Airtable Builder Certification cheat sheet
Airtable Builder Certification · Keep the key ideas within reach.
Airtable Platform and App Building
A base is the top-level container; it holds multiple tables, each table holds records (rows), and each record holds fields (columns).
Field types constrain data entry: single line text, number with precision and formatting, date with timezone and format, single select, multiple select, checkbox, attachment, linked record, lookup, rollup, formula, and autonumber.
Single select allows one choice per record; multiple select allows many. Both can be color-coded and can store new options created during entry.
A primary field is the first field of each table and is used to label linked records; it can be renamed but not deleted or hidden while it is the primary field.
Views display the same underlying records with different configuration: Grid, Form, Calendar, Kanban, Gallery, Timeline, Gantt, and List. Views never duplicate data.
Grid view supports filters, sorts, groups, field visibility, row height, and color rules; these settings are per-view, not per-table.
Forms create records only; each form exposes a chosen subset of fields and can redirect submitters or require a specific view for submission.
Collaborator permissions are set per base (owner, editor, commenter, read-only) and can be refined per table or view with share links.
Field dependencies matter: a formula, lookup, or rollup field cannot reference itself, and deleting a source field breaks dependent fields.
Databases and Connected Data
A linked record field stores references to records in another table (or the same table for self-links) and can allow linking to multiple records.
Link cardinality is controlled by 'Allow linking to multiple records' on each side; a one-to-many link has multi-link on the many side and single-link on the one side.
Lookup fields pull a specific field's value from linked records; if multiple records are linked, the lookup returns an array of values.
Rollup fields aggregate values across linked records using functions such as SUM, COUNT, COUNTA, AVERAGE, MIN, MAX, ARRAYJOIN, and ARRAYUNIQUE.
COUNT counts numeric values only; COUNTA counts non-empty values of any type. Use COUNTA to count linked records.
ARRAYJOIN(values, separator) flattens a lookup array into text; ARRAYUNIQUE removes duplicates before aggregation.
Formulas run per record and can reference other fields in the same record, including lookup and rollup fields, but cannot reference fields in other tables directly.
A junction (join) table resolves many-to-many relationships by holding two linked record fields, one to each side.
Circular linking and rollups that reference their own table can cause recursion errors; break the cycle with an intermediate field or table.
Interfaces
Interfaces are drag-and-drop layouts built on top of base data; they do not store data themselves and reflect the underlying tables live.
Interface pages can be set to a record review layout, a dashboard, a form, or a blank canvas, and pages can be grouped and reordered.
Element types include grid, list, record detail, chart, button, text, image, and form; each element binds to a table or view.
Interface visibility can be restricted by user, by a condition on field values, or by a filter, so different viewers see different records.
Record review layouts let users step through records and edit fields inline, with configurable field order and read-only fields.
Buttons in interfaces can open a URL, open a form, or trigger an update or automation, enabling guided data entry.
Interface filters and element-level filters are independent of grid view filters; changing one does not change the other.
Publishing an interface requires setting who can access it (base collaborators, specific users, or a share link) and whether editing is allowed.
A chart element requires a source table or view plus a category field and an aggregation (count, sum, average) to render.
Automations
An automation has a trigger (record created, record updated, record matches conditions, scheduled time, form submitted, or webhook) followed by one or more actions.
'When record matches conditions' fires once per record when it first meets the condition and does not re-fire while the record continues to match.
'When record updated' can watch specific fields; leaving the field list empty triggers on any change to the record.
Conditional logic uses 'Conditional' action groups with if/else branches, and 'Repeat' groups iterate over a list of linked records or an array.
Actions include create record, update record, find records, delete record, send email, send Slack message, run script, and call a webhook.
'Find records' returns a list; downstream actions must reference the list or use a repeat group to act on each result.
A scheduled trigger runs at a fixed interval (hourly, daily, weekly, or custom) and evaluates all records matching its conditions at run time.
Automation run history shows each run's status (success, failed, or skipped) and the input/output of each step for debugging.
Automations run under the permissions of the automation's owner; if the owner loses access to a field or table, the step can fail.