Google Analytics 4
blog.google
blog.google
And I hope they do.
I also have setup three implementations of Firebase and GA4. I can't for the love of god use this tool. It is in my eyes nothing but a mechanism for upselling GCE (esp. Big Query).
Answering basic management and marketing questions about marketing goals (or other conversion goals reached) is a PITA. Building basic dashboards with Google Data Studio also.
Taken together with the EU data privacy regulators regarding GA I nowadays just recommend clients to just ignore GA. Even if the USP is "but I can see my advertising numbers in there as well".
Do they want to push people away from using their product?
While with most of them you get what I would call "vanity metrics" (so not really actionable, more or less just "numbers go up or down") at least Matomo (self hosted) provides the ability to create essentially the experience of a free GA3 implementation.
If you "just" had a code drop of basic GA3 code: just use whatever floats your boat. If you have custom dimensions, user interactions tracked as events, goals and stuff like that in the free GA3 version - go with Matomo (self hosted). Configure Matomo to have more than 5 custom dimensions (easy to do) and you can replicate (nearly) everything from your GA3 implementation.
Only things you will be missing in any case is integration with Search Console and Google Ads.
This comment couldn't have come at a better time for me. I was just trying to compare the free analytics available and plausible looked okay to me. But since I haven't done anything actionable yet I think I'll research a little more.
> so not really actionable
What would you define as actionable metrics? Are you talking about the custom page triggers like scroll depth, click on obl, etc? Can you recommend what are the biggest actionable events for a saas site since you have experience with it?
My source of knowledge on this is sadly a few youtube tutorials but if there is any resource you would recommend I'd highly appreciate the link.
Sorry, couldn't resist the in joke of the team I am part of.
But let me try an example. Visits as a global metric don't tell you much. They go up, they go down. And even if you do different marketing things (like SEA, newsletters, social media marketing, organic social and so on) and only look at visits from NY of these channels that doesn't tell you much (except for the volume).
But if you look at what a visit from every one of these channels costs you on average we are getting somewhere. Add into the mix the conversion rate for every one of these channels (depending on your goals it could be e-commerce conversions, newsletter sign-ups, sign-ups for a platform and so on). Now you can already see cost per acquisition/visit and cost per conversion.
So now you can answer the question about what marketing efforts bring more bang for the buck.
That is one example of how I define actionable metrics. Metrics that help you/enable you to make better business decisions.
How I would structure and report things depends on the business. But maybe this gives you an idea already.
Really appreciate you taking the time to write this. After some initial research I see not using UTM tracking tags(1) in all my URLs (or redirects) is a big mistake. So duly noted. You're right about one thing though that since there is no direct integration with Adwords / GWM you're losing out on a lot of info using anything else other than GA.
In the end you are not loosing this data. It is just one step removed. So you need to think about extracting it from Google Ads into Excel/Google Sheets (or an external dashboarding solution) and combining it with a data extract from your analytics solution.
And if you do more, like social media advertising, Microsoft Ads (think Bing) or Twitter they all don't easily integrate into Google Analytics. For all of them you would need to look ad the marketing performance in the platform specific tool and combine this with your on site metrics as described above.
Additionally you would want to take the consent rate into account (if serving European customers). Many of the consent solutions provide this number so that you can use it to adjust things like CPO (cost per order) or cost per visit (CPV) as you are probably not seeing 20 to 60 percent of your traffic due to consent not given. This depends massively on your target audience and the countries you are serving. I have seen this range at multiple international clients of mine for different brands and markets.
So GA having this integration is nice but definitely not sufficient for marketing.
Even when thinking about the integrations for the (quite expensive) paid offerings of GA and DV360 and the likes.
I don't know why people classify fathom and Plausible in the same category as Google Analytics. There's a clear distinction between an analytics software and a counter.
Matomo has some features but I can never like it's cluttered interfaces that don't lead to immediate actionable intels.
PostHog is another alternative.
It depends on what the people needs or want. A smallish blog has different needs that a global company with multiple brands in different markets doing e-commerce, brand advertising and what not (think VW or BMW). And a medium sized B2B e-commerce shop has again differing needs.
Personally I like Snowplow. But I am a tech loving nerd. If you have the capacity to build and maintain the tracker, servers and (for reporting/visualizing) necessary ETL infrastructure you can build the perfect fit in terms of what data to collect in what way and how to use it for your org).
The cost for building and maintaining it (with redundancy, stability and everything else) on the other hand probably would be in the same ballpark as solutions like parts of the Adobe Experience Cloud. Especially the new 'Customer Journey Analytics' offering is very intriguing. Having watched the summit I was a little bit blown away by the new capabilities for reporting they just announced.
I have to admit that I actually like working with Adobe Analytics (and CJA) and the development of the solution over the last few years has been great to watch). Still it has its problems. For example using Audience Manager creates a tracking call with an unique identifier being sent to their data center in Virginia. Even if all your data resides in the European data centers this is happening. This currently is a problem for clients of mine as the explicitly choose Adobe over Google because sales told them their users' data would not leave Europe under normal circumstances.
But back to Snowplow. You need some Form of visualizing and reporting infrastructure on top. So something like Apache Superset. Works well. But another thing to build, maintain and develop further.
I like both. And I like to be able to fully own the data pipeline. Especially in cases where you are in a regulated industry or have potentially sensitive data. I know of at least on bank using Snowplow to not only track the website and online banking, but ATM usage, call center calls and visits to your personal consultant/teller. They build a real time dashboard from that data so that when you visit your bank the employee there can see in real time your last touch points and interactions. Also they build product recommendations on top of that data that in the cases known to me actually are beneficial to the customer. Like telling you when your behavior with getting money from ATMs of different banks could warrant an add on so that you would save on fees. Because they value the customer happiness as a way to generate less churn and additional WOM marketing (word of mouth).
So it totally depends on your needs and the business in my book.
I’m used to quite a bit of verbiage in HN posts about Google ruining the internet with commercialization and ads and analytics and that sort of thing.
But the comments here of which I’m a naive observer I guess, seem to be of a “how dare Google change what is an essential part of our tools!”
Is this just a case of different HN subcultures taking time to post to each other? Or is it just my lack of understanding of the nuances of which part of modern web apps is good and/or evil?
Not really, at least you observed something. It is more like a silent majority. And I have been heavily downvoted for pointing this out previously.
Sometimes post 2016 it went all the way to Ads equals evil on HN along with the rest of the internet and social media. Around 2017 / 2018 tracking becomes evil. This is further fueled in 2019 when Apple went on full PR attack ( or one could argue adding fuel to fire ) about tracking and ads. ( They subsequently pulled back and state they are not against ads after industry backlash ). Target advertising ( whatever that means [1] ). And working in FANNG ( before it was called MAAMA ) would become morally wrong for a short period of time on HN.
But there are, legitimate uses for Ads, and legitimate uses for tracking. So every once in a while you see a few people who dare to offer their contrarian view, those that are working in the ad industry, or working as a business that requires the usage of tracking, funnelling and ads for sales. And offer their views that somehow gets the upvote to top post. Assuming the upvote is not gamed in anyway, that means there are a huge number of HN readers that rarely participate in any discussions, but upvoted something that largely go against the common trend on HN.
There are also some sort of political angle to it. Although I tend to think they are separate issues rather than causes.
[1] https://www.ben-evans.com/benedictevans/2021/8/27/understand...
1. Do everything for free. Rarely tenable.
2. Charge for things. This necessarily puts focus of business on monetization, or at the very least, it frequently pits providing and extracting value at odds.
3. Delegate monetization (a.k.a. ads). Then you can just focus on providing maximum value to your customers and money scales proportionally.
As a builder of things who's not greedy, #3 sounds like positively best option, if it's financially viable and product type makes it possible.
My mom asked me to disable ad blocking I've set up on her devices, as she finds them useful.
Except it's not. Most content that I consume - like your comment here - is created by individuals that do not get paid no matter how the platform that the content is shared on is monetized. It's absurd to say that there would be no content without monetization and when people have content to shere there will be ways to share that content even when ads illegal or blocked by everyone. However as long as ads are accepted by the public at large, ad-laden sites will be more profitable and thus also have more resources to crush any alternatives that provide an ad-free user experience. I'm quite fine with anything that needs ads to survive to disappear as a result of ad-blocking - there will always be something else to fill my time.
If companies just put up ads relevant to article as a static image that took up a reasonable amount of space and scrolled with the rest of the content, people wouldn't complain.
The reality is this is adverse selection: those users probably weren't interested in a post about updates to GA, and didn't click it.
Not anti-ads and tracking doesn't mean those are pro-ads or pro-tracking.
>If that was the case we'd expect the odd pro-ad/tracking comment to always be near the top of the comments section.
It often takes lots of courage to post something contrarian. That is why it rarely happens.
I don't believe that's true. I'm pretty sure it's a no-op for me.
I made a new account, wrote a message, upvoted it with my current account.. and no change in points. I never see any message become black or "less grey" after upvote + refresh. It is a no-op for me.
When asked, more than 90% of people opt out of tracking and ads [1]. No idea what silent majority you're talking about.
> But there are, legitimate uses for Ads, and legitimate uses for tracking.
Yes, there are, and they are very limited use cases.
[1] https://arstechnica.com/gadgets/2021/05/96-of-us-users-opt-o...
Yeah, I think that's exactly it. It happens with a lot of other topics too. It's nice that HN does seem to have more success in keeping people with opposing opinions on the site than some other communities.
I have a handful of personal sites that I've slapped Google Analytics on, and while they're mostly using Google Analytics out of the box, I'd rather not have to spin my wheels reworking some custom events.
- Google - Facebook - JavaScript - Anything web after 1994 - Crypto - Paying for things
It's kind of an anti-tech comedy club.
Google is a mixed bag; they used to be a force for real good on the internet, now they're mostly mediocre with a side of evil, riding on their past benevolence. I use their tools here and there, even if I think the internet is overdue for some serious de-googlization.
Facebook is just plain evil. I don't use their stuff in my personal life, but many businesses demand that you set up Facebook ads/analytics for them. I use fake accounts for those.
Amazon is pretty evil too, maybe AWS somewhat less so. I avoid them whenever possible (freelancing, etc., where I get to choose the stack) but oftentimes AWS is baked in to existing stacks and the companies are not interested in switching to anything else.
It doesn't require massive cultural differences, necessarily, but just a degree of moral compromise that most adults have to make in order to survive. Capitalism doesn't easily make for benevolent corporations and moral purity in tech is impossible when like 90% of it is owned by a dozen huge companies.
Those page view numbers do have a relationship to my business. But it's like measuring traffic on a highway when your store is next to it. The numbers will be somewhat related, but not actionable in any way.
I guess it makes sense if you get paid by page views (ad-supported sites?). But I have a B2B SaaS and I get paid by subscriptions. I rarely care about any of the metrics that GA shows, and I found that many of them are bogus anyway. I could probably wrestle GA into measuring what I need, but why bother, there are easier ways to handle that.
A nice side benefit is that I'm not participating in the global people-tracking business and I'm not subjecting my users/customers to that.
1) Are the 2nd and 3rd layers free or do they cost?
2) Do you know the official nomenclature of these layer products so that I can google how to implement them?
Thank you.
The first level they mention is out-of-the-box, no setup Google Analytics. You install the site code and you get metrics. You don't need to be a developer or have any real skills to accomplish this. It is essentially low-effort, low-return.
The 2nd and 3rd layers do not require an investment in Google, but they do require an investment in people. You'll need someone who understands the Google Analytics API, the Measurement Protocol, how the script collects and sorts data, who can install, configure and implement Google Tag Manager, and who has at least an intermediate understanding of HTML/CSS/JS.
There is no nomenclature apart from the aspects of Google Analytics I described above, but you can read more about them here: https://developers.google.com/analytics/
1.: No - but you need people understanding your business and GA as well as being able to implement additional tracking events via Google Tagmager (for example) to be able to answer questions like: What percentage of our users came via paid social media advertising, viewed a product, added it to the shopping cart, but abandoned the checkout. And how many of them are newsletter recipients so that we might be able to send them a reminder, that they have items in their shopping cart.
2.: No official nomenclature - just deeper data analyst's knowledge to ask the business people the right questions, develop an understanding of what to measure, talk to the developers to have them help provide (if necessary) additional info in a dataLayer structure and implement the necessary tracking events in the tagmanager.
After that said analyst needs to build the necessary dashboards/reports so that management/business can have the initial questions answered.
Data/Web Analytics in a nutshell (and described way reduced).
#2 - custom GA events implemented via Google Tag Manager triggers. only takes a few hours to learn.
#3 - once you identify a few particular funnels you want to dive deep into, just skip GA altogether and use an all-purpose recording tool like Hotjar that'll let you automatically record entire user journeys starting or ending at some page. go through those with your UX team/person and look for potentially confusing interactions. takes like 5 min to set up, several days to collect data, and then a good UX person to interpret likely pain points that your users are running into. it's really hard to do a good funnel analysis in Google Analytics. Hotjar, or to some degree Facebook Insights, make that analysis a lot easier because they're GUI-driven and monitor the DOM visually, not code-and-config based like GA
I feel like your questions were answered - pretty much the answer I'd give.
1) The only cost is cost of a person who has a deep understanding of Google Analytics. Google Analytics courses are hours long but they only teach you the first layer. 2nd and 3rd come with experience - I've been working with GA for about 4 years and there are still things that surprise me, caveats that we sometimes forget about. Quick example: Beginners will quite often select 2 dates (separately), get number of users from those 2 dates, add them together and say: that's the number of users we had in those 2 dates. But if you selected those 2 dates together, you'd see a lower number. Why? Because a user who visited in both periods would count as 2 if looking at separate dates, but only as 1 if looking together.
2) There isn't anything official but there's lots of blog posts - from great people like Simo Ahava and Julius Fedorovicius. If you're just starting out, I'd recommend sitting down and writing questions you want answering. Default GA implementation will not give you answers to all your questions - they had to create a product that works OK for most, so it doesn't answer specific business questions. The greatest problem with people saying Google Analytics doesn't give them the data they need is they don't start with the question. Examples of tracking I recently implemented because I taught the team to ask questions WHEN they change/implement something: -Are users printing the pages? Tracking is set up to track when they use browsers print button. -Do we have any searches on page that return no results? Do we have searches that return too many results? -How accurate are our results - which position of the result do they normally click on.
Those questions you wouldn't answer with basic setup. By setting up custom event tracking, we can answer them.
Lastly, on the most advanced setup (keep in mind you might not actually need it) - new version of Google Analytics offers free integration with Big Query. This means you get to analyse the data using SQL.
For me, a much better approach would be my own analytics, tailored to my product using simple sql statements.
For example how many of my users are tracking mainly dividends vs tracking their portfolio? To get such data I would have to invest more time into setting the right firebase analytics flags than just creating a BI Dashboard that uses one sql query under the hood.
Are there best practices to follow? I just have a table with: id, id of thing being accessed, user email, time, event enum, and notes.
It works well enough, and I have a python script to query the table and aggregate and print graph or print the results. But it would be great if there was some "bring your own data" dashboard that would let me view the time series w/ things like filters
Sounds like Grafana
I hope you get good traction!
I have an account but have not invested yet...
As the most dominant player in the market, they could do this (and to a certain extent still can). But for how long?
So, clean cut seems like a good idea.
GA3 and GA4 are so different, it isn't like the US and Russia, it is Earth and Jupiter (not in terms of impact or size, but just vastly different).
Like a toothbrush and a table. Vastly different.
it allows you to understand what drives revenue and measure the exact MRR you get from a blog post, churned users you get from an ad campaign, etc.
here is an example dashboard: https://hockeystack.com/live-demo/dashboard-preview/Content%...
But using this solution would mean loading third-party JavaScript with third-party trackers. Which for many reasons is a complete non-starter. I'm quite happy to have a site with zero third-party scripts/trackers.
So, call me when I can send the events myself from my code :-)
How well would a new cookieless search engine hosted in the EU do charging $99 a month to use when Google/Bing exist?
it's unifying the whole customer journey with no code "while" being cookieless and hosting in the EU
I respect your opinion, but please also check out how much tools like Amplitude or Heap cost
^^^ This !
If its free, then you are the product.
I'm sure all that analytics data is a veritable goldmine for Google.
I pulled Google off all our web properties three years ago (it was originally put there by the web designers) and no amount of blog hype will ever convince me to put it back.
I built a mobile compatible website (after previously believing my product was a strictly desktop/laptop only product).
I translated my site into Portuguese, since the majority of my users were from there, and many of those users were acting like they didn't understand the site (written on English).
Those two things were probably together a 20% revenue uplift. Worth having Google analytics.
The usual story is that I should care about traffic and bounce rates, because of short attention spans and potential customers getting bored or not finding what they needed — but seriously, if you are about to start using inventory control and production control software that will make your life easier and that you intend to use for years, attention span is not a problem.
The metrics I do care about are signups, churn, and everything else that is money-related. But those I can (and do) measure myself.
Note that I am not saying that analytics are totally useless, just that they aren't useful for me.
If you apply that thinking to other areas it doesn't make as much sense e.g. 'I don't care if my site takes 2 minutes to load. I don't get paid for how fast my site loads. I get paid by customers who subscribe and pay for using my software because it brings them value.'
Like sure, that's not what you are directly paid for but it doesn't necessarily mean the data can't help you increase signups.
If you think of it in ML terms - that's extra data that would likely increase a theoretical model's accuracy as it's not fully covered by your other (more important) metrics. Except in this case the model is you and it's possible the extra work isn't worth it for the benefit which could be tiny.
I use adsense for that. It counts page views which is what i need. GA seemed to me like something full of marketing buzzwords for which i really really dont care
I wrote code for traffic source attribution and used it to find out that my money spent on super-targeted LinkedIn, Facebook and Google ads resulted in precisely $0 revenue, so I stopped advertising and the code is now rusty.
I wrote code for A/B testing (this is something you really want to have integrated with your site) using the multi-armed bandit approach and used it to find out that with low traffic of a B2B SaaS it takes months to reach conclusions. That is still sometimes used, but not a whole lot, and only because the multi-armed bandit approach with Thompson sampling is nice enough to prioritize "better" options over time, so you can leave it running safely without baby-sitting it the whole time.
Overall the problem with a B2B SaaS is that unless you are already huge, the amount of good actionable data that you can work with is small. It's much better to spend the effort on making a product that people will want to use, e.g. talk to your customers.
Properly instrumented code allows this as a side-benefit.
The main benefit of properly instrumented code being that it makes debugging a breeze.
Why are people using client side (even more, 3rd party) code for this? The server knows it, you can get it from the logs.
Google Analytics or its competitors take a few minutes to set up. Recreating that whole stack to feature parity would take forever.
The way Google mentions "privacy" in the first sentence reminds me of a mining company mentioning "the environment" in the first sentence of their plan to dig a gigantic hole somewhere.
A related note: I have concerns with Google's recent nagging of website owners to upload user data to Google, under the banner of "enhanced conversions".
The idea is that Google takes the user data and matches with their user accounts. So a user logged into Google who clicks an ad, can be tracked across devices without the need for third party cookies.
They expect the upload to happen on the conversion page - such as the thanks page on a website contact form. To me it feels like Google are trying to convince business owners that to be "leaders in your industry" they must help Google fill the void created by third party cookie banning.
Site owners are now pushed to upload every contact form submission to Google for the purpose of what is effectively next-gen third party tracking. It bothers me how blatant and obnoxious the spin is around such pressure on businesses to share their customer data with Google, when it wasn't shared before for the purpose of tracking. "Don't worry, it's hashed data" is not good enough, and not the point.
> 5. Activate Google Signals very low effort (If applicable; very low effort)
The fact that Foo1 and Foo2 are both called "Foo" is an artefact of how humans write software, but its meaningless to the computer. Foo2 usually has all the features of Foo1, but because the API is programmatically incompatible, when it comes to the dependency tree, they should be totally separate items.
You see forms of this all over the place. Eg, HTTP APIs are usually /api/v1/ not /api/v1.2/. Debian's apt repository has packages for 'llvm-10', 'llvm-11', 'llvm-12', and so on. I used to think this was silly; but now I think it might be the only way to do that in a coherent way.
> For example, let’s say you support click^7. Click 8 comes out, and someone writes a library requiring click^8. Now your library can’t be installed at the same time as that other library, your requirements do not overlap.
What I'm arguing for is that click^7 and click^8 should be seen as different packages that (like any other packages with different names) can be installed side-by-side without issue. It sounds weird, but it makes this problem goes away entirely.
Nodejs and (I think) cargo both support this. I've never run into errors like this in rust or node.
Another advantage of this approach: Integration tests in all previous versions of a package should pass on the current version of that package. I'd love to see test runners which actually enforce this - which should be pretty easy to do in a language like rust where integration tests are clearly separated.
Sometimes (rarely) it makes sense to allow an integration test for react16 v1.0 to fail on react16 v2.0, but that implies a violation of semver compatibility. So it should be the exception. Not the default.
I'm going to try being even more clear:
"Package Foo v1.2.3" should usually be spelled "Package Foo1, v2.3"
For example, the migration from react 17.x.x to react 18.x.x is actually a change from depending on package react17 to depending on react18. React17 and react18 are two different packages which have been designed by the same team, and happen to have similar APIs in order to make migration easier.
If I designed npm today I'd make "react16", "react17" and "react18" all be essentially different packages (under a namespace). They should probably only share their lists of authorized authors.
You can see this confusion all over the place. For example, when setting up peerDependencies in a nodejs project, there are suggestions[1] to use a version like `"chai": ">= 1.5.2 < 2"`. This is because chai v1 and chai v2 are essentially different packages. You want to say "chai v1 should be a peer dependency". But its unintentionally awkward to say that clearly. And with the current nomenclature, there's no way to express a dependency on both chai1 and chai2 - despite them being different, incompatible packages.
Even though nodejs is obsessed about semver, it still makes a mess of things by pretending as if major versions and minor versions are similar. They are not.
[1] Last couple paragraphs of https://nodejs.org/es/blog/npm/peer-dependencies/
And promotion and bonuses! This is the major impetus for these efforts.
From top of mind it used to be Urchin Analytics, then Google analytics, then universal analytics and now GA4.
But these Windows go to 11?
They keep adding layers of opacity to mask ever shrinking ROIs while making it easier for anyone to spend more on ads. Eventually they will get to the point where you set a budget and you get what they are willing to provide based on what their "AI" decides is best to maximize their fill rates, auction participants and their earnings per page view. No keywords or managed placements. It isn't like there is an alternative and their inertia isn't going anywhere.
I run a side business[1] that is lucky to not be dependent on search ads. We started using GA 10-ish years ago because it was free and took almost no effort to set up. But over the years it just got more and more complicated to use, to the point where it got hard to act on the results.
Last year, I switched to Plausible[2] for data protection reasons. The unexpected side effect was that I suddenly understand every single metric in my analytics tool. As a result, I use it a lot more for actual decision-making.
[1] https://push-conference.com
edit: typos
I miss seeing the user flow from page to page, but that’s it.
On my personal website, I just don’t use any analytics at all. It’s not a metric I want to be exposed to.
I’m glad that there are more and more such simple, single-purpose tools now. It took me a long time to learn this, but I’m now actively staying away from anything that markets itself as “enterprise-ready”. Too much hidden cost in keeping up with all the shifting trends and mental models.
I am not sure if I like it or if I hate it.
I agree that less features is upside for better understanding of all collected data. Simple dashboard is all I need and I guess it is same for most of websites.
As far as I can see, there's no event support, you can't filter user journeys who went to a specific page or see heatmaps.
I thought those counters were just for fun.
We’ve been creating and including both with gtag(‘config’, ‘ID’) lines for some time now in anticipation of something like this.
All I hear is:
> We will begin sunsetting Universal Analytics — the previous generation of Analytics — next year. All standard Universal Analytics properties will stop processing new hits on July 1, 2023
Great, plenty of time to replace GA with another analytics solution.
IMO, the best analytics for many businesses is customer acquisition and revenue as well as customer feedback. If you aren't adequately responding to those inherent analytics then you shouldn't yet be using The Google's form of black magic.
That doesn't mean everyone will have the same experience, but it does mean most people probably don't need to feed into Google's ability to track everyone just to look at their stats and say "yep, that's where I'd expect a person to come from."
Your website has GA4 installed on it, you might want to remove that.
Any suggestions or recommendations? I'd be keen on trying something new.
They also have a nice Fediverse presence and quite open about what they do even down to their user stats.
We're building the most comprehensive, but still easy-to-use analytics tool out there. Specifically useful for SaaS businesses that want to collect and unify marketing, product engagement, sales, and subscription revenue data.
I've seen tons of similar analytics products pop up in the last year and have avoided them all for having out-of-country data storage (and usually less functionality, etc). With how volatile data-related laws have been in the EU the past few years, IMO it feels risky to try to store anything there, let alone view it as a feature!
Fewer jurisdictions is almost certainly going to be easier, and these options offer that for the European market.
The fear is that you would now have to adhere to those "strongest laws" by hosting your data there.
There is one point from the Google announcement that I fully agree with and that is that the future of analytics is cookieless by design (only google won't be a part of this)
Edit: I am asking because I know that there are better alternatives than matomo. But there still needs to be some alternative for 'oldschool' hostings with LAMP stack.
[1]: https://matomo.org/faq/new-to-piwik/how-do-i-use-matomo-anal...
i.e. instrument a standard spec with something like www.Rudderstack.com and write it to your warehouse/datalake + whatever analytics tool you want.
That way if you ever want to move off your analytics tool (or a solution providing the data spec), you still own the data for whatever you replace it with.
By sending the data to the WH, also lets you own any proprietary data you create - like the custom PPC model referenced in this comment: https://news.ycombinator.com/item?id=30707913
full disclosure - Rudderstack employee
I'm moving my clients to Matomo and they're loving it.
GA's second-biggest moat (behind being free) is its existing install base. It's already in place on approximately every website on the Internet. And no one wants to change analytics vendors, because that breaks history, which is a no-no. You want to know if your visits are up/down vs last month or last year, you literally cannot answer that unless you are using the same tool this years as you were last year.
By forcing a migration to a non-compatible version, Google loses that big advantage. If you're going to lose that anyways, then there's less of a reason to stick with GA. Now you're forcing people to gut-and-replace their existing data collection process, what incentive is there for people to stick with Google? One of the biggest reasons to stay with GA was just killed.
The timing is also exceptionally atrocious. Google is, again, forcing people to expend the effort to rebuild their analytics implementations from scratch. Meanwhile court rulings are coming down on a monthly basis enumerating different ways in which Google Analytics violates GDPR. A lot of companies might just brush off these rulings by simply doing nothing, but now they're being forced to actively invest time and effort into Google Analytics. A lot of them are going to be doing an evaluation against privacy-conscious competitors that simply would not have happened if Google wasn't kicking people of their existing platform.
The GA4 data model is also completely different from the GA Universal data model, which is an impediment to their fourth-biggest moat, the fact that every piece of BI software in the universe has a GA plug-in. Many of those plug-ins work despite not being actively maintained since 2015; the GA4 ecosystem is smaller and will take some time to recover.
(The third-biggest moat is Google Ads integration. I suspect that Google imagines this to be bigger than it really is in practice.)
At that point we instituted a big program to move to Amplitude (which we’ve been fairly happy with, but probably won’t be perfect until we get our own data overlays through eventually getting a data lake etc etc) - now the move BACK to GA … II would never touch GA again or consider it for a new project
No, this is the same migration. This is my point: you weren't actually forced to migrate until now. The migration has been optional up until this announcement, and many people haven't. What your team did 2-3 years ago, the majority of organizations have been putting off. But now they can't anymore.
So they've basically said 'GA is closing unless you want to pay 120k/pa so move to firebase - GA is now Firebase Analytics' to now 'Firebase Analytics is Google Analytics again, but that thing we said you had to do you now have to definitively do?' ?
...and there's been very little movement. We can still create "universal analytics" properties, we still can't do the same stuff in Firebase, and the GA4 "Measurement Protocol" is still in beta.
I think our eventual move is going to be to a non-Google product.
There’s heaps of good ones around now - our shift to amplitude was a good one but I know there are a lot of other really simple and free ones about too
On the website side of things, which is GA's core user-base, Universal Analytics has remained supported up til this announcement.
We're actually not currently using a data lake. We were initially on the startup plan which was fantastic (the ability to do templated reports etc is a big one for us) but currently we couldn't justify the cost of a full subscription at that point (and now we are just making do until we can complete our next round at which point we will re-look at the whole situation)
We love that you have an API and used fivetran to ETL to (for want of something cheap + low BI tool costs) BigQuery where we were hoping to be able to attach Data Studio to it to achieve our objectives.
This may not have been the right approach.
We couldn't spare anyone from our dev team to take the time to work out the transforms/queries required to make it useful in Data Studio so we spent some time contracting the work out, but that fell over because they were on lend from a mate's startup and I think just never moved out of the blocks. Also maybe we were using the incorrect tools.
Essentially we were looking to get a fair proportion of our data out, in order to create data views for our clients on their usage (... platform in bio, but hospital analytics stuff - no patient info, all operational efficiency). This was fulfilled really nicely by Templates but in the reports we now generate (... a bit laborious but using a offshore contractor who takes all the pain out of it) we are also able to put benchmarking averages against their peer hospitals etc in which wasn't something it was immediately obvious we could achieve with Templates. Ultimately we want those individual dashboards available for the Hospitals to view, as well as be able to return things like views per period for their content on our dashboard, which we know we can also achieve through the API
Which is an abstraction that tries to handle old GA use-cases and the Segment/Amplitude style use-cases but does neither well.
This is precisely why they're doing it with urgency though. These rulings are based on Universal Analytics. GA4, along with their Consent Mode (still in beta) is supposed to be designed to work entirely without cookies and IP addresses.
More confusing controls and useless new marketing terms!
Even more useless dat ranges!
It's all here!! The tool you don't use, but we do!
Looking forward to not using it. :I
Not sure if it will be worth my time to go back and remove the old tags from those sites. Cannot see a reason to upgrade.
It was nice to know, but there are more important metrics.
The multivariate testing suite was fun at one time. However it isn't as useful when your core users are all blocking it.
Honestly though, for a huge segment of their users this hides/obscures the number one metric they look for. I understand that with SPA"s and stuff things are much more event driven, but they're absolutely going to lose some users over this.
Just take a look at https://support.google.com/analytics/answer/9267735?hl=en
It became useless to me and I haven't used it in at least five years.
If there's anyone here who still uses GA, can you tell me if they ever solved the problem of referrer spam?
At minimum they need a disavow tool.
Google Analytics 4 has a few changes that should improve their privacy aspects:
IP Anonymization: not GDPR proof
Server location: not changeable
Use of Cookie Banner: still needed
Deleting Individual User Data: improvement
Limited Data Storage: slight improvement
In short, they say they fix a lot of privacy issues, but they don't really address the real issues like the use of IP addresses and the location of your data. [1]
Disclosure: I'm the founder of Simple Analytics. [2]
[1] https://blog.simpleanalytics.com/google-to-sunset-universal-...
Apparently they are updating GA4 to remove all use of IP addresses entirely: https://www.adexchanger.com/online-advertising/google-analyt...
> Use of Cookie Banner: still needed
It still works without cookie consent though, once you incorporate their Consent Mode.
It is not clear what actually works when you incorporate Consent Mode. I haven't actually found any reports that include the data supposedly collected. I also have some doubts that Consent Mode does not actually require consent (see IP Address concerns above).
Of course, any network request is going to send an IP address... that's how the Internet works. If that alone is enough to make a service illegal then third-party analytics is going to be just flat out impossible for everyone.
Yup, that's my point.
> If that alone is enough to make a service illegal then third-party analytics is going to be just flat out impossible for everyone.
Yup, that's my point.
EU court rulings say it's illegal to transfer data to Google, because Google is subject to the CLOUD Act in the United States. IP Address is personal data, and is considered transferred just by having the user's client connect to Google's servers.
Google's description of what they will or will not do with the IP Address is legally invalid, because the CLOUD Act can compel them to break their promises. This ruling, while unfortunate, is correct.
Third-party analytics in general are not illegal. What's illegal is third-party analytics (or third-party anything) hosted by a US company subject to the CLOUD Act. Even then, it's possible to make it legal by running it through a self-hosted proxy which anonymizes the data before sending it over to Google. But a direct connection from the user-agent to Google servers is illegal.
Really poor product planning.
[0] https://pirsch.io/blog/google-analytics-declared-illegal-by-...
[1] https://european-alternatives.eu/category/web-analytics-serv...
Seems like a good time to make a break from GA and do your part to reduce centralized surveillance on the web.
Strangling it to death with ad blockers isn't working (or at least is doing less than the self inflicted damage being caused by mismanagement over at Google).
How can we make its output completely wrong, instead of just incomplete?
I'll happily stick a "do not track” header (or the new one) on the bogus requests. In my book, they've been warned to back off at that point, so any bad outcomes from tracking the requests are on them.
Any way to export historical, basic data like what you'd find in the "Audience > Overview" tab?
I should have expected this from Google, but damn, losing years and years of historical data is going to really hurt.
Couldn't be more happy.
I'll in the process of looking alternatives as GA4 is a step back. I suspect alternate analytics companies are going to love Google for this decision.
Oh yes! https://twitter.com/MarkoSaric/status/1504121261915574273
Matomo looks more like Google and the data is yours. Plausible and Simple Analytics are more similar to one another in the sense that they only collect privacy-insenstive data.
The main difference is that plausible still operates in a 'grey-area' by anonymizing IP addresses for 24 hours. This can be considered fingerprinting.
The thing is that "GA alternative" is a broad question, as GA has A LOT of features and it's also lacking many features, it depends on what you're looking for and what's your use case.
That seems to be what they're doing?
This is not an "upgrade." This is "replace product with a different product in the same space." Old GA data is incompatible with GA4 data, and there is no way to bring data from one into the other. Google is discontinuing Google Analytics, and applying the Google Analytics brand to Firebase Analytics.
Nothing carries over. Configuration, implementation, goals, user permissions, historical data, nothing. It all has to be re-built from scratch.
The Google deprecation team has really outdone themselves this time. I really would have thought they'd deprecate Gmail before Google Analytics! I've got to imagine there are legal reasons behind this, because no product team could sincerely believe that this is a good idea for the product, right?
Yes. That's what the original post and the entire discussion here is about.
I imagine you'll hear a lot of that in the near future
It's like when, e.g., Heroku sends me an email saying I need to move apps off of a stack they're retiring. They're not killing off my app; I just need to run the command or agree to a maintenance window so it keeps working after a certain date. If I for whatever reason refuse to do anything, I can't seriously claim they're shutting me down. Especially with a long notice period.
"Doing so will allow you to build the necessary historical data and usage in the new experience, preparing you for continuity once Universal Analytics is no longer available."
That doesn't seem to indicate your old data comes with.
Anyway, I don't mean to take away from your moment to hate on Google. :) It's just not a classic Google discontinuation like when they eventually shut down Stadia or whatever.
Nobody wants to move to GA4. It has nothing UA doesn't, and UA has lots of features GA4 doesn't. If they're sunsetting UA next year, there's no way it'll have improved enough by then to be a fully functioning replacement. It has a cleaner structure than UA, because it's dropping a decade of cruft, but that doesn't really help when it just doesn't do as much.
The people that are moving to GA4 are running both UA and GA4 in parallel, so that they'll have a history set when UA shuts off. Those are companies that are paying attention. I would be surprised, by the 2023 date, if even 25% of the UA installations have switched to GA4. (Of course, all the larger companies will have.)
Side note... I looked into Shopify a month or two back. They have a drop in UA implementation that is almost as easy as clicking on a checkbox. Not only do they not have a wizard for implementing GA4, they say it's not even possible for you to add manually. No due date was given. A lot of sites have UA built in as the assumed forever default.