Airtable vs Notion vs Google Sheets (2026): the ops database your automation can reach free

The short answer: stay on Google Sheets. For most Singapore SMEs it's the right operations database, and it costs nothing.

Here's why. The tool that runs your automations (Make, on this site) doesn't care how pretty the database is. It cares whether it can read and write the thing cheaply, every hour, for years. Sheets lets it do that on a free Google account, with an API that costs nothing extra and has no monthly call cap. Airtable's free tier stops you at 1,000 records per base, 100 automation runs a month and 1,000 API calls a month. That last number is the one a polling automation chews through in about ten days. Notion sits in between: a good database if your team already lives in Notion, a strange one if it doesn't.

Pay for Airtable when you need forms feeding linked tables, per-person views and an interface staff can't break, and you're fine with US$24 a seat a month. Pay for Notion when the ops data has to sit next to your wiki and your tasks. Everyone else: a Sheet with a Form in front, and Make polling it hourly. Boring, and it works.

At a glance (checked 15 Sep 2026)

Prices in US dollars unless stated. Google Sheets on a personal Google account is S$0. Terms change; click through before you budget.

What you're actually buying

Google Sheets: a free grid with a free API

Sheets is the database you already have. Any Google account gets it, and Google's limits page puts the ceiling at "Up to 10 million cells or 18,278 columns" per spreadsheet. An SME order log with 20 columns can hold hundreds of thousands of rows before that matters, and honestly, if your Sheet is that big you have a nicer problem than tool choice.

The part that decides this comparison is the API. Google's own quota page says "All standard use of the Google Sheets API is available at no additional cost" and, as long as you stay inside the per-minute quotas (300 read requests a minute per project), "there's no limit to the number of requests that you can make per day." That's what Make talks to, and it's free on Google's side. One line to watch: the same page says "Exceeding the quota request limits is planned to incur charges to your Google Cloud billing account later in 2026." At SME polling rates you won't get near it.

What you don't get: real relations between tables (VLOOKUP is not a foreign key), or per-row permissions. A Google Form in front of a Sheet is still the fastest S$0 intake system in Singapore, and the lead-to-Telegram playbook is built on exactly that.

Airtable: a real database, with meters on the free tier

Airtable is a proper relational database wearing a spreadsheet costume. Linked records, views per person, forms that write into the right table, and an Interface Designer so staff see a tidy app rather than your raw tables. If you've outgrown Sheets, this is what "outgrown" usually means.

The free tier is where the comparison turns. Airtable's plans page gives Free "1000 records per base", "1000 API calls per workspace per month", and a workspace "limited to 5 collaborators with Editor or Creator permissions". Automations get "100" runs a month, and Airtable "counts an automation 'run' each time a trigger is invoked", so failed runs count too. A hundred runs is about three a day. That's a hobby, not an operation.

Paid fixes all of it, at a seat price. Team is "$24 per collaborator/ per month billed monthly or $20 per collaborator/ per month billed annually" for "50000 records per base", "100000 API calls per workspace per month" and 25,000 automation runs. Business is "$54 per collaborator/ per month billed monthly or $45 per collaborator/ per month billed annually" with 125,000 records and unlimited API calls. Note the word collaborator: Airtable's pricing page says plans "are charged per seat" and that "No charges will apply on a Team or Business plan for read-only collaborators, form submissions, or share links." So five people editing is US$120 a month billed monthly, but twenty people looking is free. Design your permissions around that and the bill halves.

Notion: a wiki that grew a database

Notion's databases are genuinely good now: relations, rollups, and buttons that do things. The catch is that Notion is a workspace first and a database second. If your team already writes its SOPs and task boards in Notion, putting the ops data there too is sensible. If your team lives in WhatsApp and Gmail, you'd be adopting a whole way of working to get a table.

Notion's pricing page prices in USD: Free at "$0per member / month", Plus at "$10per member / month", Business at "$20per member / month", Enterprise "Custom pricing". Those figures sit under a "Pay monthly" / "Pay yearly" switch that says "Save up to 20% with yearly", and the page doesn't label which period the headline assumes. Flip it and read the month-to-month number before you budget. On Free, blocks are "Unlimited for individuals" but "limited for 2+ members", guests are capped at "10", and automations are "buttons only"; "Custom database automations" start on Plus. AI beyond the "Limited Trial" is "Free to try, then $10 per 1,000 monthly Notion credits", and the credits page says "Notion credits are an add-on for Business and Enterprise plans."

For the API, Notion's developer docs publish a rate limit of "180 requests per minute (an average of 3 per second)" on "All other plans". No monthly call meter appears on that page. On the thing that matters, that puts Notion closer to Sheets than to Airtable.

The Make maths (this is the bit that decides it)

Make charges by the step. Its pricing page says "Each module action in your scenario, like adding a Google Sheet row or fetching Gmail account data, counts as one credit." The part people miss is that checking counts. Make's operations page defines "an operation is a single module run to process data or check for new data", and its own example is "the google sheet > watch new rows module checks for new rows in a google spreadsheet = 1 operation (one per check)". The credits page adds that "by default, 1 operation equals 1 credit".

So do the sum for a polling trigger. Hourly is 24 checks a day, 720 a month, before the scenario does any actual work. Make Free gives "Up to 1,000 credits/mo" and a "15-minute minimum interval between runs". Run at that 15-minute floor and you're at 2,880 checks a month, nearly three times the free allowance. On Free, poll hourly or slower. That's true for all three databases, so it isn't the tie-breaker.

Here's the tie-breaker: the same hourly poll on Airtable Free is 720 of your "1000 API calls per workspace per month". Every 15 minutes is 2,880, and the cap is gone in about ten days. Sheets has no such meter. Notion publishes no such meter. Airtable's escape hatch is real but paid: "Watch Responses" is a webhook, so form submissions push into Make instantly with no polling, and it's the right trigger for intake. Make's docs list it as "Available on the Team plan and all higher Airtable tiers", because the form redirect it relies on is, in Airtable's words, a "Paid plan-only feature". On Free, you poll. And "Watch Records", the one you'd use for "a status changed", polls on every plan, and it needs a Created Time or Last Modified Time field in the table to work at all.

The paid Make tier changes the interval, not the arithmetic. The pricing page shows the Make Plan at "$9/mo" for "5k credits/mo" with a "Pay monthly" / "Pay annually" switch; read on 15 Sep 2026, billing period not confirmed, so treat that as the floor and read the monthly figure off the page yourself. If you'd rather not be metered by the step at all, the Make vs n8n piece covers the self-hosted route.

Singapore-specific notes

• Two of the three bill in US dollars. Airtable's prices are in "$" and Notion's page says "Price inUSD", so your bank sets the SGD figure and it moves. Sheets on a personal account is S$0. If you want Sheets under your own domain, Google Workspace's pricing page shows a different currency depending on where you open it, so read it from Singapore; the tier decision is its own article.
• GST depends on your registration status. Overseas software subscriptions can carry GST for a non-registered buyer, and a GST-registered business handles them differently. Check IRAS's guidance on buying remote services from overseas providers before you assume the sticker price is the invoice.
• Your data leaves Singapore in all three cases. Airtable stores data in the "United States (default)", with Europe and Australia on offer only when "Customers on Enterprise Scale plans" switch data residency on. Notion runs on AWS and backs up to "AWS US West-2 and AWS US East-2". Sheets sits wherever Google puts it. For customer names and phone numbers that's a PDPA transfer question to answer once, on paper, whichever you choose.
• None of this is grant money. PSG funds "sector-specific and generic solutions that are pre-approved by EnterpriseSG", "Up to 50% of eligible costs for local SMEs" to a cap of "$30,000", and "Only the actual purchase of the pre-approved solution is supported". A self-serve Airtable or Notion subscription isn't that shape. If a vendor tells you it is, check the markup and the grant reality.

Which one for which business

• Retail, F&B, services firm with orders, customers and stock in one place: Sheets. A Form for intake, a Sheet per year, Make polling hourly. The low-stock alert and receipt capture playbooks show the pattern.
• A team that already runs its wiki and tasks in Notion: Notion. Keep the ops database next to the SOPs, watch the "limited for 2+ members" block cap on Free, and go to Plus when you need "Custom database automations". Make reaches it without a monthly call meter.
• Real relational needs, staff who must see only their own rows, forms that write into linked tables: Airtable Team, budgeted at US$24 a collaborator billed monthly with viewers free. Trigger intake with "Watch Responses", not polling, and don't try to run this on the Free tier; the API cap will bite within the month.
• Anyone with a client-facing CRM need: that's not an ops database, it's a CRM, and under about 20 deals a Sheet is still the answer there too.

You might need neither if…

…your whole operation is under about 10,000 rows and fewer than five people edit it at once. That's a Google Sheet with a Form in front of it, at S$0, and Make Free polling it hourly for 720 of its 1,000 credits. You don't need a database product. You need column headers you don't change and a rule that nobody types into the raw tab. One sheet per year helps. Buy nothing. Revisit the day someone asks for "the customer's other orders" and you can't answer it with a filter.

Your first automation, step by step

1. Make a Google Form with the fields you actually need (name, phone, what they want). Google creates the Sheet for you.
2. Sign up for Make on Free. New scenario: Google Sheets "Watch New Rows", pointed at that Sheet.
3. Set the schedule to every 60 minutes, not the 15-minute floor. That's 720 credits a month of checking, inside the free 1,000.
4. Add a Telegram "Send a Text Message" module so the new row lands in your phone. Each new row costs one credit for the message.
5. Run it once by hand, submit a test form, check the message arrives. That's the whole build; the lead-to-Telegram playbook walks it click by click.
6. Only when you find yourself adding a second Sheet to answer questions about the first (which customer, which order) look at Airtable or Notion. Rebuild the same scenario on their trigger and check your credit and API counts after a week.

FAQ

Airtable's free tier is free too. Why not just start there? Because three limits arrive together: 1,000 records per base, 100 automation runs a month, and 1,000 API calls a month. The first one you hit is usually the API cap, and you hit it because Make was checking politely every 15 minutes. Start on Sheets, or start on Airtable knowing you'll pay US$24 a seat within a quarter.

Isn't a spreadsheet unprofessional as a database? Your customers never see it. Staff see a Form, a Telegram message and a filtered view. What's unprofessional is a database product nobody updates because it nags. The tool that gets filled in wins.

We're already on Notion for everything. Should ops data go there too? Probably, yes. One login and one search beats a nicer table somewhere else. Just budget Plus for two-plus members with real automations, read the monthly figure off the toggle, and keep the customer data in one database rather than scattered across pages.

Can I move later? All three export CSV. Column structure moves; relations, views and automations don't. Rebuild, don't import. It's a weekend for a small operation, which is why starting on the free grid costs you nothing even if you leave it.

What about the AI features? Airtable bundles AI credits per editor and Notion sells credits beyond a trial. Neither is the reason to pick a database. The AI you'll actually use in an ops workflow is a Make step reading the table, and it costs the same credit whichever database sits underneath.

Next step

Count your rows and your editors. Under 10,000 and five, build the Form-to-Sheet-to-Telegram scenario above this week and spend nothing. Over that, or the moment "which customer" needs a second table, price Airtable Team at the monthly rate with viewers free, or Notion Plus off the toggle, and rebuild the same scenario before you migrate anything. That's how most AI automation in Singapore starts: on a free grid, with a meter you understand.

A note on Notion's and Make's pricing pages: both show one figure per plan under a monthly/yearly switch, and neither labels which period the headline assumes. Notion's US$10 and US$20 and Make's US$9 were read on 15 Sep 2026 with the billing period not confirmed; the yearly rate is the cheaper one in both cases, so flip the switch and budget on the higher number. Airtable's figures come from its plans page, which states both periods.