It shouldn't.
Although:
- There is some variations between runs, esp. since they are so short and thus depend how the CPU will react with boosting, current voltage/temperature (if your room is hotter that day that alone will reduce boost clocks),etc.. For Corona this can easily be 2-3 seconds for 20 second run, and up to 4-5 second per minute.
- In reviews, benchmarks are run at fresh OS installation, before anything else. It's natural that there will be chance of interference from competing processes, perhaps the Defender decided to scan some files, or Photos app decided on cataloguing, on Windows started compressing files, etc.. that can be million things alone.
- Occasionally software mitigations (like Spectre for Intel CPUs, etc..) are fixed through patches that slow performance down to some degree to fix the loopholes. This is rather unfortunate.. ( I personally have these fixes disabled for my i9 for example, as they're really obscure issues)
If the discrepancy is bigger, some underlying issue can be suspected (quite often memory related).