Colquhoun.AI
← All guides

Research Guides

Reconcile Megabytes and Mebibytes Before Reporting a Smaller File

· Colquhoun.shop editorial

Two website reports can show different file sizes without measuring different files. One may divide bytes by a decimal million while the other divides by a binary multiple. Before reporting a smaller download or a broken measurement, reconcile the unit and the underlying byte count.

This guide supplies a unit-conversion worksheet for research exports and website measurements. It does not claim a performance improvement, a ranking effect or a particular Colquhoun.AI measurement feature.

1. Keep the unit definition with the number

NIST's reference on binary prefixes distinguishes one megabyte, MB, as 1,000,000 bytes from one mebibyte, MiB, as 1,048,576 bytes. It likewise distinguishes decimal and binary multiples at other scales. The symbols identify different divisors, not interchangeable spellings.

For a comparison worksheet, store the original number and unit exactly as supplied. If an exporter labels a column ambiguously, do not silently choose a divisor. Check its documentation or obtain the raw byte count before treating the value as suitable for a precise comparison.

2. Reconcile a fictional apparent reduction

Imagine a fictional public research PDF containing exactly 2,097,152 bytes. One report displays 2.097152 MB. Another displays 2 MiB. Both values describe the same count of bytes under their respective definitions.

A reviewer comparing only the displayed numbers could announce that the second measurement is about 4.63% smaller. That would confuse a change of unit with a change of file. Converting both observations back to bytes produces the same result, so there is no demonstrated size reduction.

The repair is to correct the comparison, not to optimize an unchanged document. Keep any genuine later change separate: if a revised file has fewer bytes, quantify that change using a shared measurement definition and unit.

3. Use a conversion ledger

Converting units cannot reconcile different metrics. A compressed transfer amount and an uncompressed file size still need different labels even after both are expressed in bytes. Resolve what is being counted before normalizing how it is displayed.

4. Add small fixtures to the importer

A practical free local test can include the fictional byte count above and assert that its decimal and binary representations normalize to the same integer. Include a row with a missing unit and ensure the importer flags it instead of assuming a default.

Add a rounded display value as a separate case. A value shown to one decimal place cannot necessarily recover the original exact byte count. If the raw count is unavailable, keep the resulting precision limit visible rather than presenting a reconstructed integer as original evidence.

Keep these fixtures synthetic. The test needs numbers and unit labels, not private files or customer exports. Validate the conversion independently from the user interface that formats the final report.

5. Publish the convention once and use it consistently

State the reporting convention beside the table or in its methodology note. Use the same unit in a comparison column, and keep any source-specific display in a separate field. This makes a downloadable worksheet easier for another researcher to reuse without guessing.

A defensible finding says whether the byte count changed, the unit changed or the evidence is insufficient to decide. It does not turn a formatting difference into an optimization claim. That small distinction can prevent an impressive-looking but nonexistent improvement from entering a website research report.

This guide describes the built-in Web Reader and Keyword Density behavior verified for this publication. Available data depends on the page response; these tools do not measure visits or search rankings.