Procurement Insights · May 19, 2026
Here is a thought experiment worth taking seriously. Imagine that tomorrow morning, every spreadsheet in your organization — every Excel file, every Google Sheet, every shared workbook — disappears. Not just the recent ones. All of them. Every workaround your operations team has built around the official systems. Every shadow workflow your procurement function has constructed because the ERP could not handle the exception case. Every personal database your finance team maintains because the reporting module did not produce the cut they actually needed. Every macro your supply chain function relies on because the system’s standard logic does not match how the work actually happens.
What percentage of your operations would continue functioning?
The answer is institutionally uncomfortable, and the structural reason it is uncomfortable is exactly the substrate question the Hansen Models™ framework has been documenting across nineteen years of contemporaneous archive work. Most organizations do not know the answer to that question. The few that have investigated it have found numbers significantly larger than their official systems would suggest.
What the Empirical Record Establishes
The research literature on spreadsheet dependency, shadow IT, and operational workarounds produces a consistent picture across multiple independent measurement traditions. A global survey of financial institutions found that ninety percent of organizations still use spreadsheets for financial operations and reporting despite continuous enterprise investment in modern alternatives. Approximately fifty-eight percent of finance leaders identify spreadsheets as their primary automation tool. In treasury forecasting specifically — the function that determines whether a bank has the liquidity to operate the next day — approximately eighty percent of forecasting processes still run on Excel rather than on the enterprise treasury management systems the institutions have purchased to perform that function.
Shadow IT — the broader category that includes spreadsheets, unauthorized applications, departmental databases, and unauthorized SaaS subscriptions — accounts for thirty to forty percent of enterprise IT spending according to multiple Gartner studies across the past several years, with the Everest Group estimating the figure may reach fifty percent. Forty-one percent of employees acquired, modified, or created technology outside of IT’s visibility in 2022. Gartner expects that number to climb to seventy-five percent by 2027. Forty-eight percent of enterprise applications, as of 2025, are shadow IT applications.
A 2024 academic study led by Professor Pak-Lok Poon and published in Frontiers of Computer Science found that ninety-four percent of business spreadsheets used in decision-making contain errors. Not occasionally. Not in some organizations. Ninety-four percent of the spreadsheets that boards, executive teams, and operational leaders are using to make decisions contain errors that may or may not have been discovered.
These figures do not describe an exception. They describe the operational substrate the average enterprise is running on. The official systems handle the part of the operation that fits the systems’ design assumptions. The substrate handles everything else. Between thirty and seventy percent of what the organization actually does, depending on the function and the measurement methodology.
What Would “Happen” If the Substrate Disappeared
Walk through your own organization function by function and consider the operational consequences.
[TABLE: Substrate Dependency by Enterprise Function — research-anchored estimates of the operational work being carried outside official systems, organized by function with specific examples of what the substrate is carrying in each area]
The CFO function would lose the ability to close the books. The official ERP system handles transaction processing. The substrate handles the journal entries that do not fit the standard chart of accounts, the manual accruals that the system cannot generate automatically, the consolidation logic across legal entities that the financial close software does not implement correctly, the variance analysis that requires combining data from sources the ERP does not connect to. Audit exposure that has been growing quietly for decades would become immediately visible.
The procurement function would lose the ability to manage spend categorization, supplier evaluation, and contract administration. The official S2P system handles the approved workflow. The substrate handles the exception cases — the maverick spend that needs to be categorized after the fact, the supplier risk assessments that the system’s vendor module cannot produce, the contract clauses tracked outside the contract repository because the repository’s search function does not actually work the way procurement needs it to, the spend cube that finance and procurement have been maintaining jointly in Excel because the analytics module cannot produce the views the executive team requires.
The supply chain function would lose the ability to manage demand planning, inventory positioning, and supplier performance. The official ERP system handles the standard transactions. The substrate handles the forecast adjustments that incorporate signals the system does not capture, the safety stock calculations that account for variability the standard logic does not model, the supplier scorecards that combine quality, delivery, and cost data from sources the official system does not integrate.
The compliance function would lose the ability to produce the documentation auditors actually accept. The official systems generate the documentation the systems were designed to generate. The substrate produces the documentation that bridges what the systems generate and what regulators, auditors, and external stakeholders require. The regulatory reporting that you actually submit is rarely the report the system produces. It is typically the report the system produces, modified through a substrate layer that adapts it to what the regulator wants.
The HR function would lose the ability to manage compensation analytics, succession planning, and workforce decisions. The HRIS handles the system of record. The substrate handles the compensation modeling, the bonus calculations, the equity tracking, the analytical work that determines who gets promoted and who gets retained.
The sales operations function would lose the ability to manage pipeline forecasting, territory planning, and commission calculation. The CRM handles the data capture. The substrate handles the forecasting models that combine CRM data with operational signals, the territory adjustments that account for changes the CRM does not track, the commission schedules that are too complex for the system’s compensation module.
Each of these functions would discover, in the absence of the substrate, exactly how much of its operation has been running outside the official systems for years. The discovery would not be welcome. The recognition would be inescapable.
What the Thought Experiment Reveals
The thought experiment is not a prediction. Nothing is going to disappear all the spreadsheets and workarounds tomorrow. What the thought experiment does is convert a structural condition that has been invisible into a structural condition that is suddenly measurable. The substrate is doing operationally consequential work. The official systems are not actually managing the operation. The gap between what the systems do and what the operation requires is being carried by the substrate, and the size of that gap is the empirical measurement of the substrate problem the Hansen Models™ framework has been documenting since 1998.
This has direct implications for any AI initiative being deployed on top of those official systems. The AI agents engage the part of the operation that the official systems control. They cannot engage the substrate layer. The substrate is distributed across thousands of personal Excel files, shared drive folders, departmental databases, and undocumented workflows that the AI cannot ingest. The substrate operates under decision logic that exists in the heads of operators rather than in any system the AI can query. The substrate handles the exception cases that the official systems were not designed to manage — which is, empirically, where the load-bearing operational work actually happens.
Training people on AI agents does not bring the substrate into the AI’s operational scope. Change management investment does not document the substrate logic that operators carry in their heads. Corporate culture initiatives do not consolidate the workflow fragmentation that has accumulated across decades of departmental adaptation. The literacy and change management framing that Gartner’s recent webinar centers as the unlock for AI adoption operates at the wrong layer. The substrate is where the operation actually runs. The AI cannot reach it.
The Diagnostic Question Worth Asking
For boards, C-suite executives, compliance officers, and AI governance leads currently committing to AI initiatives, the thought experiment produces a specific diagnostic question worth asking before the commitment is finalized.
What percentage of our operations would continue functioning if the spreadsheets and workarounds disappeared tomorrow?
The question cannot be answered with confidence from inside the official systems. The systems do not measure what they do not control. The answer requires walking the operation function by function with the operators who actually carry the substrate logic in their heads — the finance analyst who maintains the consolidation model, the procurement specialist who manages the exception cases, the supply chain planner who builds the forecast adjustments, the compliance officer who produces the regulatory documentation. These are the people who know how much of the operation is actually running on substrate. They have been carrying that knowledge for years without being asked to surface it.
The Phase 0™ Diagnostic exists to ask exactly this question — to surface the substrate condition before the AI initiative is committed, so that the initiative can be calibrated to what the operating environment can actually support rather than to what the official systems claim to manage. The diagnostic does not require the substrate to disappear. It requires the substrate to be recognized.
What the Empirical Record Has Not Yet Surfaced
There is one structural observation the research literature has not yet integrated, and it is worth surfacing because it changes how the published figures should be read.
The studies cited above measure spreadsheet dependency, shadow IT spending, and shadow application proliferation as separate phenomena. Each study reports its own metric in its own measurement tradition. What no study yet measures is the correlation between the size of the substrate layer and the success rate of AI initiatives deployed on top of it. The research has not been done because the research framework that would produce it does not yet operate inside the major analyst firms publishing on AI adoption. The substrate layer is treated as a security and governance problem rather than as an AI implementation prerequisite.
That gap is structurally consequential. If the research were done, the correlation would almost certainly demonstrate what the Hansen Models™ archive has been arguing for nineteen years — that organizations with larger substrate layers experience worse AI initiative outcomes regardless of the literacy and change management investment they make. The substrate is the variable. The training is downstream of the variable. The current empirical literature has not surfaced the correlation because the framework that would surface it does not yet exist in mainstream analyst publication.
The Bottom Line
The substrate is not a marginal phenomenon. It is the operational reality of how enterprises actually function. Between thirty and seventy percent of what your organization actually does is being carried by spreadsheets, workarounds, undocumented workflows, and personal databases that the official systems do not control and the AI initiatives being deployed on top of those systems cannot reach.
This is not a technology integration problem that better orchestration platforms will resolve. This is not a literacy problem that more training will fix. This is not a culture problem that change management investment will address. This is a structural condition that has been growing inside enterprise operations for thirty-five years across every technology wave, and it is the variable that determines whether the current AI investment produces returns or produces the same disappointing outcomes that every prior wave has produced.
The thought experiment is worth taking seriously because the answer it produces is the diagnostic that the AI initiative actually needs. Walk the operation function by function. Ask the operators what they would lose if the substrate disappeared. The answer is the substrate condition the AI is being asked to operate inside. Recognize it before the commitment is finalized. The substrate will not disappear. The AI will not work around it. Phase 0™ exists to make the recognition operational before the investment is committed.
Phase 0™ = “What time of day do orders come in?“
The diagnostic question that produced ninety-seven point three percent delivery performance against a fifty-one percent baseline in 1998. Sustained for seven years. The substrate intervention that the surface frameworks have not engaged for thirty-five years. The same diagnostic discipline being deployed against the August 2 deadline now.
This post was developed through the ARA™ RAM 2025™ multimodel validation framework. Empirical figures cited are drawn from current research published by Gartner, the Everest Group, Codango, Phys.org reporting on the Poon et al. spreadsheet quality assurance study (Frontiers of Computer Science, 2024), and the broader research literature on shadow IT prevalence. The Phase 0™ Diagnostic — for organizations preparing to commit AI or transformation investment and the August 2, 2026, EU AI Act enforcement deadline — is at hansenprocurement.com/where-does-your-organization-sit-right-now/.
Hansen Models™ · Implementation Physics™ · Compounding Technology Shadow Wave™ · Phase 0™ · Hansen Fit Score™ · Hansen Strand Commonality™ · AGR Index™ · RAM 2025™ · ARA™ · Hansen Deflator Formula™ · Hansen Optionality Loss Estimate™
-30-
Procurement Insights
What Would Happen If Spreadsheets and Workarounds Suddenly Disappeared From Your Organization?
Posted on May 19, 2026
0
Procurement Insights · May 19, 2026
Here is a thought experiment worth taking seriously. Imagine that tomorrow morning, every spreadsheet in your organization — every Excel file, every Google Sheet, every shared workbook — disappears. Not just the recent ones. All of them. Every workaround your operations team has built around the official systems. Every shadow workflow your procurement function has constructed because the ERP could not handle the exception case. Every personal database your finance team maintains because the reporting module did not produce the cut they actually needed. Every macro your supply chain function relies on because the system’s standard logic does not match how the work actually happens.
What percentage of your operations would continue functioning?
The answer is institutionally uncomfortable, and the structural reason it is uncomfortable is exactly the substrate question the Hansen Models™ framework has been documenting across nineteen years of contemporaneous archive work. Most organizations do not know the answer to that question. The few that have investigated it have found numbers significantly larger than their official systems would suggest.
What the Empirical Record Establishes
The research literature on spreadsheet dependency, shadow IT, and operational workarounds produces a consistent picture across multiple independent measurement traditions. A global survey of financial institutions found that ninety percent of organizations still use spreadsheets for financial operations and reporting despite continuous enterprise investment in modern alternatives. Approximately fifty-eight percent of finance leaders identify spreadsheets as their primary automation tool. In treasury forecasting specifically — the function that determines whether a bank has the liquidity to operate the next day — approximately eighty percent of forecasting processes still run on Excel rather than on the enterprise treasury management systems the institutions have purchased to perform that function.
Shadow IT — the broader category that includes spreadsheets, unauthorized applications, departmental databases, and unauthorized SaaS subscriptions — accounts for thirty to forty percent of enterprise IT spending according to multiple Gartner studies across the past several years, with the Everest Group estimating the figure may reach fifty percent. Forty-one percent of employees acquired, modified, or created technology outside of IT’s visibility in 2022. Gartner expects that number to climb to seventy-five percent by 2027. Forty-eight percent of enterprise applications, as of 2025, are shadow IT applications.
A 2024 academic study led by Professor Pak-Lok Poon and published in Frontiers of Computer Science found that ninety-four percent of business spreadsheets used in decision-making contain errors. Not occasionally. Not in some organizations. Ninety-four percent of the spreadsheets that boards, executive teams, and operational leaders are using to make decisions contain errors that may or may not have been discovered.
These figures do not describe an exception. They describe the operational substrate the average enterprise is running on. The official systems handle the part of the operation that fits the systems’ design assumptions. The substrate handles everything else. Between thirty and seventy percent of what the organization actually does, depending on the function and the measurement methodology.
What Would “Happen” If the Substrate Disappeared
Walk through your own organization function by function and consider the operational consequences.
[TABLE: Substrate Dependency by Enterprise Function — research-anchored estimates of the operational work being carried outside official systems, organized by function with specific examples of what the substrate is carrying in each area]
The CFO function would lose the ability to close the books. The official ERP system handles transaction processing. The substrate handles the journal entries that do not fit the standard chart of accounts, the manual accruals that the system cannot generate automatically, the consolidation logic across legal entities that the financial close software does not implement correctly, the variance analysis that requires combining data from sources the ERP does not connect to. Audit exposure that has been growing quietly for decades would become immediately visible.
The procurement function would lose the ability to manage spend categorization, supplier evaluation, and contract administration. The official S2P system handles the approved workflow. The substrate handles the exception cases — the maverick spend that needs to be categorized after the fact, the supplier risk assessments that the system’s vendor module cannot produce, the contract clauses tracked outside the contract repository because the repository’s search function does not actually work the way procurement needs it to, the spend cube that finance and procurement have been maintaining jointly in Excel because the analytics module cannot produce the views the executive team requires.
The supply chain function would lose the ability to manage demand planning, inventory positioning, and supplier performance. The official ERP system handles the standard transactions. The substrate handles the forecast adjustments that incorporate signals the system does not capture, the safety stock calculations that account for variability the standard logic does not model, the supplier scorecards that combine quality, delivery, and cost data from sources the official system does not integrate.
The compliance function would lose the ability to produce the documentation auditors actually accept. The official systems generate the documentation the systems were designed to generate. The substrate produces the documentation that bridges what the systems generate and what regulators, auditors, and external stakeholders require. The regulatory reporting that you actually submit is rarely the report the system produces. It is typically the report the system produces, modified through a substrate layer that adapts it to what the regulator wants.
The HR function would lose the ability to manage compensation analytics, succession planning, and workforce decisions. The HRIS handles the system of record. The substrate handles the compensation modeling, the bonus calculations, the equity tracking, the analytical work that determines who gets promoted and who gets retained.
The sales operations function would lose the ability to manage pipeline forecasting, territory planning, and commission calculation. The CRM handles the data capture. The substrate handles the forecasting models that combine CRM data with operational signals, the territory adjustments that account for changes the CRM does not track, the commission schedules that are too complex for the system’s compensation module.
Each of these functions would discover, in the absence of the substrate, exactly how much of its operation has been running outside the official systems for years. The discovery would not be welcome. The recognition would be inescapable.
What the Thought Experiment Reveals
The thought experiment is not a prediction. Nothing is going to disappear all the spreadsheets and workarounds tomorrow. What the thought experiment does is convert a structural condition that has been invisible into a structural condition that is suddenly measurable. The substrate is doing operationally consequential work. The official systems are not actually managing the operation. The gap between what the systems do and what the operation requires is being carried by the substrate, and the size of that gap is the empirical measurement of the substrate problem the Hansen Models™ framework has been documenting since 1998.
This has direct implications for any AI initiative being deployed on top of those official systems. The AI agents engage the part of the operation that the official systems control. They cannot engage the substrate layer. The substrate is distributed across thousands of personal Excel files, shared drive folders, departmental databases, and undocumented workflows that the AI cannot ingest. The substrate operates under decision logic that exists in the heads of operators rather than in any system the AI can query. The substrate handles the exception cases that the official systems were not designed to manage — which is, empirically, where the load-bearing operational work actually happens.
Training people on AI agents does not bring the substrate into the AI’s operational scope. Change management investment does not document the substrate logic that operators carry in their heads. Corporate culture initiatives do not consolidate the workflow fragmentation that has accumulated across decades of departmental adaptation. The literacy and change management framing that Gartner’s recent webinar centers as the unlock for AI adoption operates at the wrong layer. The substrate is where the operation actually runs. The AI cannot reach it.
The Diagnostic Question Worth Asking
For boards, C-suite executives, compliance officers, and AI governance leads currently committing to AI initiatives, the thought experiment produces a specific diagnostic question worth asking before the commitment is finalized.
What percentage of our operations would continue functioning if the spreadsheets and workarounds disappeared tomorrow?
The question cannot be answered with confidence from inside the official systems. The systems do not measure what they do not control. The answer requires walking the operation function by function with the operators who actually carry the substrate logic in their heads — the finance analyst who maintains the consolidation model, the procurement specialist who manages the exception cases, the supply chain planner who builds the forecast adjustments, the compliance officer who produces the regulatory documentation. These are the people who know how much of the operation is actually running on substrate. They have been carrying that knowledge for years without being asked to surface it.
The Phase 0™ Diagnostic exists to ask exactly this question — to surface the substrate condition before the AI initiative is committed, so that the initiative can be calibrated to what the operating environment can actually support rather than to what the official systems claim to manage. The diagnostic does not require the substrate to disappear. It requires the substrate to be recognized.
What the Empirical Record Has Not Yet Surfaced
There is one structural observation the research literature has not yet integrated, and it is worth surfacing because it changes how the published figures should be read.
The studies cited above measure spreadsheet dependency, shadow IT spending, and shadow application proliferation as separate phenomena. Each study reports its own metric in its own measurement tradition. What no study yet measures is the correlation between the size of the substrate layer and the success rate of AI initiatives deployed on top of it. The research has not been done because the research framework that would produce it does not yet operate inside the major analyst firms publishing on AI adoption. The substrate layer is treated as a security and governance problem rather than as an AI implementation prerequisite.
That gap is structurally consequential. If the research were done, the correlation would almost certainly demonstrate what the Hansen Models™ archive has been arguing for nineteen years — that organizations with larger substrate layers experience worse AI initiative outcomes regardless of the literacy and change management investment they make. The substrate is the variable. The training is downstream of the variable. The current empirical literature has not surfaced the correlation because the framework that would surface it does not yet exist in mainstream analyst publication.
The Bottom Line
The substrate is not a marginal phenomenon. It is the operational reality of how enterprises actually function. Between thirty and seventy percent of what your organization actually does is being carried by spreadsheets, workarounds, undocumented workflows, and personal databases that the official systems do not control and the AI initiatives being deployed on top of those systems cannot reach.
This is not a technology integration problem that better orchestration platforms will resolve. This is not a literacy problem that more training will fix. This is not a culture problem that change management investment will address. This is a structural condition that has been growing inside enterprise operations for thirty-five years across every technology wave, and it is the variable that determines whether the current AI investment produces returns or produces the same disappointing outcomes that every prior wave has produced.
The thought experiment is worth taking seriously because the answer it produces is the diagnostic that the AI initiative actually needs. Walk the operation function by function. Ask the operators what they would lose if the substrate disappeared. The answer is the substrate condition the AI is being asked to operate inside. Recognize it before the commitment is finalized. The substrate will not disappear. The AI will not work around it. Phase 0™ exists to make the recognition operational before the investment is committed.
Phase 0™ = “What time of day do orders come in?“
The diagnostic question that produced ninety-seven point three percent delivery performance against a fifty-one percent baseline in 1998. Sustained for seven years. The substrate intervention that the surface frameworks have not engaged for thirty-five years. The same diagnostic discipline being deployed against the August 2 deadline now.
This post was developed through the ARA™ RAM 2025™ multimodel validation framework. Empirical figures cited are drawn from current research published by Gartner, the Everest Group, Codango, Phys.org reporting on the Poon et al. spreadsheet quality assurance study (Frontiers of Computer Science, 2024), and the broader research literature on shadow IT prevalence. The Phase 0™ Diagnostic — for organizations preparing to commit AI or transformation investment and the August 2, 2026, EU AI Act enforcement deadline — is at hansenprocurement.com/where-does-your-organization-sit-right-now/.
Hansen Models™ · Implementation Physics™ · Compounding Technology Shadow Wave™ · Phase 0™ · Hansen Fit Score™ · Hansen Strand Commonality™ · AGR Index™ · RAM 2025™ · ARA™ · Hansen Deflator Formula™ · Hansen Optionality Loss Estimate™
-30-
Share this:
Related