Free Data Dictionary Generator – Document Your Dataset Online
Turn CSV rows or pasted spreadsheet cells into field documentation. Check inferred types, missing values and distinct counts, write your own definitions, and download a data dictionary without signing in.
- ✓CSV & spreadsheet paste
- ✓Editable field descriptions
- ✓CSV, JSON & Markdown export
2. Review your data dictionary
Example datasetDataset overview
Statistics apply to all imported rows, not a sample.
Data quality and review notes
Open to see observed issues and review notes.
Write a business description for each field.
Editable field documentation
Field 1
Field details
Checking confirms your review, not the database's physical column type.
| Field | Data type | Description | Missing | Unique | Review |
|---|
Missing counts and types come from observed records. They are not verified database constraints or business definitions. Type changes and descriptions are included in every export.
3. Export or save your work
Exports include all fields in their original order, regardless of table filters. Review sensitive information before sharing any downloaded document.
Save your editing progress (no account needed)
Download an editable project and reopen it later. It saves field definitions and statistics, not your raw dataset rows or example values. Store the file safely if it contains confidential field information.
What Is a Data Dictionary?
A data dictionary explains the fields in a dataset: their names, intended meanings, data types, formats, accepted values and any rules the team has agreed to use. It is not the dataset itself. A list of 5,000 orders contains individual purchases; the dictionary explains what order_id, quantity and unit_price stand for.
That distinction becomes useful when someone else opens a file you created. They can see that a field named status exists, but the name alone cannot tell them whether “Cancelled” means a customer cancelled an order or a payment failed. That definition must come from a person who knows the process. AmriData checks the observed structure and helps you record those decisions.
How to Generate a Data Dictionary Online
- Bring in the data. Paste a table from Excel or Google Sheets, paste CSV, or choose a CSV/TSV file. Check First row contains field names and specify additional missing-value tokens if your file uses them.
- Generate Data Dictionary. The browser checks row lengths, identifies the fields and profiles all imported records within the stated size limits.
- Inspect the overview. Check field count, record count, missing values, mixed types and possible identifiers. Read any cautions before making decisions.
- Complete definitions. Use the table's Data type and Description controls. Open Edit details to document allowed values, format, business rules, sensitivity and notes.
- Export. Download CSV, JSON or Markdown, or use Print / PDF. Example values are omitted unless you deliberately include them.
If you only have a planned schema rather than observed records, select Manual fields. You can still write definitions, but the generator will not manufacture missing counts or uniqueness statistics.
What Information Does a Data Dictionary Include?
Field names provide a stable point of reference. A business-friendly label helps nontechnical readers. The data type describes how a value should be interpreted; a format might say “YYYY-MM-DD” or “two decimal places.” Descriptions explain business meaning. Allowed values and validation rules can specify what an upstream system expects.
For imported records, the generator adds missing counts, percentages and distinct nonmissing values. These are observed statistics, not promises about future files. For example, a field can contain five unique values in a sample without having a declared uniqueness constraint in a database.
The tool also records the user's chosen type separately from the inferred type. An analyst can decide that a digit-only customer code should be treated as text; the source values remain unchanged. Every metadata edit stays with its original field when you search or sort the displayed results.
Data Dictionary Example: E-commerce Orders
Consider a CSV containing order_id, order_date, product_category, quantity, unit_price and order_status. The generator sees a row of field names and inspects the subsequent values. A human editor then completes the definitions.
| Field | Possible type | Illustrative business definition |
|---|---|---|
| order_id | Text identifier | Identifier assigned to an order |
| order_date | Date | Calendar date the order was placed |
| product_category | Categorical | Merchandise grouping used in reporting |
| quantity | Integer | Number of units on this record |
| unit_price | Decimal | Price per unit, with currency recorded separately |
| order_status | Categorical | Order processing state |
These definitions are examples, not information the tool can verify from the CSV. One row might represent a full order or one line item from a larger order. Without that context, even a well-labeled column can be misunderstood. The preloaded example includes five fictional records, one blank quantity and leading-zero customer identifiers for testing.
Generate a Data Dictionary From CSV
CSV files are convenient for field-level documentation because the first row often provides headers, and the remaining rows give evidence about typical values. AmriData handles commas, semicolons and tabs, standard double-quoted fields, escaped quotation marks and quoted line breaks. If a file has no header, uncheck the header option; the fields will be labeled Column 1, Column 2 and so on.
The current limits are 1.5 MB, 8,000 data rows, 80 columns and 250,000 total cells. The page checks all accepted records; it does not secretly profile a smaller sample. Files beyond those limits are rejected with a message. Large database exports should be reduced to an appropriate documented extract before import, with the limitations of that extract recorded in your notes.
NULL and N/A are also missing; you can change that list. Nonmissing distinct counts compare original strings exactly, including letter case and meaningful spaces.Generate a Data Dictionary From Excel
Open the worksheet, select a rectangular region that includes headers, copy it and paste into the tool. Spreadsheet copy operations generally provide tab-separated cells, which the importer accepts. Alternatively, save the worksheet as CSV and choose that file.
Direct .xlsx upload is not available in this release. Exporting a sheet as CSV can also discard spreadsheet-specific cell types and formatting. If a code such as 00123 must keep its leading zeros, check the exported CSV before using it: once Excel has already converted the code into the number 123, the original zeros cannot be recovered from the file.
How Data Types and Data Quality Are Inferred
Type inference checks the nonmissing field values for integer and decimal representations, ISO-style dates, datetimes, clock times, booleans and text. Repeated low-cardinality text can be classified as categorical. A column with inconsistent observed types is marked Mixed. A completely empty field is Unknown.
Column names that look like identifiers and number strings beginning with a zero are handled conservatively as text. The tool does not guess whether 03/04/2026 means March 4 or April 3. Date-like strings with slashes prompt a caution instead. None of these heuristics can prove a database's actual storage type.
Quality notes cover blanks, duplicate headers, type mismatches and possible sensitive fields. A possible identifier is a column whose imported rows have no missing values and all distinct observed values. That observation is useful during review, but it is not equivalent to a formally declared primary key.
Data Dictionary vs. Data Catalog, Database Schema and Data Profiling
Data dictionary vs. data catalog
A dictionary explains individual fields in one dataset or data model. A catalog usually inventories many data assets, helping people find tables, reports and dashboards. Catalogs can also capture owners, lineage and access rules. This free tool generates individual dictionaries; it is not an enterprise catalog.
Data dictionary vs. database schema
A database schema specifies structures such as tables, storage types, relationships and constraints enforced by a database engine. A dictionary can document the schema and add human definitions. A CSV alone cannot tell this generator whether a column has a foreign key, a declared NOT NULL constraint or a particular SQL column type.
Data dictionary vs. data profiling
Profiling inspects actual records for patterns, missing values and diversity. The dictionary explains what those fields are for. AmriData includes limited profiling with the dictionary so reviewers can notice problems before publishing their definitions. It does not alter the source records.
Why Document a Dataset?
A field named net_sales sounds straightforward until two teams discover that one includes refunds and the other does not. A definition that states the aggregation level, exclusions, currency and reporting period can prevent incompatible reports. Similarly, researchers can record variable definitions and coding decisions so another person can reproduce the analysis.
Documentation also reduces onboarding questions. A new teammate should not have to search a lengthy SQL file or ask the dataset's creator what every abbreviation means. Where there is uncertainty, a visible “Needs review” label is more useful than a confident-looking but invented explanation.
Data Dictionary Best Practices
- Write definitions that explain what one record and one field represent, not merely a longer version of the column name.
- Document units, currencies, date timezones and code-list values when you know them.
- Keep inferred metadata separate from confirmed business and database rules.
- Review identifiers, leading-zero values and any dates whose format is ambiguous.
- Mark sensitive or personal information before sharing documentation. Keep examples excluded when they are unnecessary.
- Regenerate and review a dictionary after major changes to the source fields. This tool does not automatically monitor upstream schema changes.
Common Data Dictionary Mistakes
Leaving every description blank produces a list of columns, not useful documentation. Treating all digit-only fields as quantities can destroy the distinction between measurements and account codes. Claiming that a unique sample is proof of a primary key makes the dictionary stronger than the evidence permits.
Other common problems include combining values with different units, guessing what statuses mean, treating empty cells as zero, and exporting real customer values in a sample column. The generator reports observations; the person using the dataset still has to confirm definitions, legal obligations and business rules.
Data Dictionary Generator questions
Answers about file formats, inferred metadata and exports.
What is a data dictionary?+
A data dictionary documents field names, data types, definitions, formats and any known rules for a dataset. It differs from the individual records in that dataset.
How do I generate a data dictionary automatically?+
Paste CSV or tab-separated data, or choose a CSV file. Check the header option and click Generate Data Dictionary. Review the inferred metadata and add definitions before exporting.
Can I create a data dictionary from a CSV file?+
Yes. Files up to 1.5 MB are parsed locally in the browser, with limits of 8,000 data records, 80 fields and 250,000 cells. The tool rejects over-limit files instead of silently sampling them.
Can I generate a data dictionary from Excel?+
Copy a rectangular range of cells from Excel or Google Sheets and paste it into the editor, or export the sheet as CSV. Direct .xlsx file import is not included.
Is the AmriData Data Dictionary Generator free?+
Yes. CSV analysis, manual field documentation and CSV, JSON and Markdown exports are free.
Do I need an account?+
No. The standalone generator runs without signing in.
What information should a data dictionary include?+
At minimum, field name, data type, meaning and applicable format or allowed values. This tool also reports observed missing and unique counts when records are provided.
How is a data dictionary different from a database schema?+
A database schema formally describes database structures and constraints. A data dictionary adds business descriptions and usage notes. CSV analysis cannot verify database primary keys or foreign keys.
Can the generator detect data types automatically?+
Yes. It uses deterministic checks for numbers, dates, times, booleans and text. Results are inferred from the imported values and can be overridden.
Can I edit field descriptions?+
Yes. Open Edit details for a field to change its description, name label, selected type, allowed values, format, business rules and notes. Your edits are included in exports.
Can I export the dictionary as CSV or JSON?+
Yes. Download CSV for spreadsheets, JSON for programs, or Markdown for project documentation. The page also supports browser print to PDF.
Does the tool upload my dataset to a server?+
No. Parsing and profiling occur in your browser using JavaScript. The tool does not submit the dataset to AmriData for processing.
Can it identify missing values?+
Yes. It counts empty or whitespace-only cells and any optional null tokens you specify, then calculates missing percentages over the imported records.
What is the difference between a data dictionary and a data catalog?+
A dictionary details fields within a dataset. A data catalog inventories many datasets and may track owners, lineage, relationships and access policies.
Is the generated dictionary always accurate?+
No. Data types, possible identifiers and sensitive-field flags are inferred from the observed records or names. Confirm business meaning, database constraints and classifications yourself.