The Reporting Gap, Two Ways
Spreadsheets and BI tools fail in opposite ways, one from too little structure, the other from structure nobody uses.
Two organisations can have completely different reporting infrastructure and land in the same place: neither one trusts the number on the screen. One is still building reports by hand. The other bought a BI platform two years ago and mostly stopped using it. Both problems look different from the outside, but they share the same root cause, and neither shows up on a balance sheet until a decision goes wrong.
The Finding
Organisations without a business intelligence tool tend to assume the fix is simple: buy one. Organisations with one tend to assume the problem is solved because the tool exists. Neither assumption holds up. Spreadsheet-based reporting carries a well-documented error rate that has not meaningfully improved since researchers began measuring it decades ago, and BI tools, once purchased, frequently stall well short of the adoption their cost would suggest. The two failure modes sit at opposite ends of the same curve: too little structure, or structure nobody actually uses.
What This Actually Costs
Put a number on either failure mode and the picture sharpens. Analysts relying on manual spreadsheet prep spend 10 to 11 hours a week just collecting and preparing data before any actual analysis starts (Alteryx, 2025), time that a working BI tool is specifically meant to reclaim. Yet a Forrester study of 63 organisations using Power BI found self-service adoption alone saved users 125 hours a year, meaning most of that potential saving goes unclaimed wherever adoption stalls. The cost isn't hypothetical on either side, it's just hidden in different places.
Why It Happens
Excel
Spreadsheets scale error, not insight. A single incorrect formula or a stale copy-paste can propagate silently across dozens of downstream cells, and nothing in the format itself catches it. The more a spreadsheet is relied on, the more surface area it has for an error nobody's checked.
Power BI
BI tools get bought, not adopted. Purchasing a platform solves a procurement problem, not a workflow one. Without a deliberate rollout that changes how people actually build and request reports day to day, most staff default back to whatever got the job done before the tool arrived.
Neither failure mode announces itself. A wrong spreadsheet number and an unused dashboard both look, from the outside, exactly like a working reporting process. The gap only becomes visible when a decision built on either one turns out to be wrong.
What the Research Shows
The spreadsheet-error research is unusually consistent for a finding this uncomfortable: independent studies across different organisations and time periods keep landing close to the same number, roughly 94% of spreadsheets in active use containing at least one error. The BI adoption picture tells a related story from the other direction. Gartner's research places most organisations' BI and analytics use at around the 30% scaling mark, meaning the majority never move past piloting despite the initial investment. Forrester's finding sharpens why: when only one in five non-IT staff can actually self-serve their own reports, the tool has effectively been handed to a small group of gatekeepers, and everyone else is back to asking someone else for the number, the same trust bottleneck spreadsheets create, just with better graphics.
What Fixes It
1. Treat spreadsheets as a stage, not a destination. A spreadsheet used for a first pass at a new report is a reasonable starting point. A spreadsheet still being relied on as the permanent system of record a year later is where the error rate research starts to matter.
2. Adoption is a workflow problem, not a training problem. Getting a BI tool used at scale means changing how a report request actually moves through a team, not just running a session on how the software works.
3. Measure trust, not just usage. Login counts and dashboard views don't tell you whether people believe the number they're looking at. That has to be asked directly, and tracked over time, not assumed from activity data.
Whichever end of this gap an organisation sits at, the underlying question is the same: does the reporting process the team actually uses produce a number people are willing to stand behind, or one they've simply stopped questioning. It's worth noting that spending more doesn't automatically move an organisation from one side of this gap to the other. A well-funded BI rollout with low adoption is still, functionally, a spreadsheet problem wearing better software, the number reaching decision-makers is only as trustworthy as the process behind it, regardless of what that process cost to build.
Sources & References
Panko, R. Spreadsheet error research (cumulative body of work, ongoing since 1995). Gartner. Survey on business intelligence and analytics adoption. Forrester Research. Self-service BI accessibility for non-IT staff. Forrester Consulting (commissioned by Microsoft). The Total Economic Impact of Microsoft Power BI, 63 organisations studied. Alteryx. State of Data Analysts in the Age of AI, 2025.