Tag performance
Third-party tags don’t just send data — they cost time. A misbehaving vendor script can slow every pageview on your site, and nothing in the vendor’s own reporting will tell you. Tag performance (app.monita.ai/performance) answers “is a tag slowing the site?” continuously, from your real traffic.
How it’s measured
Section titled “How it’s measured”The monitoring script (v1.28+) passively reads timing data the browser already records — it never instruments or wraps vendor code:
- Network time — per-request duration for every third-party host, from the browser’s Resource Timing data.
- Main-thread blocking — time a vendor’s script held the main thread, attributed by source via Long Animation Frames (Chromium browsers; the of samples include main-thread attribution stat shows your coverage).
- Page weight — transfer size per vendor, where the vendor’s responses allow it (
Timing-Allow-Origin). A blank weight means the vendor withholds sizing from every measurement tool, not just Monita.
Collection runs on a small sample of pageviews (2% by default) and reports once per page through the same batched call the script already makes — there is no extra request and no measurable cost to your pages. To change the rate, set perf_sample (0–1) on __monita_settings; 0 disables collection for a property.
Reading the table
Section titled “Reading the table”Each row is one third-party tag, aggregated across sampled pageviews in the window:
| Column | Meaning |
|---|---|
| Time / page (p75) | Total network time the tag added to a typical slow pageview |
| Main-thread blocking (p75) | Script execution time that froze the page, attributed to this vendor |
| Pages blocked | Share of sampled pageviews where this vendor blocked the main thread at all |
| Requests / page | How chatty the tag is |
| Weight / page | Bytes transferred, where the vendor permits measurement |
| Slowest request | The worst single request observed |
A slowing pages badge appears when a vendor’s p75 main-thread blocking crosses 200 ms — sustained blocking at that level is user-visible jank, and worth raising with the vendor or your tag manager owner.
Field data vs audits
Section titled “Field data vs audits”Tag performance is field telemetry: it catches a vendor misbehaving for real users, intermittently, or only in certain regions — the situations scheduled lab checks miss. Audits are the complement: on-demand lab runs that isolate a single tag’s cost deterministically, useful for proving a regression to a vendor after the field data flags it.