Pingsivo · 1519 words · Updated 7 October 2026
English editorial adaptation prepared with AI assistance. Hypothetical examples are labelled. Technical sources accompany the relevant explanations.
Before comparing a speed test with a download, check the units. Mbps and Mbit/s both mean megabits per second. MB/s means megabytes per second; French interfaces may write Mo/s. One byte contains eight bits, so divide a decimal Mbit/s value by eight to obtain its theoretical decimal MB/s equivalent. This converts the unit, not an advertised connection speed into a guarantee for every file transfer.
The small letter that changes the comparison
Speed tests usually display bits, while download applications often display bytes. Two visually different numbers can therefore describe similar transfer rates. Reading the number without the unit can make a working connection look much slower than expected.
For example, 80 Mbit/s is theoretically equivalent to 10 MB/s using decimal units. That does not mean every download will maintain exactly 10 MB/s. The destination, transfer method and conditions still influence useful application throughput.
Write the full unit in your notes. “Download: 80” is not enough. “Download: 80 Mbit/s, HTTP test towards Cloudflare” establishes a much clearer comparison. Combine this habit with the reliable testing protocol.
Convert units without changing the measurement
Divide bits per second by eight to obtain bytes per second. Multiply by eight for the reverse conversion. Thus 5 MB/s equals 40 Mbit/s using the same decimal convention. The arithmetic does not repair a connection; it makes two displays comparable.
Check prefixes too. Some applications use binary units such as MiB while others use decimal MB. Preserve the convention shown by the software. Start by checking units and orders of magnitude before trying to explain small residual differences.
Keep original precision in exported data and round only for presentation. Early rounding can hide differences between nearby measurements. Separate the underlying value from the simplified number shown in a summary.
Estimate download duration
A basic duration estimate divides file size by useful transfer rate after aligning units. A hypothetical 1,000 MB file transferred at a constant 10 MB/s takes 100 seconds. The calculation assumes a stable rate and a file size matching the data actually transferred.
Real transfers may ramp up, fluctuate or pause. A download application's average may also use a different time window from a speed test. Treat the estimate as a planning aid, not a promise accurate to the second. A documented range is more helpful than false precision.
To check the estimate, use an authorised file of known size and record total duration and destination. Avoid simultaneous transfers that change the conditions. A discrepancy means one or more assumptions require investigation; it does not automatically prove that the measurement tool is wrong.
Subscription, link rate and measured throughput
Several speed figures can coexist on one device. Your subscription describes capacity under its contractual conditions. The operating system may show the local link rate. A speed test measures transfers towards a particular destination. These numbers describe different parts of the chain.
A Wi-Fi link rate is not a direct Internet download measurement. A network port's capacity also does not establish that every component between the application and the destination can sustain that rate. Ask which part of the path each value actually describes.
The Pingsivo Internet test measures the browser-to-Cloudflare path within its selected profile. The engine documentation explains how transfers are organised. Keep measured results separate from advertised or negotiated hardware capabilities.
Why a particular download may still be slow
A good speed-test result does not guarantee identical performance from every service. The file server, route, application and current conditions all contribute. One measurement cannot identify responsibility for the entire chain.
Check whether the problem affects one destination or several independent activities. A destination-specific slowdown calls for a different investigation from a problem affecting every device. Do not compare services by starting several large downloads simultaneously, because they would compete for the same resources.
Consider device work as well. Writing the file, scanning it or processing it can affect the application's displayed duration. Keep security protections in place and distinguish network transfer from subsequent processing. Total task time is not always network time.
Read CSV exports correctly
Pingsivo's Internet CSV exports store rates in bit/s, while the interface normally displays Mbit/s. A raw value of 25,000,000 therefore corresponds to 25 Mbit/s. Latencies are in milliseconds. Read column definitions before calculating statistics or drawing charts.
In a spreadsheet, create a separate presentation column instead of overwriting raw data. Keep numeric cells numeric and put the unit in the heading or display format. This makes formulas easier to review and preserves the original result.
Missing values must remain missing. An incomplete test with no upload result does not mean zero upload. Replacing empty cells with zero would artificially lower averages. Separate complete and incomplete results before producing a summary.
Put the comparison in context
Imagine a household seeing 96 Mbit/s in a test and approximately 11 MB/s in a download. The theoretical conversion gives 12 MB/s. The apparent eightfold difference disappears, leaving a smaller difference that still needs context.
This is a teaching example, not a measured Pingsivo customer result. For your own investigation, retain time, destination and concurrent activity. Different tools do not necessarily observe the same duration or network path, so do not force their results to match exactly.
For video problems, useful throughput is only part of the story. Read the 4K streaming guide and the stability guide to connect unit calculations with the actual viewing experience.
Build a comparison sheet that preserves meaning
Start a worksheet with separate columns for timestamp, destination, measurement method, completion status, raw value and original unit. Add a calculated presentation column only after checking the unit convention. For an Pingsivo CSV download value in bit/s, dividing by one million produces decimal Mbit/s; dividing that result by eight produces decimal MB/s. Keep the original column unchanged. Someone reviewing the sheet can then reproduce the calculation and distinguish the data supplied by the tool from formatting added for readability. This is particularly helpful when a colleague later imports a second file with different units.
Use a different set of columns for latency and variation. Dividing a millisecond value by eight would have no relevance to a bits-to-bytes conversion, yet spreadsheet formulas copied across mixed columns can create exactly this type of mistake. Name calculated columns explicitly, and check one row by hand before filling the formula through the sheet. Where a measurement is missing, make the display say unavailable or leave an intentionally empty numeric cell. The calculation should not silently transform absence into a performance claim.
Suppose a hypothetical support conversation contains a speed result of 40 Mbit/s and an application display of 4.7 MB/s. The conversion creates a theoretical reference of 5 MB/s. It does not explain every remaining difference, but it shows that the two observations are much closer than the raw numbers suggest. The next questions concern destination, duration and conditions rather than an alleged eightfold loss. Document whether the application figure is an instantaneous display or a longer average, because that distinction can matter when a transfer begins slowly and then settles.
When estimating a completion time, keep the estimate separate from the measured duration. Record the assumptions: known file size, constant useful rate and no competing activity. If the observed operation takes longer, inspect which assumption changed. Perhaps the rate fluctuated, perhaps the destination differed, or perhaps the application performed work after receiving the file. Do not automatically subtract all of the extra time from the network's reputation. A useful report explains the calculation and its limits so the reader can decide whether the difference actually warrants further investigation.
Avoid reporting unnecessary decimal places in the human summary. A long string of digits can imply accuracy that the experiment does not have. Retain full precision in the original data, then present a sensible rounded value with its unit, method and sample context. The purpose of conversion is to make reasoning clearer. It should not produce an impressive-looking table that hides inconsistent measurements or encourages a stronger conclusion than the evidence supports.
FAQ: common unit questions
Are Mbps and Mbit/s different?
In this context they both describe megabits per second. Do not divide by eight when converting between those two spellings. Division by eight applies when converting bits to bytes.
Why is the download number eight times smaller?
First check whether the download uses MB/s and the test uses Mbit/s. Convert before interpreting the remaining difference. Different-looking numbers alone do not establish a fault.
Does 100 Mbit/s always produce 12.5 MB/s downloads?
The decimal conversion is exactly 12.5 MB/s, but a particular file transfer is not guaranteed to sustain that useful throughput. Mathematical equivalence and application performance are different claims.
Should I add upload and download together?
Not to advertise one representative Internet speed. They describe different directions and uses. A combined number would hide weaknesses in either direction. Show them separately alongside latency and context.
Which unit should a report use?
Choose explicit, consistent units throughout the comparison. Mbit/s works well for a human-readable summary; bit/s can suit storage and calculations. Label columns and retain the original file rather than multiplying unnecessary conversions.