Hacker News new | past | comments | ask | show | jobs | submit login
Migrating Away from Google Analytics (freshman.tech)
134 points by ayoisaiah on July 11, 2020 | hide | past | favorite | 42 comments



"Google Analytics is free to use (for up to 10 million hits per month), but it’s only because the company recoups the cost of running the service by using the aggregated data to track users for the purpose of showing personalised ads."

The above quote is a bit misleading. Google Analytics by default uses first party cookies, the data associated with which isn't used for targeting.

For users of Google Analytics who are using Google Advertising products they can also enable the use of the third party double click cookie for remarketing and other advertising targeting use cases. This can also be done without Analytics just using Adword tagging.

I strongly disagree with the premise of the quote that Google makes money off of Analytics by gathering data for targeting. Instead, the value proposition for Google is better described here, https://www.quora.com/How-does-Google-make-money-from-Analyt... .

In a nutshell, Google hopes to show advertisers the value of their advertising buys on Google by attributing conversions on their sites to the correct marketing channel.


> The above quote is a bit misleading. Google Analytics by default uses first party cookies, the data associated with which isn't used for targeting.

My understanding is that in Chrome all the GA data from a browsing session is sent to Google over the same HTTP/2 connection, which means activity can be correlated across sites by the unique TCP connection.

So they have the motive and the means to do this tracking, are you saying that Google does not do this? Seems like some proof is in order here.


Seems you want proof that something _isn't_ happening, wouldn't it be more appropriate for you to provide proof that something is happening?

Innocent until proven guilty and all that yadda yadda


Perhaps it's merely coincidental, but Google created this TCP connection tracking data where it wasn't there before or was of much lower fidelity; they made a specific exception for GA to reuse one connection across domains for 6 minutes in HTTP/1 in Chrome and then later had their protocol rubber-stamped that reuses one connection to GA for 30+ minutes in HTTP/2.

So yes, I would like to see some proof, for example a specific corporate denial, that this tracking data that they created isn't being used.


I agree that a public statement would be good because that shows more commitment to the policy as opposed to maybe a temporary state of affairs.

My belief is that first part cookie data is currently not used for ad targeting which is why third party cookies are also used and is one of the reasons that the whole impending death of the third party cookie cook have such a big effect on the online advertising industry.

Whether that means Google will look for workarounds involving the analytics first party cookie, I don't know.


If not proof, then an official ironclad promise that everyone including my mom can understand would be nice


If it data isn't used for targeting, how exactly do they deliver you demographic information about your visitors?

it seems pretty obvious visitors are absolutely associated with profiles of people.


Looks like the answer is here and in agreement with what I said above.

https://support.google.com/analytics/answer/2799357?hl=en

In short, demographics are from the third party doubleclick cookie and are only available if you enable advertiser features.

Whether this enables Google to click a significant amount more of data on these third party cookies is unclear to me. So my original claim could be wrong. I would guess that a significant amount of the information would still be collected via AdWords tagging for conversions or display Ads.

There are also Apple and Android device IDs for mobile. I don't know if those are app specific or device specific and if they are only logged if advertiser features are enabled.

So the story is a bit murkier than I claimed.


That's good question. I could be totally wrong. I'll look into this and reply back.


You have to enable it separately from settings, accept its eula etc. Definetly not enabled by default.


Props to the author for moving to an analytics tool that's more respectful to users.

For anyone who is interested in doing the same, I maintain an open source (and self-hostable) analytics tool called Shynet [0] that works without cookies or requiring any JS.

And to be clear, it isn't a SaaS -- the only way to run Shynet is to self-host it.

[0] https://github.com/milesmcc/shynet


I found your project when you commented on a similar HN post, and it's great. I've switched all of my sites from Fathom to Shynet. Nice work :)

(Fathom was good but the self-hosted version is in disarray)


That’s great to hear! I’m glad it’s working well for you. And yeah, Shynet is definitely in part a response to half-baked self-hostable tools like Fathom’s offering.


This looks great, thanks! I'm going to switch from fathom to this.


The opposite of not using Google Analytics is NOT use these new minimalist analytics tools. I'm so sick of seeing this narrative! There are plenty of mature analytics tools that do a thorough job for a small amount of money. For the curious, I've been using Clicky.com. I don't see how using tools that give you only an iota of information makes any sense.


The author says that he doesn't need a majority of the features Google Analytics has:

> I also realised that I wasn’t even using up to 10% of the features provided by the service. For the most part, all I cared about is the number of visitors, page views, and where visitors are coming from (search engines, social media e.t.c.) so all the other metrics that Google collects are not important for me.

But I agree, not everyone shares that thought. I believe that Matomo is the only real self-hosted alternative for more advanced analytics.


I have been also building a more complex self-hosted alternative, similar to Matomo: https://usertrack.net


I've opened the following issue to get input about how woke hosts files should treat Plausible and Fathom. I'm interested to hear good takes, if you have any, on the subject of ethical tracking.

What to do about Plausible and Fathom? https://github.com/StevenBlack/hosts/issues/1346


I have been considering giving up Google Analytics for my websites too, but the thing that stops me is that GA integrates really well with Google search engine, providing detailed info about which keywords were used to find your site.

Do other analytics services also provide this feature? If so, I'm curious of how the manage to do it without being integrated with Google.


Matomo (ex Piwik) does have an "importer" for search keywords (from GA and Bing and maybe something else) but it's not free: https://plugins.matomo.org/SearchEngineKeywordsPerformance

It uses the APIs of the search engines to download/import the search keywords.


Yeah, Plausible has a Google Search Console integration so you can see all the keyword queries people find your site with directly in the Plausible dashboard.


So glad to see any alternative to a Google product. It looks like we’re slowly moving to a world where it will be basically just Big Corps using Google products and then eventually even they will succumb under sheer weight of “why are you supporting Evil Corp?” It’s becoming uncool to use Google, and they deserve it.


I reccomend GoAccess https://goaccess.io/ as Google Analytics alternative

Not a service but a self-hosted server log analyzer and viewer, no javascript or any other page code necessary, plus it's fast and free.


I agree, works really well for us. Above all, no javascript!


This article has its heart in the right place at least.

Author mentions privacy as a factor but does not state why they think their new system is better.

> Unsurprisingly, Plausible’s reporting for page views was consistently about 30-40% higher presumably because ga is blocked by some clients. ga is gone for good now though.

What? Why is the higher number de-facto the accurate one? Maybe ga excludes and/or bots recognizes return visits better. More != more accurate, as the YouTube ad money scandals showed

I like simpleanalytics a lot but it can’t differentiate between one visitor viewing 3 pages and 3 visitors. Still, I appreciate their offering for visitor privacy


Just two weeks ago I moved to Matomo from Google Analytics- it works great, and it feels like a tiny little contribution to make the world a better place.


Yep I did the same a few days ago and its working well so far. Many of the same features and layout but without all the Google baggage.

Also with 1 or 2 line changes to the default config, you can allow it to get around any adblock/privacy filters that may have their default JS blacklisted.


Curiously, matomo analytics, a more privacy friendly competitor, keeps getting blacklisted by Edge for distributing malware. It happens so often that matomo has a faq about it. I suspect this means our sites will not show up in search engines and you have to fill out annoying forms to get re listed. It’s very annoying to get punished for spending a day trying to track people less


"I also realised that I wasn’t even using up to 10% of the features provided by the service."

Most of these people could get everything they want from just analysing the webserver logs. But I guess that isn't the modern way, a true web pro has just gotta load 12 frameworks to display a page of text.


That's a bit exaggerated. In OP's case, Plausible is one <1 KB file and it's easier to set up that than a log parser.

My personal favourite thing about analytics that I haven't seen log parsers do is live statistics, showing people moving on the website. I made my own analytics thingie but Plausible and GA also provide this.


If I use a CDN or another service like Netlify, I don't get any server side logs, so I must do tracking client side.


Absolutely not. That’s how we all used to do website analytics and there is a reason client-side packages (like GA) won in the marketplace. They just work way better.


The load time of the Google Analytics script differs quite a lot depending on the time of day. https://royal.pingdom.com/google-analytics-script-loads-97-s...


My time is limited. I'd rather spend that time gluing together other bits of functionality than reinventing each bit one by one, unless I have to


I’ve deployed this in the past.

Analog[0], AWStat[1], W3Perl[2] are some that come to mind.

[0] - https://www.c-amie.co.uk/software/analog/

[1] - https://github.com/eldy/awstats

[2] - http://www.w3perl.com


For a paid product dashboard is a bit basic: long lists have no pagination and opens in modals; Linux is split into distros, while Windows and Mac are just single entry for each; no browser versions and no way to check what browsers are used on what operating system.


Wish there was free competitor


PostHog Analytics [1] is open source, and sofar a very nice alternative (or complement) to Google Analytics.

It's more like Heap Analytics, in that it collects user clicks and other events from your site (or app) and allows you to retro-actively define "actions" based on these events.

For instance, if you decide today to keep track of how many users click your sign-up button on page X, PostHog can graph this metric for you for every moment since you first installed it on your site.

You can combine actions into funnels, graph nearly everything, the GDPR support is first-class, there's support for heatmaps, feature flags (rolling out new features to just a percentage of your visitors), and user cohort analysis.

[1] https://posthog.com/


Free is a big part of the problem. The existence of free tools in a space, that are paid for with data gathering, crowds out the market for paid competitors that operate in a more privacy friendly manner.


GoatCounter is open source and there’s a free hosted option for non commercial use:

https://www.goatcounter.com/


plausible is open source (https://plausible.io/open-source-website-analytics) and can be self-hosted.


You can have a look at https://ownstats.cloud if you happen to have a AWS account available




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: