> ## Documentation Index
> Fetch the complete documentation index at: https://docsv4.mile.app/llms.txt
> Use this file to discover all available pages before exploring further.
# Create or Update Data Source
> Write a finished task's values into a data source record, creating it or updating the existing one.
**Create or Update Data Source** writes the values of a finished task into a [data source](/pages/workflow/data-source/introduction). If a record with the same primary key already exists, it is updated; otherwise a new record is created. Field work keeps your master data current: customer addresses, stock counts, service history, sales leads.
Available for: **On Task Finished** only, because it needs the task's complete data.
Required permission:
* View automation
* Create automation
## Automation detail
1. **Data Type**: required. The [data type](/pages/workflow/data-type/introduction) whose records are created or updated, for example **Store**.
2. **Field Mapping**: every field of that data type. For each one, type a fixed value or click the search icon to copy a field of the finished task, exactly as in [Create Task](/pages/workflow/automation/actions/create-task#filling-a-field). The data type's primary key field is marked **(primary key)**.
3. **Data Source Access**: shown only for data types whose records can be limited to certain users. Tick it to control who can see the records this automation writes, then choose how:
* **Update**: add the chosen users to the record's existing access list.
* **Replace**: replace the record's access list with the chosen users.
4. **Limit Data by User**: who gets access. Either one option from **Auto limit data based on**, which takes the person from the task when it runs:
* **Same assignee**: the user assigned to the task.
* **Assigned by**: the user who assigned it.
* **Done by**: the user who finished it.
or one or more named users from the **User** group.
## How it works
1. A task of the automation's task type is finished, and the [rules](/pages/workflow/automation/rules), if any, are checked.
2. The value mapped to the **primary key** is looked up in the data type.
3. **A record with that key exists**: it is updated with the mapped values. Fields that aren't mapped keep their values.
4. **No record has that key**: a new record is created with the mapped values.
* Map a value to the primary key field. Without it, records can't be matched.
* The primary key is matched exactly, including upper and lower case.
* Required fields of the data type must get a value, fixed or copied.
* The data type's field validation applies.
## Examples
**Customer database after delivery**
| Setting | Value |
| - | - |
| Event | On Task Finished |
| Task Type | Delivery |
| Data Type | Customer Database (primary key: Customer Name) |
| Customer Name | Customer Name (from Delivery) |
| Customer Address | Customer Address (from Delivery) |
| Customer Coordinate | Delivery Coordinate (from Delivery) |
| Customer Phone | Customer Phone (from Delivery) |
**Inventory after a stock check**
| Setting | Value |
| - | - |
| Task Type | Stock Inspection |
| Data Type | Inventory Database (primary key: Product ID) |
| Product Name | Inspected Product |
| Current Stock | Stock Count |
| Warehouse Location | Inspection Location |
**Service history**
| Setting | Value |
| - | - |
| Task Type | Home Cleaning |
| Data Type | Service History (primary key: Service ID) |
| Customer Name | Customer Name |
| Service Type | Service Category |
**Sales leads from canvassing**
| Setting | Value |
| - | - |
| Task Type | Field Canvassing |
| Data Type | Sales Leads (primary key: Prospect Name) |
| Contact Number | Phone Number |
| Interest Level | Survey Result |
| Data Source Access | Update, Limit Data by User: Done by (the canvasser sees their own leads) |
## Good practice
* Choose a primary key that really is unique, such as a customer ID or order number.
* Prefer copied values over typed ones, and map only fields that matter.
* Use [rules](/pages/workflow/automation/rules) to write only when it makes sense, for example only successful deliveries.
* Test with a sample task, then check the record on the Data Source page and the run in the [log](/pages/workflow/automation/log).
## Troubleshooting
**No record is created or updated.**
* Check that the automation is **Active**, the task type matches, and the rules don't exclude the task.
* Check that the field mapped to the primary key had a value in the task.
* Check that every required field of the data type has a value.
**A new record is created every time instead of updating.** The primary key value doesn't match the existing record exactly (case included), or the wrong field is mapped to it.
**Some fields don't change.** They're not mapped, or the task's field was empty.
**The wrong data type is updated.** Edit the automation and check **Data Type**; look for other automations on the same event.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.