Netlify Analytics: Two questions - Service Workers and rule of large metric numbers

So, I finally gave in and was too curious NOT to enable Analytics on my main site.

I noticed two different interesting things.

First, I have a service worker that when it installs itself, it grabs a copy of an offline page. Every time my SW is installed, it registers a page view (about 5,000 from June 30-July30). That’s not necessarily a bug, but it definitely feels like it falls into “unintended consequence” territory. Curious if anyone has thoughts about that.

Second, if I take that 5,000 view bloat into account, my site reports 10,999 more page views in Netlify Analytics than with Google Analytics. I expected it to be bigger than GA… but not quite that large…

The stats (June 30 - July 30):
Google Analytics with my spam filter (a few different things built into this filter): 5,228
GA without filter: 5,384 (+ 156)

Netlify Analytics: 21,277 (+16,049)

Netlify Analytics without /offline bloat: 16,227 (+10,999)

Just curious to see if this is in keeping with everyone else’s experience. Just caught me off guard

I use Fathom analytics, and see some differences myself on a tiny 12-page site. No spam filters. Fathom has some bloat from local testing calling the tracking script, on the order of a few hundred page views.

Source Uniques Views Referrer-1 Referrer-2 Google
Fathom 986 2,400 134 36 109
Netlify 2,880 (hourly avg * 24 * 30) 4,709 324 35 751 (150 google.com, 601 google.com/url)

Hello,

Yeah, we’ve noticed that this will happen where you’ll see traffic on resources that you may not have seen in JS tracking systems - as you say, not really a bug, but not ideal either.

We’re working on some improvements so that you’ll be able to see what sort of traffic you’re getting so you can filter down better whether it’s actual human-being traffic or bots or other types of requests etc.

Frances

2 Likes

Hi Frances, and Netlify Team.

I have the same issues as Brob. I am trying to fathom the differences between GA and NA, and derive some useful conclusions from the NA statistics

Netlify Stats Aug 14 - Sep 12
Total pageviews: 324,846
Total unique visitors: 10,774

GA Stats Aug 14 - Sep 12
Pageviews: 6,915
Users: 2,021

I’ve been told that cookiebot adds an additional pageview to every new visitor. This accounts for some of it. But still - 7k vs 162k views is a disparity I cannot plausibly account for.

If I estimate 40% bot traffic, that gets the user disparity between GA and NA to 3x, which is plausible.

Can you suggest any way to derive meaningful pageview stats from NA?

Many thanks

I think all of this will become clearer once we roll out the ability for you to see what’s bots and what’s real people (on the way!).

Hey there, are you an official Netlify Dev? This feature would really be handy soon since I want to know the amount of humans on my sites not robots :grinning::robot:

hey Kyle,

the above was a former netlify employee who has since left the company - we don’t anonymize accounts when someone leaves any more because it doesn’t work well (as is obvious by this thread)

We still hope to provide greater capabilities for analytics and more fine grained information, but at the moment, there have still been no changes.

1 Like