Build a Spec Sheet Readers Can Filter and Verify

A spec sheet, short for specification sheet, records the characteristics of a product or model: dimensions, capacity, operating limits, materials and other defined properties. To publish those specifications as an explorable table, you need more than a column of numbers. Each value needs its unit, applicable conditions and source.

Start with the original manufacturer documentation, build a clean spreadsheet, then publish the reviewed data. The result should let a reader find a model, compare a property and trace the number back to its source.

Choose what one row represents

Use one of two layouts:

Reader task	Row structure	Example columns
Compare models with the same properties	One row per model or variant	model_id, width_mm, depth_mm, mass_g
Preserve specifications with different units or conditions	One row per model–parameter–condition combination, distinguishing value basis where needed	model_id, parameter, value, unit, condition, value_basis

The first layout makes side-by-side comparison straightforward. The second is safer when one property has several values under different conditions. You can maintain the detailed table as your source record and create a narrower comparison table from it.

Keep variants separate. If size, configuration or revision changes a specification, give that version its own identifier rather than combining several versions in one row.

Preserve the condition, not just the number

For a concrete example, Siemens’ data sheet for miniature circuit breaker 5SY6450-6 lists width as 72 mm and depth as 76 mm. It also lists rated operational current as 48 A at 40 °C and 45.95 A at 50 °C. Those current values cannot be represented faithfully by a single unqualified “current” cell. See the manufacturer’s data sheet, pages 1–2.

A working extraction could look like this:

model_id	parameter	value	unit	condition	source_page
5SY6450-6	width	72	mm	—	2
5SY6450-6	depth	76	mm	—	2
5SY6450-6	operational_current	48	A	at 40 °C	1
5SY6450-6	operational_current	45.95	A	at 50 °C	1

Here, “—” means no condition was recorded for that row; it is not a numeric value. In the working file, use a blank condition cell and document that convention.

Add these fields to the full table:

• value_basis: Preserve labels such as rated, typical, minimum or maximum. The two current rows above are rated values.
• source_url: Link to the original document, not merely a search result or reseller listing.
• source_revision: Record the document’s stated revision or modification date when available.
• checked_on: Record when you checked the source, separately from its revision date.

This is a data extraction, not a replacement for the manufacturer’s installation instructions, approvals or complete operating limits.

Make the spreadsheet behave like data

Before exporting:

1. Use one header row. Remove merged headings, decorative title rows and blank separator rows from the publication table.
2. Keep numbers and units separate. Store 72 and mm, not 72 mm, in a detailed table. In a model-comparison table, a header such as width_mm makes the shared unit explicit.
3. Normalize only comparable measurements. Convert units consistently, but retain the original value and unit when conversion affects precision. Do not treat width, installation depth and overall depth as interchangeable.
4. Leave unknown numbers blank. Do not substitute zero for “not stated.” If you need to distinguish unknown from not applicable, add a status column.
5. Check duplicate keys. Review repeated model–parameter–condition combinations, accounting for unit, value basis and source revision before treating them as duplicates. A minimum and maximum for the same property are separate records, not an extraction error.

CSV itself does not declare column data types or uniqueness requirements. The W3C CSV on the Web primer explains how accompanying metadata can describe and validate that structure. For a reader-facing version, build a dataset reference sheet covering field definitions, units, missing-value rules and provenance.

Publish and check the reader’s view

As of October 5, 2026, TablePage accepts CSV, TSV, XLSX and XLS uploads and generates a public dataset page with a filterable table and a shareable link. Upload only specifications you are permitted to publish—never confidential supplier quotes, private contact details or unreleased product information.

After publication, check that:

• Model identifiers remain intact, including leading zeros where applicable.
• Numeric values appear correctly, including decimals and negative limits.
• Conditions, value-basis labels and source references remain visible.
• Filtering to one model or parameter produces the expected records.

In the detailed layout, compare numeric values only after selecting the same parameter, compatible units, matching conditions and the same value basis. Sorting a mixed value column would otherwise put unlike measurements together.

Share the page with a short scope statement: which models it covers, when sources were checked and what a blank means. Readers should be able to compare the data without losing the qualifications that make each specification meaningful.