Counting requests
What counts towards your monthly limit, why that is every request, and what happens when you go over.
Your plan has a monthly limit: 500K, 5M or 20M. This page says exactly what counts towards it, because that is where an expectation most often goes wrong.
Everything on your measurement address counts
Not just your events. Every request arriving at your measurement address counts, and there are more of them than you would think:
- the events your website sends;
- the
gtm.jsyour visitor fetches; - traffic from preview mode;
- checks that see whether your server still responds.
On one of our own servers, over four days, roughly a third of the requests were the loader and half were other traffic. Reckon with your event volume alone and you will consistently come out too low.
Why it is counted this way: this is the number your server actually runs on, and it is the same number you see back in your traffic log. Stape and TAGGRS count the same way.
What happens at 80, 100 and 110 percent
| At | What happens |
|---|---|
| 80% | you get an email |
| 100% | you get an email |
| 110% | your measurement address closes |
That ten percent of headroom above your limit is there so you can act on the warning without anything going wrong. Do not act, and it closes.
Closed means closed. Requests are refused from then on — a
402 Payment Required — and those measurements are lost. Not an empty response
that looks like success: your website would believe it was being measured while
nothing happened, and there would be no trace at all.
Nothing is billed afterwards
Go over your limit and no per-request invoice follows. If you need more room, you move to the next plan. That is the whole arrangement.
What the counter itself runs on
Usage is totalled per day, per measurement server. That has one consequence worth knowing: period boundaries are rounded to the day. If your subscription starts on the fifteenth at two in the afternoon, the whole fifteenth counts. That falls in your favour.