Is a spreadmart always an Excel file?

No. The term can include desktop databases and related local reporting systems.

Must every spreadmart be replaced immediately?

The appropriate response depends on its role, controls and alternatives. Documenting and managing an existing process may be a useful step.

Does automation remove the problem?

It can reduce manual work, but automatically applying inconsistent rules still produces inconsistent information.

AI Glossary / Data science and analytics / Advanced

Spreadmart

Definition

A spreadmart is a locally maintained reporting or analysis system, commonly built with spreadsheets or desktop databases, that performs functions associated with a data mart. These can include gathering data, transforming it, defining measures and distributing reports. The term often highlights the problems that arise when such systems become recurring organisational dependencies without shared definitions and adequate management.

What it means

A spreadsheet is not automatically a spreadmart. The concern is a local system that becomes an independent source of business data and rules, especially when other teams rely on its outputs without knowing how they were produced.
A spreadmart may also reveal a genuine unmet need. Staff may have built it because existing systems do not answer an important question or cannot provide results in time. Understanding that need is essential before replacing the system. Removing a workbook without preserving its useful function can simply encourage another unofficial version to appear.

How it works

A typical local reporting process imports extracts, applies formulas or manual adjustments, joins reference lists and publishes a summary. Over time, new tabs, linked files and exceptions accumulate. The workbook can become both a report and the place where business definitions are implemented.
Different copies may refresh at different times or use different inclusion rules. Two reports can therefore disagree even when both display a measure called “active customers.” Discrepancies may reflect timing, definitions or errors.
Review the data flow, transformations, ownership and downstream use. Decide which shared rules need managed implementation and which exploratory work should remain flexible. The solution concerns responsibilities and reproducibility as well as the choice of tool.

Three practical examples

A sales team refreshes a workbook from several exports each week and maintains its own customer grouping rules. Other departments reuse the totals without seeing the adjustments, creating a risk of inconsistent interpretation.
An editorial office combines article trackers into a launch report. Each contributor defines “complete” differently, so the combined workbook becomes a local reporting system whose numbers require reconciliation.
An operations analyst builds a temporary capacity model because the standard dashboard lacks a needed scenario. The model becomes a recurring planning dependency, prompting the organization to document and support it rather than leaving it with one owner.

Think of it this way

A spreadmart resembles a small workshop that grows into an essential production facility while still relying on handwritten instructions and one person's knowledge. The workshop may make something valuable. The question is whether its records, controls and continuity match the importance of the work it now performs.

Worked example

Suppose two fictional teams report completed glossary articles from the same 100-record extract. Team A counts every row marked “draft complete” and reports 80. Team B counts only rows that are both drafted and editorially approved and reports 65. Both use the heading “Completed articles.”
The difference is 80 − 65 = 15 articles. Team B's total is 15 / 80 = 18.75% lower than Team A's, but that does not establish that either team made an arithmetic error. Their definitions differ. Reconciliation shows that 15 drafted articles are awaiting approval.
The team creates two explicit measures: drafted articles = 80 and approved articles = 65. If the total scope remains 100, the corresponding rates are 80% and 65%. These invented figures demonstrate definition reconciliation, not the actual glossary project's status. A shared file would still mislead if the two meanings remained hidden under the same heading.

Illustrative case study

In a fictional project, Harbour Services depends on a monthly workbook maintained by one analyst. It combines service records, local category mappings and several manually entered exceptions. Managers trust the report, but the analyst's absence exposes that nobody else can reproduce the refresh reliably.
The team maps the workflow with the analyst, identifies the source for each adjustment and documents why the exceptions exist. It moves recurring shared calculations into a supported process while retaining a spreadsheet view for planning scenarios. Published figures include their refresh date and definition, and scenario changes are kept distinguishable from official results.
Harbor tests the replacement against representative historical cases before retiring the old process. The outcome is a clearer reporting responsibility and reproducible calculation path, not an assumed financial saving. This fictional case illustrates preserving useful local knowledge while reducing dependence on an undocumented system.

Common misconceptions and mistakes

Not every spreadsheet is a spreadmart, and moving the same unclear logic into a database does not automatically solve the problem. Shared infrastructure can still contain conflicting definitions or poorly maintained rules.
Blaming the workbook owner can obscure the business need that produced the system. Conversely, treating a successful local prototype as permanently sufficient can overlook growing dependence, access requirements and continuity needs.
A single published number is not necessarily a trustworthy number. Review its definition, source, timing and transformations. Reconciliation should explain discrepancies rather than merely force one report to match another without understanding the reason.

Business use

Use the concept to identify reporting processes whose importance has outgrown their management arrangements. Inventory recurring local systems and prioritize them by operational dependency, complexity and the consequences of incorrect outputs.
Work with their owners to preserve useful rules and context. Establish clear definitions, refresh procedures and responsibility for changes. A managed data source can coexist with spreadsheets used for analysis and presentation. Evaluate improvements through reproducibility, reduced reconciliation effort and continuity, using actual measurements rather than assuming a universal saving from migration.

Check your understanding

Why did the two reports disagree in the example?

Answer: They used different definitions of completion.

Why understand the owner's workflow before replacing it?

Answer: It may contain essential business rules and unmet requirements.

What is insufficient about moving the file to a shared folder?

Answer: Shared access alone does not resolve definitions, refresh logic or accountability.

Related terms

Data mart
Data governance
spreadsheet
shadow IT
business intelligence
metric definition
data lineage
reconciliation
self-service analytics

Learn next

Map one recurring spreadsheet report from source extracts to published totals. Identify where definitions, manual changes and refresh timing are recorded, then test whether another person can reproduce a result.

Sources

Eckerson and Sherman, Strategies for Managing Spreadmarts, TDWI, 2008: original research-based definition, business context and management approach. https://tdwi.org/articles/2008/05/27/strategies-for-managing-spreadmarts.aspx Wayne Eckerson interview, DSSResources, 2005: spreadmarts as surrogate data marts and the wider management issue. https://dssresources.com/interviews/eckerson/eckerson08052005.html Examples and calculations are original. Historical survey prevalence and cost estimates are not presented as current or universal facts.

Previous: Spreading activation Next: Spurious Tuple