Most people get into Airtable by replacing a spreadsheet. Within a few weeks, they hit the same wall: how do I avoid retyping the same client name on every project record? How do I see all the invoices for one customer without filtering and re-filtering?
The answer is linked records — Airtable's relational core. Linked records turn separate tables into a connected database, and once you understand them, half of Airtable's power becomes available. Lookups and rollups, the two features that make linked records actually useful, depend entirely on this idea.
This guide explains what linked records really are, when to use them, and the field patterns that show up in nearly every well-designed base.
The Mental Model: Relationships, Not Repeats
A spreadsheet stores data in rows. When you need to associate a row with another piece of information — the client a project belongs to, the items in an order — your only tool is to retype the data on every row. Client Acme Corp on row 1, row 4, row 7, and row 12. Now imagine Acme rebrands to Acme Industries. You're hunting through every row to update the name.
Linked records fix this by storing the relationship instead of the data. You have one Clients table with one row for Acme. You have a Projects table where each project's Client field points at the Acme row. If Acme renames itself, you change it in one place and every project that links to Acme sees the new name immediately.
This isn't unique to Airtable — every real database works this way. What Airtable does is make it visual: click a cell, search for the record you want to link, done. No SQL, no foreign keys, no joins.
For Airtable's official primer, see the Linked Records documentation.
Creating Your First Linked Record
The flow takes about ten seconds.
- Open the table where you want the link to appear (e.g. Projects).
- Add a new field. Choose the Link to another record field type.
- Pick the table you want to link to (e.g. Clients).
- Decide whether to allow one or many linked records per cell.
- Save the field.
A reverse link automatically appears on the other table. Open Clients and you'll see a new field showing which Projects link to each client. You didn't have to create it — Airtable created it because every linked field is bidirectional by definition.
When you click into the linked cell, Airtable shows a search popup. Type a few letters of the record's primary field, pick the match, and the link is in place.
One-to-Many vs Many-to-Many
Two relationship shapes show up in almost every business base.
One-to-many. One record on one side links to many on the other. One Client has many Projects. One Project has many Tasks. One Order has many Line Items. The "many" side has a linked field that allows one record (restricted), the "one" side has a linked field that allows many records (the default).
Many-to-many. Records on both sides link to many on the other side. One Student attends many Classes; one Class has many Students. One Tag is on many Articles; one Article has many Tags. Use a linked field on each side, both allowing multiple records. Airtable handles the underlying relationship automatically.
If you need to store data about the relationship itself — the date a student enrolled in a class, the role a contact plays at a company — introduce a third table that sits between the two. This is called a junction table. Each row in the junction has one linked field to the left table and one to the right. The relationship-specific fields (Enrollment Date, Role) live on the junction record.
Lookup Fields: Showing Linked Data Inline
A linked record field shows only the primary field of the linked record — the client's name, the project's title. Most of the time you want to see more than that without leaving the table.
Lookup fields solve this. A lookup pulls a specific field from the linked record and displays it on the linking record.
Example. On the Projects table, you've linked each project to a Client. Now you add a Lookup field called Client Industry and configure it to look up the Industry field from the linked Clients record. Open any project and you can see at a glance which industry its client is in — without opening the Clients table.
Lookups update automatically. Change a client's industry in the Clients table, and every project linked to that client immediately shows the new industry.
A few rules to keep in mind:
- Lookups are read-only. You can't edit a lookup field directly. To change the value, edit the linked record.
- Lookups across multiple links show all values. If a Project links to three Clients, the lookup shows all three industries comma-separated.
- You can lookup through lookups. If a Project links to a Client, and Client has a lookup of
Account Managerfrom a Users table, the Project can have its own lookup that goes Client → Client's Account Manager. Two hops, no problem.
Rollup Fields: Aggregating Across Linked Records
Lookups show values. Rollups calculate over them.
A rollup field aggregates values across all records linked through a specific linked field. The most common aggregations are SUM, COUNT, AVERAGE, MIN, MAX, and CONCATENATE, but Airtable supports more.
Example. On the Clients table, each client links to many Invoices. You want to see each client's total revenue. Add a Rollup field called Total Revenue that rolls up the Amount field from linked Invoices using the SUM aggregation. Now every client row shows the sum of all their linked invoice amounts. Add a new invoice and the rollup updates automatically.
Rollups can include conditions. Configure the rollup with a filter like Status = 'Paid' and it only sums invoices that match — Lifetime Paid Revenue becomes a one-field calculation instead of a separate table.
Common rollup patterns:
| Source | Aggregation | Result |
|---|---|---|
| Linked Invoices, field = Amount, condition = Status is Paid | SUM | Total revenue |
| Linked Tasks, field = ID | COUNT | Number of tasks |
| Linked Reviews, field = Rating | AVERAGE | Average rating |
| Linked Project Phases, field = Due Date | MAX | Final project deadline |
| Linked Tags, field = Name | ARRAYJOIN, ", " | Comma-separated tag list |
Linked Records vs Multi-Select: Pick the Right Tool
A common confusion. Both linked records and multi-select fields can hold multiple values. When do you use which?
Use a multi-select when the options are a fixed, short list that doesn't need its own data — Priority (Low/Medium/High), Category (Marketing/Sales/Support), Status tags. The values are just labels.
Use a linked record when each option could become a record with its own fields — Tags that have descriptions, Categories that have owners, People who have email addresses. Anything that you'd ever want to know more about beyond just the name belongs in its own table.
Rule of thumb: if you ever find yourself wanting to add a description to a multi-select option, you should have used a linked record.
Patterns You'll See in Real Bases
A few schemas show up over and over in client projects.
CRM Pattern
Companies (one) ←→ (many) Contacts (one) ←→ (many) Deals
↓ (many)
Activities
A Company has many Contacts. Each Contact belongs to one Company. Each Contact can have many Deals. Each Deal has many Activities (calls, emails, meetings). Add rollups on Company for total open deal value and on Contact for last activity date.
Inventory Pattern
Products (one) ←→ (many) Stock Movements
↓ (many)
Warehouses
Each Product has many Stock Movements (in or out). Each Stock Movement links to a Warehouse. Add a rollup on Product that sums the quantity from Stock Movements to get current stock by warehouse.
Project Management Pattern
Clients (one) ←→ (many) Projects (one) ←→ (many) Tasks (many) ←→ (many) Team Members
Standard agency setup. Clients have Projects. Projects have Tasks. Each Task can be assigned to many Team Members. Rollups on Project for task completion percentage, on Client for total project hours.
For a complete CRM build using this pattern, see How to build a CRM in Airtable.
Performance Considerations
Linked records are powerful but they're not free. A few things to watch as your base grows.
- Avoid linking thousands of records into a single field. A
Customerslinked field on an Event with 10,000 customers will make the Event record slow to open. Move the relationship the other way — on each Customer, link to the Event. - Rollups across large linked sets are expensive. A rollup that sums 50,000 invoices per client recalculates on every change. Limit with a filter (only invoices from this year) or use a script to write the value periodically instead.
- Don't link across bases when you can avoid it. Sync tables work but they're slow and have limitations. Keep related tables in the same base when possible.
- Primary fields are what you see in linked cells — make them descriptive. A primary field of "Contact 1," "Contact 2" makes linked cells useless. Use a formula primary field that concatenates Name + Company instead.
Common Mistakes
Three patterns we see on most rescue projects.
Treating linked record fields like text fields when importing. When you import a CSV with client names and try to map them to a linked field, the import creates duplicate Clients records every time the spelling differs. Always clean the source data, or import in two passes — first the Clients table, then the linking records with linked fields matched by name.
Building everything in one table. Adding columns is fast, so the temptation is to keep adding. But Project Name, Client Name, Client Email, Client Phone, Client Industry, Account Manager Name, Account Manager Email — those last six belong on Clients and Users tables. Split early.
Not using junction tables when you need them. A many-to-many relationship that needs to store metadata (when did they enroll, what role do they play) requires a junction table. Without one, you have nowhere to put that data and you'll end up duplicating things or losing information.
Where to Go Next
Linked records, lookups, and rollups are the three concepts that unlock Airtable. Once you've built one base around them, the rest of the platform — interfaces, automations, scripting — becomes much easier because everything else assumes you understand relational data.
For practical applications, the CRM build guide and the project management tutorial both show linked records used in production. The migration guide from spreadsheets covers how to restructure flat data into a relational model.
Airtable's official Linked Records documentation covers every edge case in detail and is worth a read when you're modeling a complex relationship for the first time.
