Almost every HR team we speak to says the same thing: they would like to stop running the department out of spreadsheets. A few months later, a new one has appeared. It is usually called something like headcount_final_v3, it lives in one person’s drive, and by now three other people depend on it.
This is worth understanding properly, because the usual explanation is wrong. HR teams do not reach for spreadsheets out of habit or because they dislike the systems they were given. They reach for spreadsheets because the systems they were given cannot answer the question in front of them.
The spreadsheet is a symptom, not the problem
Think about what actually happens. Someone in finance asks what the salary cost of the engineering team will be after the promotions that take effect next quarter. The HR system holds current salaries. The promotions live in a review cycle that has not been actioned yet. The team structure sits in an org chart that was last updated when two people left.
No single screen can answer that question, so somebody exports three reports, pastes them side by side, cleans up the mismatched names, then works out the number by hand. At that moment a spreadsheet is the most reasonable tool available. The trouble starts afterwards, because that file does not get deleted. It becomes the place where the answer lives, and the next person who needs a similar answer starts from it rather than from the system.
Why the copies drift apart
Once a number exists in two places, those two places begin to disagree. An employee changes their bank details in the HR system, but the payroll sheet still carries the old ones. Someone’s last working day moves by a week, yet the leavers tab was built in advance and nobody goes back to it.
Each individual drift is small. Together they produce the situation most HR leaders recognise, where two reports from the same department give different headcounts and a meeting gets spent deciding which one to believe. The cost is not really the time spent reconciling, although that is significant. The bigger cost is that people quietly stop trusting the reports, so decisions get made on instinct instead.
The pattern behind it
If you look at where the spreadsheets cluster, they almost always sit at the seams between systems. Recruitment does not talk to onboarding, so somebody tracks the handover manually. Attendance does not talk to payroll, so somebody builds a monthly sheet to bridge them. Performance does not talk to compensation, so promotion decisions get modelled outside both.
Every seam produces a spreadsheet, because a seam is precisely the place where the software stops being able to help. Adding another specialised tool rarely fixes this. It usually adds two more seams.
What actually reduces them
The honest answer is that you do not remove spreadsheets by banning them. You remove them by removing the reasons people open them. In practice that means a few specific things.
First, the employee record needs to be one record rather than several that get synced. If hiring, payroll, attendance and performance all read from and write to the same record, a promotion updates pay without anyone retyping it, because there is nothing to retype.
Second, the questions people actually ask need to be answerable inside the system. Cost per hire, attrition by team, salary cost after planned changes: if these require an export, an export is what you will get, along with the file that outlives it.
Third, permissions have to be part of the record rather than a property of the file. Much of the reason sensitive data ends up in shared drives is that the system made sharing hard, so someone worked around it. A record that knows who is allowed to see what removes that incentive.
A reasonable expectation
None of this means spreadsheets disappear entirely, and you should be suspicious of anyone who promises that. Modelling a restructure or sketching a compensation scenario is legitimate work, and a spreadsheet is a good tool for it.
What should disappear is the category of spreadsheet that exists only because two systems could not speak to each other. Those are the ones that quietly become a source of truth nobody chose, and they are the ones worth designing out.