Terms of use

TableWorkbench offers browser-based tools for preparing and reviewing tabular data. General operations include cleaning, deduplication, appending, joining, comparing, filtering, sorting, and reshaping. Dedicated tools handle selected workbook data, calculations, nested records, time-related tasks, and specialised output files. Supported inputs and outputs differ by page. Read the individual tool description and the options shown in its workspace rather than assuming that every format or operation is available everywhere.

The current tools require no TableWorkbench account and do not offer a paid subscription, a cloud project store, or manual review of your data. They help apply selected rules to the inputs you provide. They do not decide which business source is authoritative or certify that a dataset describes the real world correctly. Choose a tool whose operation fits your task, and use the result only after reviewing the decisions and outputs relevant to your work.

Process only data that you are authorised to access and use. Selecting a file for local processing does not change ownership of that source file. Keep an original copy and follow any restrictions that apply to the information in it. If an organisation controls the data, its rules about devices, browsers, processing, storage, and sharing still matter even when the work happens locally. Do not assume that a convenient file picker establishes permission to use someone else’s records.

The application reads the selected data and generates new output files without actively overwriting the source. You decide where downloads are saved and what you do with them afterward. Check filenames and destinations before replacing an existing file through your browser or operating system. Keep the supporting reports needed to explain an output, and avoid sharing fields or reports that the recipient does not need for the intended purpose.

Input settings affect the dataset the tool receives. A delimiter, header choice, skipped row, encoding, or worksheet range can change the apparent structure. Inspect the preview and confirm that important values are in the right columns before running an operation. A CSV extension does not guarantee a particular separator or date convention. Workbook values, displayed text, and formula text can also produce different inputs depending on the mode a tool supports.

Date and number transformations require an interpretation of the source values. Review ambiguous dates, separators, missing markers, and rounding choices. Text that contains digits may be an identifier rather than a quantity. The workbench does not know the intended meaning of an undocumented field. If a source value is already damaged or missing, another transformation cannot reliably recreate it without an appropriate reference or an explicit rule that you are prepared to justify.

Check result counts, field names, changed values, and exception reports before exporting or applying a transformation. Use a few known records to verify how your rules behave, including relevant edge cases. An operation that finishes without an error has completed its supported computation; that is different from establishing that the rules answer your business question. A comparison with no changes can still compare two incomplete exports or use a key that was inappropriate for the task.

The tools do not replace a financial audit, tax filing process, professional review, or another system’s acceptance checks. Arrange the independent review appropriate to the consequences of an error. For a routine catalogue cleanup, a count and sample review may be useful. For a critical import or reconciliation, additional checks may be necessary. You remain responsible for deciding whether to rely on the output and whether the chosen process meets your organisation’s requirements.

Selecting matching fields defines which records may be related. Duplicate keys, blank identifiers, or similar names can make that relationship ambiguous. Fuzzy candidates are suggestions for review, not proof that two records describe the same entity. Where a tool asks you to confirm matches or resolve conflicts, inspect the relevant values before accepting them. A high similarity score does not replace knowledge of the source or the meaning of its identifiers.

Aggregation changes the level of detail. Grouped totals and pivot cells depend on which records participate and how missing or invalid values are handled. Expanding one record into several rows may repeat an amount, and summing that amount later may count it repeatedly. Running calculations depend on ordering and groups. Read the page-specific explanation and preserve enough detail to trace a surprising total back to its contributing records.

A downloaded file must be interpreted by the application that opens it. CSV and TSV contain text fields rather than the complete feature set of a workbook. Data-focused XLSX output is not a promise to reproduce all formatting, charts, macros, or formulas from an original document. A receiving application may infer types, change displayed values, or apply its own import rules. Review the downloaded representation and use appropriate import settings in the destination.

Calendar, contact, and product files have additional conventions. Generating an event file does not create a live subscription or send invitations. Generating contacts does not establish how a particular address book will update existing entries. A product validation report covers the checks implemented by that tool and cannot guarantee that a store will accept every image, value, or account-specific requirement. Test a small destination import before committing a complete batch.

Working data is held in browser memory, not in an automatically saved cloud project. Refreshing, closing, or clearing the relevant workspace ends that in-memory work. Compatible tools can share an applied table in the general workflow, while specialised workspaces may have separate state. Follow the controls shown on the page and export anything you need before leaving. Do not treat an open tab as the only copy of a result you cannot afford to lose.

A saved rule file contains configuration rather than a complete dataset backup. It can help repeat a task, but new input still needs review and compatible fields. Downloaded results, reports, archives, and rules remain under your own file-management process. Clearing the page does not remove those files. If you share a package, inspect both the main result and supporting material for information that should not be included.

Browser capabilities, device resources, input shape, and output expansion can affect whether an operation completes. Limits on file size, record count, columns, cells, archive expansion, and task duration are safeguards rather than performance guarantees. A small compressed workbook may expand substantially, while a join with repeated keys may create a large result. Use coherent smaller batches when necessary and retain a record of what each batch contains.

Do not interfere with the service or attempt to bypass its protective limits. The current site does not promise permanent availability or complete offline operation. Features, supported formats, and limits can change as the application changes. Read the current explanation on the page you are using. If a result appears inconsistent with that explanation, preserve a fictional reproduction and use the contact options that are actually displayed on the contact page.

The privacy page describes local processing, network requests for website resources, memory sessions, and downloaded copies. Read it alongside the tool instructions when deciding how to handle a particular dataset. Local processing does not erase restrictions on the information or control what happens after you deliberately send a result to another service. Consider the complete path from source to destination rather than just the transformation visible in the browser.

For questions or suggestions, provide the tool name, the relevant settings, the expected output, and the observed result. Avoid sending complete business files when a few fictional rows demonstrate the issue. A report does not give the maintainer remote access to your session or recover an unsaved table. The contact page explains the currently available route and the details that can help make a problem reproducible.