You are about to point a sync at an Airtable table with twenty-something fields in it. Some of them you built two years ago and have not thought about since. Before you commit, you want to know which of those columns will land in Google Sheets in a state you can build a client report on.
Most of them arrive fine. About five do not. The problem is that the five that do not look exactly like the ones that do. A rollup that lands in the cell as 48,200 sits there looking every bit as solid as a number someone typed by hand. It is not the same thing. It is a photograph of a number, taken at the moment of the sync.
This is a pass through the whole field list, one type at a time. For each: what you actually see in the cell, what you can safely sort, filter and pivot on, and what will catch you out three weeks later when a client questions a total.
The short answer
Every Airtable field falls into one of three groups once it reaches a spreadsheet.
Stored fields are things a person typed or picked: text, number, currency, single select, checkbox, date, collaborator. They arrive as themselves. Sort them, filter them, pivot on them, total them.
Computed fields are things Airtable works out: formula, rollup, lookup, count. They arrive as an answer with no working behind it, correct at the moment it was fetched, and nothing in the spreadsheet will update it again except another sync.
Container fields hold more than one thing at a time: multiple select, linked record, attachment. A spreadsheet cell holds one thing, so they arrive squashed into a comma-joined string. Read them freely. Do not filter or pivot on them until you have split them apart.
This article exists because the cell gives you no clue which group you are looking at.

What every field type looks like in the cell
Formatting varies a little between tools. What does not vary is the shape of the value and what you are allowed to do with it.
| Airtable field | What lands in the cell | Analyse it? | Watch for |
|---|---|---|---|
| Single line text | The text, unchanged | Yes | Nothing |
| Long text | The text, line breaks and all, inside one cell | Read only | Paragraph breaks live inside the cell |
| Long text, rich text on | The text with its formatting marks showing | Read only | Bold and bullets arrive as stray symbols |
| Number | The number | Yes | Decimal places are formatting, not the value |
| Currency | The number, with a currency format applied | Yes | The symbol is formatting. The value is a bare number |
| Percent | A decimal. 25% arrives as 0.25 | Yes | Do not multiply by 100 a second time |
| Duration | A h:mm:ss string, or a count of seconds | Careful | Not something Sheets can total unaided |
| Rating | A number, or a row of stars | Careful | If it arrives as stars, you cannot average it |
| Checkbox | TRUE or FALSE | Yes | Blank and FALSE both mean unticked |
| Single select | The option name | Yes | The option colour does not travel |
| Multiple select | Every chosen option in one cell, comma joined | Split it first | Filtering on it matches partial words |
| Date | The date | Yes | Check the timezone before you trust it |
| Date and time | The date and the time, in UTC | Careful | This is the off-by-a-day bug |
| Created time | A timestamp, in UTC | Yes, for order | Same timezone question as any date |
| Last modified time | A timestamp, in UTC | Yes, for order | It moves for any edit by anyone, including a sync |
| Collaborator | The person's name and email | Yes | Several collaborators comma-join into one cell |
| Attachment | The filename, then a link | Filename only | The link stops working within hours |
| Linked record | A comma-joined list of linked record names | Read only | The relationship itself does not travel |
| Lookup | The values pulled from the other table, comma joined | Careful | Frozen, and often several values per cell |
| Rollup and count | A single number | Careful | Frozen at sync time. Looks live. Is not |
| Formula | The result | Yes, if plain | The formula does not travel. Nothing recalculates |
| Barcode | The code, as text | Yes | Leading zeros can vanish |
| Button | A clickable link | No | Nothing there to analyse |
| Autonumber | A whole number | As an ID, yes | It is creation order, not a quantity |
| Airtable record ID | A short code beginning with rec | As a join key | Never sum it, sort it, or chart it |
The rest of this article is the parts of that table that cost people money.

One record, read from the Airtable API and the synced sheet on the same run, 23 July 2026.
The text columns that break formulas
Long text keeps its line breaks. A field holding three paragraphs arrives as one cell holding three paragraphs. Nothing is lost, and the row is now four lines tall. If you later export that sheet to CSV for another system, those breaks inside the quoted cell are what break the import. If something downstream parses the file, add a formula field in Airtable that strips the breaks and sync that instead.
Rich text arrives with its markup visible. Switching on rich text formatting means the field no longer stores plain text. It stores text plus formatting marks, so a bolded phrase comes through wrapped in asterisks and a bulleted list comes through as lines beginning with dashes. It reads fine and it looks unpolished in a client-facing tab.
Multiple select is a trap dressed as a column. A record tagged Retail, Wholesale and Online arrives as one cell reading Retail, Wholesale, Online. Everything you now want to do with it is wrong. A pivot treats that string as one category, so you get a row per combination of tags rather than a row per tag. A filter for "Retail" also matches "Retail Partners". A COUNTIF for one tag misses every record where it is not the only one.
Split the column before you use it, with SPLIT or a QUERY on a helper tab, and pivot the split result. If a report is really built around those tags, the cheaper fix is upstream: a checkbox in Airtable per tag that matters, pivoted on directly.
Numbers that are not the number you saw
Percent arrives as a decimal. Airtable stores percentages the way spreadsheets do, so 100% is stored as 1 and 25% is stored as 0.25. Whether the cell shows 25% or 0.25 depends only on the number format applied on the way in. Every report that comes out four orders of magnitude too small has divided by 100 one time too many.
Currency arrives as a bare number. The dollar sign and the decimal places are display settings on both sides. That is good, because the column sums correctly. It is bad if two tables use different currencies and you total them together, because nothing in either cell says which currency it is.
Duration and rating stop being numbers. Airtable holds a duration as a count of seconds and a rating as a small whole number, and both often arrive formatted for human eyes: 1:30:00, or a row of filled and empty stars. Formatted that way they sort and read fine and average not at all. To total time tracked or average a rating, add a formula field in Airtable that outputs the plain number and sync that column.
Count and autonumber are both whole numbers and are not alike. A count is a real measure you can sum. Autonumber is an identifier that happens to be numeric, and averaging autonumbers gives you a number that means nothing.
