Field note · performance / 3 MIN READ
A perfect score was not the point. The bytes visitors received were.
A homepage speed investigation became a lesson in measuring the deployed site, not just the local build.
ACT: ask for the delivered URL, device profile, date and actual file response before accepting a speed claim. This case is a dated author account, not a promise of a score or conversion gain.
What is checked
The author’s build log records a video reduction from 25.7 MB to 6.6 MB, a stale cache response and a later lab result. The linked September 27 report is historical; this briefing has not independently rerun that experiment.
What this cannot prove
No visitor field data or conversion study is supplied. Raw trace files, repeat counts and the full random-seek test setup are not attached. Current service-page scores and this dated log may describe different versions.
The first screen looked finished. The trace disagreed.
The author’s homepage investigation found that an entrance animation kept a ready headline invisible. A loading score can therefore reflect how the opening is revealed, as well as how quickly its files arrive. The reported change kept the visual sequence while adjusting when the text became visible.
For a buyer, the acceptance question is concrete: can someone read the offer on the first screen while the visual experience loads? Ask for the tested page and conditions. A source-code change alone is not evidence of the delivered experience.
The local file got smaller. The live file did not.
The log records a hero video changing from 25.7 MB to 6.6 MB. The first deployment still delivered the older file because the CDN—the service caching files near visitors—kept its previous URL. A versioned filename addressed that cache mismatch.
The reported random-seek timings are a limited comparison, not proof of equal playback quality. An HTTP 206 response means partial file delivery; it does not prove smooth scrolling on every device. The lab scores describe that dated run, not today’s whole site.
What a business owner should take from it
ACT when a visual site is being built or accepted. Record the URL, date, mobile and desktop profiles, actual asset version and any available visitor data. Ask the engineer to show the response received from the deployed site, then inspect reading and navigation on a phone.
PASS on a redesign driven only by a score screenshot. Agree what must remain readable and interactive, what is being measured, and what counts as failure before buying further visual work. No sales improvement is established by this case.
Sources and artefacts
APPLY THE METHOD TO YOUR BUSINESS
A useful answer starts with your own evidence.
For a video-led or visually ambitious website, discuss acceptance on the delivered first screen, navigation and real asset responses. This case does not promise your site the same score.
Explore website engineering ↗Service details and current pricing are on VKVstudio.com. Project scope and agreements are handled by email.