Isn't some manual work in every business normal?

Sure. The test isn't whether a spreadsheet exists. In fact, a simple spreadsheet is often the best answer. The real test is whether that spreadsheet gets pulled the same way on a regular cadence, and whether there's a system that could be doing that work automatically. A one-off analysis in a spreadsheet is normal. A manual process that runs every week is an issue.

Our team says the system is too hard to use. Is that the same thing?

It's usually the same root. "Too hard to use" often means it was never configured for how they work, or nobody showed them the part that would have helped.

We just paid a lot for this ERP. Do we have to replace it?

Almost certainly not, and I'd push back hard on anyone telling you to. The spreadsheet next to it is evidence the implementation stopped early, which is a much cheaper thing to fix than a migration.

How do I raise this without it sounding like I'm blaming my team?

Start from the assumption they're right. They built the workaround because the system didn't answer their question, and they're the only people who can tell you what the question was.

What if the spreadsheet is better than the system?

Then it's telling you something specific about a gap, and that's useful. Find out what it does that the system doesn't, because that's the requirement nobody wrote down when the software got bought.

Why ERP Implementations Fail | Janaitis Business Systems

At one company I worked with, somebody hand-built a sales report from HubSpot data every week. It took a while. They started by filtering to the leads that came in during the past week, then copied over the number of leads in each category (form fill, MQL, SQL, opportunity, and so on). Along the way they did some manual transformations: excluding form fills that were clearly spam, adjusting revenue on deals that had changed since they were entered, and so on. The bones of the report were recorded in the CRM, but the person pulling it made a number of judgment calls along the way.

That ran every week so that somebody could answer a question on the Monday sales call. HubSpot was already paid for, and it was not a cheap subscription.

A shadow spreadsheet is a spreadsheet your team maintains next to a system the company bought, doing a job that system was supposed to do. It isn’t a discipline problem, at least not on the reporting side. It’s the cheapest available answer to a question the system can’t currently answer, and its existence tells you the software was purchased but never finished.

What was wrong

Not the people, and not HubSpot.

Nobody had ever agreed what a qualified lead was. That’s the whole root of it. There was no shared definition, so there was no field anyone could fill in honestly, so nobody on the sales team was updating the records. And because it wasn’t being updated, the only way to get the number was to reconstruct it afterward from raw exports.

Meanwhile the tool sat there able to do all of it. You can set up the pipeline so each stage carries a probability of closing, and read the value sitting in the pipeline in real time, any day you like. None of that was configured. They were using maybe half of it as a place to collect email addresses, and doing the categorization by hand in Excel, because nobody had taken the time to set the system up properly.

Two things follow from building the number that way, and the second one is worse. It’s labor intensive, which everybody notices. And it’s unreliable, which almost nobody notices. A borderline lead classified as qualified last week gets classified as unqualified this week, because somebody different pulled the report, or because the same person doesn’t remember how they called it last time. So the numbers don’t tie week over week, and there’s no way to tell whether a change in the funnel is a change in the business or a change in whoever had the spreadsheet open.

What a shadow spreadsheet tells you about the system

When I find one, it’s usually one of two things.

The tool isn’t configured to do what it can do. Somebody bought it, somebody ran an implementation, and the rules and features that would have answered the question were never switched on. This is the HubSpot case above.

Or the question spans two systems that don’t talk. You can pull the leads out of the CRM easily enough. But if deal size lives in your invoicing software and not in the CRM, and those two aren’t integrated, then any report tying the lead funnel to revenue has to be assembled by a person out of two exports. Connecting the two systems is the harder project, so people naturally start doing it themselves, and then keep doing it.

Either way the spreadsheet is a report on the software, not on the team.

Why does anyone keep doing it by hand?

It’s not because they’re lazy. It’s because it’s usually the most efficient alternative available to them, given their current constraints.

Shadow spreadsheets get born out of convenience. Usually the person knows there’s probably a better way. What they don’t have is any of the three things that would let them take it: the access or ability to change the system themselves, the knowledge that the feature even exists, or a reason to stop, since this is how it’s always been done and it works well enough to get through Monday.

None of that is a failure of will. Given what’s in front of them, the spreadsheet is the rational move.

The ten-minute analysis you can run today

You don’t need anybody’s help for this one.

Pick a recurring meeting where a number gets presented. Your weekly sales review is the easiest place to start.

Ask whoever brings that number two questions. Where does it come from, and how long does it take you to put together. Ask it plainly, not as an audit, because you won’t get an answer if everyone’s afraid of getting in trouble.

Listen for the word “export.” Or “pull,” or “I just drop it into my sheet.” Any of those means the system isn’t producing the number and a person is.

Ask one more: what happens when you don’t have time to do it. You’ll learn whether the meeting has ever run without it, and what that meeting was like.

Run that on three or four recurring numbers and you’ll have a count. Most owners are surprised by it, and the surprise is the point.

Follow the number downstream

This is the part that decides whether a shadow spreadsheet is an annoyance or a risk.

That spreadsheet gets reviewed on the weekly call, and the team gets coached against the KPIs in it. Then the summary numbers come off the bottom of it and go into a report for senior leadership, or get screenshotted straight onto a slide in the weekly leadership deck. So a hand-assembled, hand-classified figure becomes the official account of how the sales department is doing, one level up, where nobody present has any idea how it was made.

Now think about the week it doesn’t get filled in. Somebody got busy the day before, which is the most ordinary thing in the world. The team shows up to the review and there’s no report. As the owner, you have to take their word for how things are going, with nothing to check it against. And when it’s your turn to report upward, you’ve got no artifact to bring.

The business didn’t lose a spreadsheet that week. It lost its ability to verify anything about the sales department, and it lost it because one person had a busy Thursday.

How to fix it

Almost never by buying different software.

Sometimes it’s finishing the configuration that got skipped at implementation. Sometimes it’s connecting two systems so the question stops requiring a human courier. Often it starts before either, with agreeing what the term even means, because you can’t build a qualified-lead field until the room agrees what a qualified lead is. That order is the same one on the nine-constraint map: definitions first, connections second.

And part of it is the change management nobody budgeted for. Buying the technology and integrating the technology are two different projects, and training your people on what the new system can do for them is a third. Skip the second or the third and you get exactly this: an expensive system, a team that works around it, and a spreadsheet holding the business together.

We bought the ERP. The spreadsheets never left.