> ## Documentation Index > Fetch the complete documentation index at: https://docs.vern.so/llms.txt > Use this file to discover all available pages before exploring further. # Column options > Every setting on a template column, what it enforces, and the error it produces. A template is a named set of columns, and a column is the unit of validation: it carries the name the agent matches your source data against, the description it reasons from, and the rules every cell has to pass. ## Where to configure them Go to [Templates](https://app.vern.so/templates), open a template, then click a column to edit it or **Add Column** to create one. Changes save to the template and apply to every workbook built from it. A template can't be saved while a migration that uses it is running. Wait for the run to finish, or stop it first. The same lock applies to [`PATCH /templates/{slug}`](/migration-api/update-a-template), which returns `409`. ## Name The column header in every sheet built from this template, and the handle everything else uses: the import agent matches source columns against it, chat resolves `@Column Name` to it, and exports use it as the field name. **Columns are matched by name, not by an internal id.** Renaming a column on a template that already has workbooks is therefore treated as deleting one column and adding another — the renamed column starts empty on sheets that already exist, and any validation errors recorded against the old name are cleared. Settle names before you onboard customers; afterwards, use chat to move the values across. ## Description Plain English describing what belongs in the column. This is not decoration — it is the main thing the import agent reads when it decides which source field maps here and how to transform it. It is also what the AI uses to draft a validation pattern, and what chat consults when you ask it to fix a column. Write what a *good* value looks like, including the cases that trip people up: > Australian Business Number. 11 digits, no spaces or hyphens. Source system > stores it with spaces — strip them. Blank for sole traders without one. If the agent keeps misreading a column, the description is almost always the fix. ## Required Flags a cell as invalid when it has no value. * A cell counts as empty when it is missing, `null`, an empty string, or whitespace only. * The error recorded on the cell is `Required field is empty`. * It does **not** block an import or an export. Required marks the cell so it shows up in the sheet's invalid filter and in the import agent's review — it is a flag to resolve, not a hard stop. ## Unique Flags every cell in a repeated group as invalid. * Uniqueness is scoped to **one sheet** — the rows of this template inside one customer's workbook. Two different customers may hold the same value. * Comparison is **case-insensitive and trimmed**: `ACME `, `acme` and `Acme` are the same value. * **Blank cells are skipped.** Empty is never a duplicate, however many rows are empty. Combine with **Required** if a value must be both present and distinct. * The error names the count, so you can see the size of the collision: `Duplicate value: "acme" appears 3 times`. ## Validation Rule The dropdown picks how a cell's *content* is checked. Three options: | Rule | What it does | | - | - | | **None** | No content check. The column still honours Required, Unique and any link rule. | | **Strict** | The value must match a pattern. | | **Link** | The value must exist in a column on another template. | ### Strict Strict holds a regular expression that every non-blank cell must match. You author it three ways, all of which produce the same stored pattern: * **Visual blocks** — the default. Build the pattern from labelled pieces (`Digits (11)`, `Letters (1+)`, `@`, `Choice Options`, …) rather than writing regex. Each block takes parameters such as an exact length or a min/max range. * **Presets** — start from a ready-made pattern for a category (Email, Phone, Text, Number, Date, ID / Code) and adjust its parameters. * **AI** — switching a column to Strict while it has a name and a description generates a pattern from them. Review it; it is a starting point, not an answer. Two things to know about how it is applied: * **Blank cells are skipped.** A pattern never makes a column mandatory — that is what Required is for. * **The pattern is not anchored for you.** `\d{11}` matches an eleven-digit run *anywhere* in the value, so `abc12345678901x` passes. Anchor it — `^\d{11}$` — when you mean the whole value. A failing cell records `Value does not match required pattern`. ### Link Link means the value has to exist somewhere else — a supplier code that must name a real supplier, a parent id that must resolve. Reference checking is the one rule complex enough to need its own page: Single targets, polymorphic targets chosen per row, and references that live inside a structured value. ## Structured values Everything above treats a cell as one value. A column can instead hold a list, a record, or a list of records — addresses, phone numbers with types, line items. That shape is declared separately from the Validation Rule and is checked alongside it: Declaring a list, a record, or a list of records, and the raw JSON Schema escape hatch. ## What actually gets enforced The Validation Rule dropdown names **one** check, but a column can carry several at once, and each is enforced on its own: * Required, Unique, the pattern, the value's shape and every link rule are enforced **whenever they are present** — not only when the dropdown happens to name them. A column set to **Strict** that also has a link rule gets both checks. * Setting the dropdown to **Link** additionally turns on the grid's link behaviour: link errors are styled in the sheet and re-checked live as you edit a cell, and the [nested webhook export](/help-center/send/webhooks) sends the referenced record instead of the raw value. Those two surfaces are the only things that read the dropdown rather than the rule itself. Practical upshot: pick the dropdown value that matches what the column is *for*, and don't worry that adding a second rule will switch the first one off. ## When changes take effect Saving a template re-runs validation across every sheet already using it, but only for what changed: * Toggling **Required** re-checks that column for empties, and clears the old `Required field is empty` errors when you turn it off. * Changing the **pattern or the shape** re-checks that column's content. * Toggling **Unique** re-runs duplicate detection, and clears duplicate errors when you turn it off. * **Removing** a column clears every error recorded against it. Each sheet's invalid count is recalculated and its cache cleared afterwards, so the workbook reflects the new rules without anyone reopening it. ## Next * [Link columns](/help-center/templates/link-columns) — references between templates. * [Structured values](/help-center/templates/structured-values) — lists and records in one column. * [Spot-check and fix](/help-center/import/spot-check-and-fix) — working through what validation flags. * [Update a template](/migration-api/update-a-template) — the same settings over the API. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.