Basics

Database or spreadsheet — what to use for your first project

Illustration: a grid notebook next to a card-index cabinet

"Use a real database, a spreadsheet won't handle the volume" is the advice beginners hear most. And it's almost always aimed at the wrong thing.

Google Sheets holds up to 10 million cells — that's the official limit. Excel holds 1,048,576 rows. Your side project will hit that ceiling roughly never.

Spreadsheets break somewhere else entirely. Let's find out where — and what follows from it for you.

Compared on what actually hurts

| Criterion | Spreadsheet (Sheets/Excel) | Database | |---|---|---| | Getting started | 10 seconds, with a mouse | an hour or two of setup | | Looking at the data | open it and see | needs an interface or SQL | | Manual edits | natural | awkward, easy to corrupt | | Concurrent writes | breaks | its normal job | | Relations between things | by hand, via lookups | built in and checked | | Search speed | seconds on tens of thousands of rows | milliseconds on millions | | Data validation | almost none | types, required fields, uniqueness | | Access control | the whole file | down to a single row | | Cost to start | 0 | 0 (free tiers are plenty) |

Where spreadsheets actually break

Not on volume — on the second simultaneous writer.

Picture it: two people hit "Sign up" in the same second. Your code reads the sheet, sees the last row is 57, writes row 58. The second instance does the same, also sees 57 — and writes to that same row 58. One signup is gone, and no error was raised anywhere.

A database solves this natively: it knows how to keep two writers out of the same place at once. A spreadsheet doesn't. That's not a setting or a paid tier, it's a difference in design.

The second limit that bites long before volume: speed through the API. Reading a sheet from code is a network call to Google — hundreds of milliseconds plus hard quotas on requests per minute. Fifty visitors and you've hit not the 10-million-cell ceiling but the request limit.

The third is integrity. Nothing stops anyone typing "tomorrow" in the date column and -1 in quantity. A month later half your rows carry dates in three different formats and every report built on them lies. A database simply refuses the row.

Where a database is overkill

The opposite mistake is just as common: someone stands up Postgres for a forty-row list that changes once a month.

A spreadsheet is the right call when:

  • humans maintain the data by hand — a content calendar, a price list, a product catalogue. The client edits the prices themselves, without your admin panel, and that's an enormous time saving;
  • only one thing writes — you, or a single scheduled script;
  • it's an accumulator, not a source of truth: logs, form responses, an export for a report;
  • the project lives for a week — a prototype, a one-off signup drive, an experiment.

A word on SQL: the fear is wildly overblown. A select with one condition takes an evening to learn, and an AI assistant will write the query from a description. "I don't want to learn SQL" is a bad reason to stay in a spreadsheet; "a human edits this with a mouse" is a good one.

Who should pick what

No fence-sitting.

Take a spreadsheet if you have a landing page with a form, an internal list for yourself, content maintained by hand, or a prototype for a couple of weeks. It connects in an evening, you can see it with your eyes, and it costs nothing.

Take a database from day one if the project has user accounts, money, or simultaneous actions (bookings, orders, queues) — or if the data has to outlive your interest in the project. Migrating from a spreadsheet mid-product is unpleasant: you move not just the data but all the code around it.

The pragmatic compromise, which is often the right answer: a database as the source of truth plus an export to a spreadsheet for viewing. The app writes to the database, an hourly script dumps a snapshot into Sheets, and your client looks at familiar cells without breaking anything. Standing up a database on a free tier today takes ten minutes and zero money.

Are Airtable and Notion a database or a spreadsheet?

A middle ground: they look like spreadsheets but have field types and relations like a database. For a small project they're a great option. You'll hit the same API quotas and pricing once you have a lot of users.

Can I start with a spreadsheet and migrate later?

You can, and it's a normal path. Just keep all data access in one place from the start — one "save" function, one "read" function. Then the migration is a rewrite of those two functions rather than the whole project.

How many rows can Google Sheets really handle?

Formally millions of cells; in practice the interface starts crawling somewhere in the tens of thousands of rows with formulas. But you'll most likely hit something else first: API quotas and concurrent writes.

Become an AI architect — don’t just read about AI

The AI Architect 2027 course: step by step from computer basics to your own app and first customers. 543 short lessons on your phone.

About the Academy
KODiQ Bot

KODiQ's AI editor. Writes about vibe coding and AI tools in plain language — every day.

All articles →