Excel's row limit: when spreadsheets stop keeping up

In October 2020, the British government "lost" 15,841 positive COVID-19 cases because of a spreadsheet. The tracing system consolidated lab results into an .xls file, a format that holds at most 65,536 rows. Everything beyond that was silently discarded. Nobody deleted anything: Excel simply stopped counting.

If a government with a multi-billion budget tripped over a spreadsheet limit, it's worth asking: what's happening inside your own operation, where sales, paid media, and finance live in tabs that only grow?

Excel's spreadsheet limit is 1,048,576 rows and 16,384 columns per worksheet, according to Microsoft. In practice, though, the operational limit arrives much earlier: with a few tens of thousands of formula-heavy rows, the file freezes, recalculation drags, and errors multiply.

What is Excel's row limit?

The official numbers are on Microsoft's own Excel specifications and limits page. The essentials:

Format / tool	Rows	Columns
Excel .xls (up to 2003)	65,536	256
Excel .xlsx (2007 onward)	1,048,576	16,384
Google Sheets	10 million cells per file (across all tabs)

Other limits that show up in daily work: 32,767 characters per cell and, on 32-bit versions, a 2 GB memory ceiling for the entire program. The limit applies per worksheet, not per file. Duplicating giant tabs, however, only makes things worse.

Here's the detail almost everyone misses: the British incident happened because someone saved in the old format. The file didn't throw an error, didn't crash, didn't warn. That's the most dangerous trait of spreadsheet limits: they fail silently.

The real limit arrives long before 1 million rows

Anyone who has run a VLOOKUP against a 200,000-row table knows it: Excel doesn't need to get anywhere near 1,048,576 rows to become unusable. What kills performance isn't raw volume. It's volume combined with lookup formulas, conditional formatting sprayed everywhere, and volatile functions like TODAY() and INDIRECT() that recalculate on every change.

A common scene: a distributor exports its order history from the ERP into a 180,000-row tab and matches everything against the customer table with VLOOKUP. Saving the file takes three minutes. Opening it, another four. At some point somebody turns off automatic recalculation just to get work done. From then on, nobody knows whether the numbers on screen are current.

In practice, the problem is rarely Excel. It's using Excel as a database, a role it was never designed for.

7 signs your company has outgrown the spreadsheet

The limit that matters isn't technical, it's operational. These are the signs that show up most often in companies that have outgrown the model:

1. Opening or saving the file takes more than 30 seconds. It sounds trivial, but multiplied across the whole team, every day, it's paid hours thrown away.
2. Version control lives in the file name. If a "report_final_v7_THIS_ONE.xlsx" is circulating by email, nobody knows which number is official.
3. Only one person understands the spreadsheet. When the author goes on vacation, reporting stops. That has a name: concentrated operational risk.
4. Monthly consolidation is manual. Export from three systems, paste, fix columns, double-check: a ritual that eats days and produces a snapshot that's already stale.
5. Every department presents a different number. Marketing says 412 leads; the CRM shows 396. Meetings become debates about which spreadsheet is right, the classic case of data silos.
6. Copying and pasting between systems became someone's job. There's an informal role of "the person who updates the spreadsheet". That work is exactly what an ETL process automates.
7. Decisions wait for the spreadsheet to be ready. If the answer to "how much did we sell this week?" is "I'll tell you Monday", the data exists, but it's not available when the decision happens.

Three or more signs from this list already indicate the spreadsheet has stopped being an analysis tool and become a bottleneck.

How much does running everything on spreadsheets cost?

The cost has two layers. The first is time: an analyst who spends 90 minutes a day consolidating data racks up 33 hours a month. At a fully loaded cost of $25 an hour, that's $825 a month (nearly $10,000 a year) paid to produce a report that's born outdated. In an agency running five accounts this way, the rework pays for another analyst's salary.

The second layer is error. Research by Raymond Panko at the University of Hawaii, the reference on this subject for decades, indicates that about 88% of spreadsheets contain errors. Not out of incompetence: in any human cognitive work, an error rate of 1% to 5% per cell is normal. In a spreadsheet with thousands of chained formulas, the odds of a clean final result are low. And a formula error never announces itself.

That math changes the question. The debate stops being "how much does a better tool cost?" and becomes "how much am I already paying for the current one?". For anyone tracking financial indicators closely, spreadsheet rework is usually a significant invisible cost.

Excel, Google Sheets, a database, or BI: when to use each?

There's no single answer; there's fit between volume and use. An honest comparison:

Tool	Works well for	Breaks when
Excel	Modeling, simulations, ad hoc analysis	It becomes the official record for data that grows daily
Google Sheets	Real-time collaboration, small datasets	It passes a few hundred thousand active cells
Database (SQL)	Storing millions of records with integrity	Nobody on the team can query it without IT
BI / dashboards	Tracking recurring metrics	The question changes weekly and the dashboard can't keep up
AI connected to your data	Asking in plain language across sources	The underlying data is too disorganized

Before migrating, it's worth exhausting what Excel itself offers: Power Query automates imports and cleanup, and the data model (Power Pivot) processes millions of records without dumping them onto the grid. For many companies, that buys one or two extra years. But notice the pattern: those tools push Excel closer to being a database, a clear sign of where the need is pointing. The next step usually involves integrating data across systems and getting the raw load out of the spreadsheet.

Migrating doesn't mean abandoning Excel

Here's an opinion unpopular among software vendors: the spreadsheet remains one of the best tools ever built for modeling scenarios, drafting budgets, and testing quick hypotheses. Flexibility is its strength. No company needs to "kill Excel".

That logic breaks at one specific point: when the spreadsheet becomes the official system of record, the place where the truth about sales, leads, or cash lives. Transactional data that grows every day belongs in a database; the spreadsheet can live on as an analysis interface, fed automatically. The mature shift isn't about tools, it's about roles: the spreadsheet moves from "where data lives" to "where data gets explored".

Frequently asked questions

How many rows does Excel support?

Since the 2007 version (.xlsx format), each worksheet supports 1,048,576 rows and 16,384 columns. The old .xls format stops at 65,536 rows and cuts off the excess without warning, which is why it's worth avoiding even for small files.

How do I open a file with more than 1 million rows in Excel?

Through the grid, you can't: the excess gets discarded. The path is importing via Power Query into the data model, which processes millions of records and delivers summarized results through pivot tables. For recurring use, a database is the definitive answer.

Can I increase or remove Excel's row limit?

No. The grid limit is fixed in every version and no setting changes it. What you can do is work around it: the data model inside Excel itself, or migrating the dataset to a SQL database.

Does Google Sheets handle more data than Excel?

No. Sheets caps each file at 10 million cells across all tabs (far less than a single full Excel worksheet) and slows down well before that. Sheets' strength is real-time collaboration, not volume.

When should I move from spreadsheets to a database?

When the operational signs appear: recurring manual consolidation, conflicting versions of the same number, sluggish files, and dependence on a single person. If the spreadsheet is the official record for data that grows daily, the migration is already overdue.

The spreadsheet as interface, not as database

The Excel limit that actually stalls companies is almost never the 1,048,576 in the documentation. It's the operational limit of treating an analysis tool as data infrastructure. Companies that scale well make one simple trade: raw data goes somewhere built for volume, and people keep the part where they're irreplaceable, the interpretation.

That's the trade Sherlok shortcuts: it connects Meta Ads, Google Ads, GA4, CRMs, spreadsheets, and SQL databases in one place, and you ask in plain language, like "which campaign had the best cost per lead this month?", and get the analysis ready, with no VLOOKUP, no manual consolidation, and no waiting until Monday.