Rank Saved First-Byte Timings With Context
Admin ·
Start with a capture and a clear measurement basis. Request wait, navigation elapsed time and a field percentile answer different questions. Compare the slowest entries against evidence recorded alongside them.
Compare the two requests in the HAR example
Choose Load HAR example and then Analyze TTFB. The catalog request appears first at 1,900 ms. It was the second entry in the capture, so Entry 2 still points to the original order.
The example records a cache miss, a 1,350 ms origin Server-Timing value and separate DNS and connection timings. The asset request waited 240 ms and records a hit. Inspect the differences before deciding what to measure in a new capture.
| Endpoint | Recorded HAR wait | Cache marker |
|---|---|---|
| /catalog | 1,900 ms | MISS |
| /assets | 240 ms | HIT |
Open TTFB AnalyzerHTTP Archive 1.2 timing fieldsChrome DevTools network timing and HAR export
Use the timing basis when comparing values
HAR wait measures the captured request wait. A Navigation Timing entry instead lets the tool subtract startTime from responseStart. If requestStart is present, the result also shows that narrower phase.
The numeric Lighthouse server-response-time audit and PageSpeed field TTFB percentile can both appear in one saved report. Use the Measurement basis filter to review them separately. Use a HAR if the saved Lighthouse report contains only a newer document latency insight and no supported numeric audit.
HTTP Archive 1.2 timing fieldsW3C Navigation Timing Level 2Chrome Lighthouse server response time
Separate missing measurements from fast responses
Choose Load text example and analyze again. The checkout row retains its Retry limit outcome below measured observations. It has no numeric value. In your own notes, start each observation with an HTTP URL and give recorded timing values with ms or s.
Export JSON to preserve the source order, measurement basis and visible evidence. Review endpoint paths before sharing. Query strings, fragments, URL credentials and unrelated headers are omitted from the report.
Open TTFB AnalyzerChrome DevTools network timing and HAR export
Common questions
Does a 1,350 ms origin marker explain all of a 1,900 ms wait?
It is one duration reported by the service. Other recorded phases and repeated captures help test the explanation.
Why is one HAR entry unavailable?
The capture lacks a valid timing or has no HTTP response. The tool retains the row without turning unavailable data into zero.