login statistics. This guide is written for CIOs, CFOs and project leads in post-deployment. It does not repeat the generic adoption indicators (see our 8 KPIs to detect silent rejection). It focuses on one specific object, the parallel Excel file, and on the method to inventory it, assess it and wind it down.

The Symptom: The ERP Is Live, but the Real Numbers Live in Excel

A ghost spreadsheet is a file that duplicates, corrects or extends a function the ERP is supposed to handle. It appears in no application inventory, nobody has validated it, and yet decisions rely on it.

The signals are easy to spot once you know where to look:

  • Meetings where people cite “Sophie’s file” instead of an ERP report.
  • Systematic ERP exports: the same extract is run every Monday to be reworked by hand.
  • Gaps between official and actual reporting: two versions of the truth are in circulation.
  • Excel attachments in month-end emails: the close calendar depends on files travelling by email.
  • Indispensable individuals: if only one person understands the file, you have a continuity risk on top of a control risk.

These files are not born of bad faith. They are the rational response of a user who has to deliver a result and cannot find, in the new system, a way to do it fast enough or reliably enough. Understanding that logic is the precondition for winding them down without alienating your teams.

Six Root Causes: The Telltale Signal and a Concrete Fix

1. A Need Missed During Scoping

Signal: the file performs a calculation or tracking task that appears in no specification document (warranty tracking, sales commissions, particular intercompany recharges).

Fix: requalify the need as a formal change request, with a business owner and a priority. Depending on the case, it is handled through configuration, a custom field, a standard report or an extension. If full coverage is disproportionate, explicitly document the decision to keep a controlled file (see below).

2. Reporting That Is Too Slow or Too Rigid

Signal: the user exports data to cross-reference, filter or present it differently, because the standard report does not answer their question or takes too long to display.

Fix: build a short list of expected reports based on the exports people actually run, then deliver them as views, dashboards or a lightweight data warehouse. The acceptance criterion is simple: get the answer faster than in Excel. If you cannot, the file will survive.

3. Access Rights That Are Too Restrictive

Signal: users without access to certain data (another site’s inventory, margins, cost prices) get it passed to them as files.

Fix: review the permission matrix with process owners, distinguishing viewing from editing. Opening targeted read-only views costs less, and carries less risk, than letting uncontrolled extracts circulate. Segregation-of-duties constraints must of course still be respected.

4. Poor Data Quality

Signal: users “fix” in their own file data they consider wrong in the ERP (addresses, supplier lead times, bills of materials, prices).

Fix: treat the file as an anomaly detector. Every local correction must be escalated, fixed at the source and assigned to a data owner. Without that loop, the file becomes a second reference database and the gap widens. Master data governance relies on the same mechanisms as those described in our article on post-deployment ERP governance.

5. Lack of Training

Signal: the user does not know the function exists, or does not know how to trigger it. The spreadsheet manually reproduces what the ERP does natively.

Fix: short sessions centred on the real task (“prepare customer dunning”, “track budget commitments”), run on the team’s own data rather than a demo dataset. The role-based programme is detailed in our guide to ERP user training.

6. Habits

Signal: no identifiable technical obstacle. The file has existed for years, the user masters it, and it makes their day-to-day feel safe.

Fix: this is the most underestimated cause, and the only one where technology is not enough. You have to make the alternative visibly simpler and support the transition: a key user alongside the team, a defined cohabitation period, a clear personal benefit (less re-keying, fewer late-night closes). Banning Excel by memo has the opposite effect: files migrate to local drives and become invisible.

Auditing Spreadsheets: A 4-Step Method

Step 1. Inventory Through Interviews

Do not start with an automated scan of file servers: it surfaces thousands of workbooks without telling you which ones matter. Work through interviews with department heads and key users, asking three questions: which files do you use every week to do your job? Which do you pass on to someone else? What happens if you are absent for a week? Present the exercise as help with simplification, not a compliance check, otherwise the most sensitive files will not be declared. For each one, record the owner, frequency, data source and recipients.

Step 2. Rank by Criticality and Risk

Assess each file on two axes. Criticality: does a decision, an accounting entry or a figure communicated externally depend on it? Risk: undocumented complex formulas, macros, manual entry, personal data, dependence on a single person. A simple four-quadrant grid is enough. Files that are both critical and risky are handled first.

Step 3. Map the Data Flows

For priority files, draw the path of the data: where it comes from in the ERP (which screen, which extract), what the file does to it, and where the result goes (email, accounting, management, re-entry into the ERP). This mapping often reveals that a single file feeds three others: removing one breaks the rest.

Step 4. Decide: Keep, Integrate or Delete

  • Delete: the file duplicates an existing function; you only need to support the user.
  • Integrate: the need is legitimate but belongs in the ERP (configuration, report, field, workflow) or in the BI tool.
  • Keep with safeguards: the use is legitimate and occasional (see next section).

Each decision has an owner, a deadline and a tracking indicator.

The Real Risks of Parallel Spreadsheets

The arguments against spreadsheets are well known, but they are best framed as business risk rather than dogma.

Formula errors. A spreadsheet has no input validation and no built-in testing. A range extended incorrectly, a cell overwritten by a value, a link to a file that was moved: the error is silent and spreads. The academic literature on spreadsheet errors is extensive, and cases of published figures distorted by a single formula are not rare. The takeaway for you: a file not reviewed by a third party is an unmeasured risk.

Internal control. The ERP provides traceability (who entered, who approved, when) and segregation of duties. A file anyone can edit, with no reliable history, bypasses those controls. If that file feeds a journal entry or a closing figure, the audit trail is broken.

Audit. External and internal auditors increasingly ask how end-user computing (EUC: spreadsheets, macros, small databases) that contributes to financial information is controlled. A critical file with no owner, no documentation and no version control becomes a point of attention, even a written recommendation. On these traceability issues, see also our article on the ERP audit trail.

GDPR. An export of the customer, HR or payroll database saved on a local machine, sent by email or copied to a USB stick escapes the record of processing activities, retention rules and the ERP’s security measures. An absence tracker with reasons, or a list of employees with their pay, are typical cases. The risk is not only a leak: you also cannot answer an access or erasure request if you do not know where the data sits. For the general framework, see our GDPR and ERP guide.

Which Spreadsheets to Keep: When Excel Stays Legitimate

The goal is not zero spreadsheets. Excel remains an excellent tool for:

  • one-off analyses (one question at a time, no recurrence);
  • simulations and scenarios (alternative budget, impact of a price change, plan assumptions);
  • prototyping a future report or indicator before industrialising it in the ERP.

The useful rule: a spreadsheet is legitimate when it consumes ERP data without replacing it. It becomes a ghost when it corrects, stores or feeds that data back without control.

Safeguards for the files you keep:

  1. A named owner and a deputy.
  2. A single source: data comes in by export or connection, not by re-keying.
  3. A shared, backed-up location with version history, not a local drive.
  4. A one-page document: purpose, sources, calculation logic, recipients.
  5. A third-party review for files that feed a decision or a close.
  6. An annual review: does the need still exist, can it be integrated?

A 90-Day Wind-Down Plan

The schedule below is a template to adapt to the size of your organisation. It assumes a sponsor (CFO or CIO) and a designated lead.

WeeksActionOwnerIndicator
1Kick-off: announcement, reassuring message, appointment of the lead and relay key usersSponsor + leadScope and schedule approved
2 to 3Interview-based inventory, department by departmentLead + key usersNumber of files identified (baseline)
4Criticality / risk rankingLead + internal control% of files ranked
5 to 6Data flow mapping for priority filesKey users + ITFlows documented for critical files
7Keep / integrate / delete decision, priority arbitrationSteering committee% of files with a decision
8 to 10Quick wins: reports, access rights, data fixes, targeted training sessionsIT / vendor / trainersFiles deleted, reports delivered
11Heavier integrations planned in the ERP roadmapIT + businessChange requests estimated and scheduled
12Safeguards put in place on retained filesFile owners% of files documented with an owner
13Review: recount, team feedback, integration into governanceSponsor + leadNumber of spreadsheets before / after

Measuring before and after is the essential point: without a baseline in week 3, you can neither demonstrate the result nor sustain momentum. Track three figures: the number of critical spreadsheets identified, the number deleted or integrated, and the number kept with safeguards.

The Role of Key Users

Key users are the main lever of the plan. They know the real reasons a file exists, they have the legitimacy to propose an alternative, and they can sit next to the user to test it. Free up their time explicitly: a wind-down plan handed to overloaded key users fails through exhaustion. They fit naturally into the logic of the post-go-live hypercare phase, of which this plan is an extension.

An Illustrative (Fictional) Case

Take a fictional industrial SME, used here for illustration only. Six months after its ERP went live, the finance department discovers through interviews that management accounting maintains a workbook tracking margin by job. The reason: the standard report ignores late-entered subcontracting hours, and it takes several minutes to display. The analysis shows three overlapping causes (a poorly scoped need, performance, late data entry). The response was a dedicated report, an entry reminder in the workflow and a hands-on session for the team. The workbook was retired at the following close, after a one-month cohabitation period in which both sources were compared. The case is deliberately simple: its value lies in the approach (understand the cause before deleting), not in the figures.

Prevention: Build the Spreadsheet Inventory Into Every Future Project

The best time to deal with ghost spreadsheets is before they appear. Three practices to build into your project method:

  1. File inventory at scoping. Before specifying, list the spreadsheets in use in each department and ask of each: what does it do that the current tool does not? It is a goldmine of real needs, often missing from formal workshops. Our guide to migrating from Excel to an ERP details this approach.
  2. An exit criterion in acceptance testing. Each identified spreadsheet receives a decision (covered, to be developed, kept) and testing verifies that critical cases are handled.
  3. Reviews at D+90 and D+180. Run a short inventory again after go-live and hand it to lasting governance, for example the ERP center of excellence.

The guiding principle stays the same throughout: do not ban Excel, but offer a faster and more reliable alternative. Users only leave their file if the new path saves them time from the first week.

Going Further

To quickly assess your situation, download our ERP evaluation grid: 30 criteria to objectify the gaps between what your system covers and what your teams actually do. For a broader framework on adoption tracking, also read our 8 KPIs to detect silent rejection.