38

Update Sep 14: GA4 has been enabled network-wide.

Update Sep 8: GA4 has been enabled on Super User.


TL;DR: We’re preparing to upgrade Google Analytics to version 4.

A few months ago, the Google Analytics (GA) cookie was recategorized as a “Performance Cookie”. Users who have access to site analytics saw a drop in reported page views for their respective communities. Earlier this year, Google announced that Universal Analytics (UA) will be going away and will stop processing new data after July 1, 2023. Our longer term solution was to migrate to the latest version of Analytics, Google Analytics 4 (GA4), which includes more privacy controls with Consent Mode and a default IP anonymization feature out of the box, which means that GA4 will not store the IP addresses of users.

In this post, we will provide a brief overview of the Consent Mode feature, what changes will be made, and a staged rollout plan for when the upgrade will take place across Stack Exchange network sites.

What is Consent Mode?

Consent Mode is a beta feature that adjusts the behavior of Google tags based on the visitors’ consent choice of granted or denied as it relates to their use of non-essential cookies which are disabled by default. Until consent has been granted, Google will receive cookieless pings that communicate the consent state and user activity, instead of storing cookies.

These pings include some functional and aggregate, non-identifying information which enables Google to fill in the data gap using machine learning to model users who decline cookies against similar users who accept cookies. What this means for us is we will be able to better understand user behavior and any other aggregate traffic data, without collecting any identifying information.

What changes will be made?

We will be running our existing UA integration in parallel with GA4. We are still relying on our existing analytics to report on high-level aggregate metrics and integrating GA4 in parallel gives us time to verify the implementation and start building the historical data. The reason for this is because it is not possible to migrate our existing data from UA to GA4 due to the differences in data schema and dimension definitions.

For UA, the analytics cookie will still be classified as a “Performance Cookie”. If consent has not been provided, data will not be sent to Google Analytics. This is the current behavior and no changes have been made.

With GA4 and Consent Mode enabled, the Javascript snippet will always be loaded across the pages you visit regardless of the consent status. What this means is you will see cookieless pings to Google Analytics for each page you visit; however, data will not be stored or read from browser cookies (e.g. _ga cookie) until consent has been granted. Your consent status will continue to persist across every page you visit. For more information about Consent Mode behaviors, please refer to Google’s Help Center article.

For a high level overview of our existing analytics setup and what our setup will look like with GA4, please refer to the diagrams below.

Here’s a simplified flowchart of our existing UA setup:

Current Universal Analytics setup

And here’s what our setup will look like when we run both UA and GA4 in parallel:

Universal Analytics and Google Analytics 4 setup

Timeline

Here’s the rollout plan, just so you know when these changes will be made:

WHERE WHEN STATUS
Super User Week of Sep 5 (or sooner) Live as of Sep 8
Network-wide Week of Sep 12 Live as of Sep 14

While there will not be changes to the Privacy Policy you can also expect to see some updates to our Cookie Policy page in our legal portal in the near future.

If you come across any bugs related to GA, please add an answer (one answer per bug) so that we can triage and prioritize them accordingly. We’re also happy to answer any questions you may have regarding the GA4 upgrade.

2
  • 7
    Before the 'actual' cutover - will the analytics visible to high rep users and mods be based off the old or new analytics? Commented Aug 31, 2022 at 15:09
  • 1
    @JourneymanGeek great question, but I don’t have a good answer at this time. I will be working with the team responsible for the site analytics tool to understand how it currently works before we can figure out what changes will need to be made to minimize impact for those who rely on the tool today. I know this isn’t the answer you’re looking for, but the first step is to conduct an investigation.
    – tanj92 Staff
    Commented Sep 2, 2022 at 15:09

7 Answers 7

85

Am I understanding this correctly, the new GA4 system will still send information about every page view to Google for all users even if they do not give consent?

From Google support:

The pings described above can include:

  • Functional information (such as headers added passively by the browser):
    • Timestamp
    • User agent (web only)
    • Referrer
  • Aggregate/non-identifying information:
    • An indication for whether or not the current page or a prior page in the user's navigation on the site included ad-click information in the URL (e.g., GCLID / DCLID)
    • Boolean information about the consent state
    • Random number generated on each page load
    • Information about the consent platform used by the site owner (e.g. Developer ID)

And of course the IP address will be sent to Google as well, that's a side effect of every HTTP request. Though they might not store it without consent, not sure about the details here.

The privacy issue never was cookies themselves, it was tracking. Especially tracking across different sites by powerful entities like Google. The absence of a unique identifier makes tracking users a lot harder, but likely not impossible for a company like Google.

From a privacy perspective this change is clearly worse than disabling GA entirely without consent. I understand why SE is doing this, as it restores large parts of the analytics they lost by disabling GA without consent. I'm not very happy about this, but I assume this is reasonably legal. The interesting question is one nobody here can answer anyway, whether Google can and actually does identify users without using the tracking id like in previous versions.

12
  • 7
    Wasn't it the user agent that was almost unique if you include all the add-ons, OS version and other details? Commented Sep 1, 2022 at 11:05
  • 73
    "OK, so we heard y'all don't like cookies, so we're going to do exactly the same thing cookies do, but without cookies. Are you happy now?" No information is non-identifying when aggregated enough to track someone's activity online. Yes, I understand why companies want to do this, but I don't care. In the olden days, media managed to fund itself without knowing that I was looking for eggplant recipes yesterday. Figure out a better way.
    – ColleenV
    Commented Sep 1, 2022 at 14:38
  • 5
    Mad Scientist, yes, with latest version of Analytics (GA4), some non-identifying information is still sent with every page view even if users do not give consent, which has been outlined in the “What is Consent Mode?” section as well as in the flowchart for when we run both Universal Analytics (what we use today) and GA4. In GA4, IP anonymization is activated by default and Google recently announced that it has taken it a step further, with GA4 no longer logging or storing IP addresses. Additionally, the information collected by GA4 will now be regionalized.
    – tanj92 Staff
    Commented Sep 1, 2022 at 14:44
  • 13
    I'll just leave this here en.wikipedia.org/wiki/Device_fingerprint
    – jaskij
    Commented Sep 1, 2022 at 21:33
  • 1
    I don't understand what you mean when you say "disabling GA entirely without consent". Are you referring to end users disabling Google Analytics? Or to SO/SE disabling it? Why is consent needed? Where does that come into play? I don't see how this is any worse for tracking. It increases the anonymization of individual users, which seems like a win for privacy.
    – TylerH
    Commented Sep 2, 2022 at 14:02
  • 4
    @TylerH It is more tracking than not sending data to GA at all (which is what SE is doing right now if you don't consent). It is a lot better than sending unique tracking beacons as they did (likely illegally) before switching off GA. Commented Sep 2, 2022 at 14:05
  • @MadScientist Thanks, but what about "disabling GA entirely without consent"? Are you thinking they need your consent to disable Google Analytics on their own websites?
    – TylerH
    Commented Sep 2, 2022 at 15:13
  • 2
    @jaskij, According to this help center article, "Google doesn’t allow fingerprint IDs or other attempts to identify individual users. Instead, Google aggregates data (such as historical conversion rates, device type, time of day, geo, etc.) to predict the likelihood of conversions." support.google.com/analytics/answer/10710245?hl=en
    – tanj92 Staff
    Commented Sep 2, 2022 at 15:27
  • 30
    "We don't let people identify individual users with IDs. We just accumulate so many small bits of information about them that an AI can distinguish that person from someone else fairly accurately." There is more to protecting someone's privacy than making it hard for a human to look at the raw data and know who that person is. I realize I'm tilting at windmills, but I am compelled to keep pointing this out. The problem isn't "How do we preserve someone's privacy and know everything we want to know to market to them." The problem is "Why do you have to know all that to market to me?"
    – ColleenV
    Commented Sep 2, 2022 at 19:16
  • 1
    @TylerH I believe "disabling GA entirely without consent" means that some time ago, SE disabled the use of GA entirely, unless or until a user gave their consent. My guess is that prior to that time GA was active unless a user turned it off. Presumably this resulted in much less GA data for SE.
    – TripeHound
    Commented Sep 2, 2022 at 21:28
  • 6
    @TripeHound GA was actually active even when the user did not give consent (meta.stackexchange.com/questions/377822/…). SE fixed that at some point after I complained about it and disabled GA entirely for users that had not consented to tracking. So this change means that instead of no GA at all, users that did not consent get the reduced tracking GA4 implements. Commented Sep 2, 2022 at 21:57
  • 3
    So having ads perform device fingerprinting wasn't enough? Commented Sep 14, 2022 at 21:04
64
  • Have you considered other analytics systems?
  • If so, what does Google Analytics have that's important enough to send all user traffic to Google?

Google's understanding of what is and isn't okay has been challenged repeatedly by data protection authorities. If I understand your plan correctly, you do not intend to ask for any consent before transferring traffic and usage data to Google.

2020-03-11: The Swedish Data Protection Authority imposes administrative fine on Google:

The Swedish Data Protection Authority imposes a fine of 75 million Swedish kronor (approximately 7 million euro) on Google for failure to comply with the GDPR. Google as a search engine operator has not fulfilled its obligations in respect of the right to request delisting. […] Google does not have a legal basis for informing site-owners when search result listings are removed and furthermore gives individuals misleading information by the statement in the request form. That is why the DPA orders Google to cease and desist from this practice.

2022-01-12: Austrian DSB: EU-US data transfers to Google Analytics illegal:

[…] die Standarddatenschutzklauseln, die die Erstbeschwerdegegnerin mit dem Zweitbeschwerdegegner abgeschlossen hat, kein angemessenes Schutzniveau gemäß Art. 44 DSGVO bieten, […] die Maßnahmen, die zusätzlich zu den in Spruchpunkt 2. b) genannten Standarddatenschutzklauseln getroffenen wurden, nicht effektiv sind, da diese die Überwachungs- und Zugriffsmöglichkeiten durch US-Nachrichtendienste nicht beseitigen,

Soweit die Beschwerdegegner ins Treffen führen, dass keine „Mittel“ verwendet würden, um die hier gegenständlichen Kennnummern mit der Person des Beschwerdeführers in Verbindung zu bringen, ist ihnen neuerlich entgegenzuhalten, dass die Implementierung von Google Analytics auf www.[REDACTED].at eine Aussonderung iSd ErwGr 26 DSGVO zur Folge hat. Mit anderen Worten: Wer ein Tool verwendet, welches eine solche Aussonderung gerade erst ermöglicht, kann sich nicht auf den Standpunkt stellen, nach „allgemeinem Ermessen“ keine Mittel zu verwenden, um natürliche Personen identifizierbar zu machen.

2022-02-10: Use of Google Analytics and data transfers to the United States: the CNIL orders a website manager/operator to comply:

The 10 February 2022, the CNIL, which was cooperating with its European counterparts, has issued and order to comply to several organizations using Google Analytics because of illegal transfers of data to the United States. […] A simple change in the tool's settings is not enough

2022-04-22: Austrian DPA rejects “risk based approach” for data transfers to third countries:

In seiner Stellungnahme vom 9. April 2021 im Parallelverfahren zur GZ: DSB- D155.027 hat der Zweitbeschwerdegegner bei Frage 9 zwar vorgebracht, dass er eine derartige Information nur bekommt, wenn gewisse Voraussetzungen erfüllt sind, wie etwa die Aktivierung von spezifischen Einstellungen im Google-Account. Nach Auffassung der Datenschutzbehörde vermag dieses Vorbringen nicht zu überzeugen. […] Dabei kann das Vorbringen der Beschwerdegegner rund um die „Anonymisierungsfunktion der IPAdresse“ dahingestellt bleiben, da die vollständige IP-Adresse jedenfalls für einen gewissen – wenn auch sehr kurzen – Zeitraum am Server von Google LLC verarbeitet wird. Dieser kurze Datenverarbeitungszeitraum ist ausreichend, damit der Tatbestand des Art. 4 Z 2 DSGVO erfüllt wird.

Es ist nämlich nicht erforderlich, dass die Beschwerdegegner jeweils alleine einen Personenbezug herstellen können, dass also alle für die Identifizierung erforderlichen Informationen bei diesen sind (vgl. die Urteile des EuGH vom 20. Dezember 2017, C-434/16, Rz 31, sowie vom 19. Oktober 2016, C 582/14, Rz 43). Vielmehr ist ausreichend, dass irgendjemand – mit rechtlich zulässigen Mitteln und vertretbarem Aufwand – diesen Personenbezug herstellen kann (vgl. Bergauer in Jahnel, DSGVO Kommentar Art. 4 Z 1 Rz 20 mVa Albrecht/Jotzo, Das neue Datenschutzrecht der EU 58).

2022-06-23: Italian SA bans use of Google Analytics No adequate safeguards for data transfers to the USA:

The Italian SA found that the website operators using GA collected, via cookies, information on user interactions with the respective websites, visited pages and services on offer. The multifarious set of data collected in this connection included the user device IP address along with information on browser, operating system, screen resolution, selected language, date and time of page viewing. […] In determining that the processing was unlawful, the Italian SA reiterated that an IP address is a personal data and would not be anonymised even if it were truncated – given Google’s capabilities to enrich such data through additional information it holds.

2022-07-14: Datatilsynet nedlægger behandlingsforbud i Chromebook-sag

”Der er risiko for, at Google eller andre tredjeparter anvender personoplysninger om lærere og elever til brug for markedsføring eller andre formål, som Helsingør Kommune som dataansvarlig ikke ønsker, at personoplysninger skal behandles til. Særligt kontaktoplysninger, IP-adresse og digitale spor (almindelige oplysninger) er relevante i denne sammenhæng. Det bemærkes, at der er tale om personoplysninger relateret til elever, der nyder en særlig beskyttelse i henhold til databeskyttelsesreglerne, hvorfor adgangen til og behandlingen af personoplysninger om eleverne udgør et yderligere element i relation til risikovurderingen.”

I det konkrete tilfælde, hvor det er alment kendt, at de teknologier, der benyttes til levering og systemunderstøttelse af den valgte service – Google Chromebooks og Workspace for Education – også benyttes til at levere øvrige dele af Googles produkter, og disse benyttes til oplysningsindsamling, målrettet markedsføring og salg af disse oplysninger. […] Datatilsynet finder, at Helsingør Kommunes risikovurdering ikke til fulde dokumenterer de risikoscenarier, som kan opstå som følge af databehandlerkonstruktionen og de foretagne systemvalg.

2022-09-21: Press release: Use of Google Analytics for web analytics

The Danish Data Protection Agency has looked into the tool Google Analytics, its settings, and the terms under which the tool is provided. On the basis of this review, the Danish Data Protection Agency concludes that the tool cannot, without more, be used lawfully. Lawful use requires the implementation of supplementary measures in addition to the settings provided by Google.

“Since the decisions by our European colleagues, we have looked into the tool and the specific settings available to you when you intend to use Google Analytics. This has been particularly relevant as Google, following the first Austrian decision, has begun to provide additional settings in relation to what data can be collected by the tool. However, our conclusion is that the tool cannot, without more, be used lawfully.”

If you go ahead with this plan, will you at least consider respecting the Do Not Track header?


I've had a little look at Google Analytics 4, and – based solely on what Google claims about its functionality – it's not something you can opt users into without their consent.

The Internet Archive did not let me save a current version of the User exploration page, so I've documented the differences between the current version and that version:

Before:

To see all of a user's event data identified by only the anonymous ID, you need to change the reporting identity from By User-ID and device to By device only:

Current:

To see all of a user's event data identified by only the anonymous ID, you need to change the reporting identity from Blended Identity which factors in User ID, Google Signals and Modeling to By Device only:

Aside from that, the page is identical, so I will quote:

When a user is logged in, event data associated with that user is identified by the user ID you provide. When the user is logged out, that user's event data is identified by an anonymous ID.

  • Do you intend to give Google our user IDs, in association with our activity?

This information should be written in your privacy policy, but it's not. And the point is somewhat moot, since Google could easily deduce such information from just an "anonymous" ID.

I'm no Google Analytics expert, but it feels like you'd need to give them some ID anyway, so you could comply with GDPR Article 17 "Right to erasure" requests. It's a bit of a catch-22 for you, though: if you give Google this data, then you're giving them explicitly personally-identifying information that they can use to connect more dots about us. If you don't, though, what Google gets is already personally-identifying to Google – but we have no way of requesting it deleted, because you can't identify which information is ours!

Speaking of, [GA4] Data-deletion requests said (in June):

If you need to delete data from the Analytics servers for any reason, then you can use a data-deletion request to issue a request for its removal.

You can have a maximum of 12 active (in grace period / pending deletion) requests per property at any one time.

You can cancel any request within 7 days from the time it was created. Use the request-details page. During the first 7 days, you can also see a preview of the effects of the deletion in your reports and in Explorations.

A data-deletion request can take between 7 and 63 days to be processed, depending on the amount of data to be deleted, and data must be more than 12 days old before it can be deleted.

If you're using consent mode for your property, you may need to add seven calendar days to the end date of your data-deletion request to ensure that any value you are specifically targeting for deletion is deleted from behavioral models. This is because Analytics trains machine-learning models on data from consented users (i.e., the data you may be trying to delete) and approximates future user behavior based on that training data.

This raises red flags. Are "consented users" actually consenting to this use of their data? Your current consent system isn't consent, and the big button is labelled:

Accept all cookies

with the small print:

Your privacy

By clicking “Accept all cookies”, you agree Stack Exchange can store cookies on your device and disclose information in accordance with our Cookie Policy.

Your cookie policy says nothing about training machine learning models. Even if you stretch and assume people consent to what's in the cookie policy, they're not consenting to this. This terminology is misleading.


I've said all this before, more than once, so this time I'll just quote the UK's ICO.

How do we comply with the cookie rules? | ICO:

A number of services exist that provide an analytics function, and it could be easier for you to use these instead of building your own. However, it can be more difficult to obtain consent for third-party analytics cookies as there is no direct relationship between the third-party organisation and the user of your site. In these cases you need to ensure the information you provide to users about these cookies is absolutely clear and is highlighted in a prominent place – for example you can’t just include it through a general privacy policy link.

What are cookies and similar technologies? | ICO:

Functions usually performed by a cookie can be achieved by other means. This could include, for example, using certain characteristics to identify devices so that visits to a website can be analysed.

PECR applies to any technology that stores or accesses information on the user’s device. This could include, for example, HTML5 local storage, Local Shared Objects and fingerprinting techniques.

Example

Device fingerprinting is a technique that involves combining a set of information elements in order to uniquely identify a particular device.

Examples of the information elements that device fingerprinting can single out, link, or infer include (but are not limited to):

  • data derived from the configuration of a device;
  • data exposed by the use of particular network protocols;
  • CSS information;
  • JavaScript objects;
  • HTTP header information,
  • clock information;
  • TCP stack variation;
  • installed fonts;
  • installed plugins within the browser; and
  • use of any APIs (internal and/or external).

It is also possible to combine these elements with other information, such as IP addresses or unique identifiers, etc.

PECR also applies to technologies like scripts, tracking pixels and plugins, wherever these are used.

From now on, this guidance uses the single term ‘cookies’ to refer to cookies and similar technologies that PECR applies to, […]

Let's compare this to Mad Scientist's non-exhaustive list of what Google claims they collect from "non-consenting" users. Or maybe you could look at what the Austrian DSB has found that Google Analytics actually collects?

7
  • 4
    I haven't included the formatting from those quotations; quite a bit of it is underlining-for-emphasis, which Stack Exchange's markup doesn't support. I recommend reading the original rulings, which are – like the text of GDPR itself – quite clear on these matters.
    – wizzwizz4
    Commented Sep 2, 2022 at 17:23
  • 16
    At this point, I don't even know why I bother. I wonder whether I'll get directed to send an email to [email protected] again. I wonder whether they'll close the ticket, having responded to just one point, and not respond to my reply. I wonder whether anyone at the company actually cares about their users' right to not have their metaphorical library records sent to one of the world's largest surveillance corporations – or is our exercising of choice to inconvenient for you?
    – wizzwizz4
    Commented Sep 2, 2022 at 20:01
  • 7
    You can't "fill in the data gap" like this, anyway. The class of users who clicks "Accept all cookies" is not typical of your audience. That missing data is missing.
    – wizzwizz4
    Commented Sep 2, 2022 at 20:02
  • 7
    I suspect this is perceived by management as Google's problem, as they would be the ones who would get in trouble. It's not like they're doing this out of altruism. They're just trying to find a way to get analytics back without breaking the law themselves.
    – trlkly
    Commented Sep 3, 2022 at 3:08
  • 5
    @trlkly There are many, many other analytics providers, even if they don't want to self-host one of the half-dozen free software implementations (which run on Windows Server, or in Linux containers!). Maybe Google Analytics has something special that justifies using it, but nothing I've seen in the Site Analytics (20k or moderator) is GA-specific stuff.
    – wizzwizz4
    Commented Sep 3, 2022 at 11:07
  • 7
    Thanks for posting, "Dieser kurze Datenverarbeitungszeitraum ist ausreichend" (translated: "this short time during which [Google] processes the IP is enough"...) says it all for me.
    – bad_coder
    Commented Sep 4, 2022 at 2:46
  • 2
    An alternative: matomo.org
    – Florian F
    Commented Sep 4, 2022 at 10:21
49

When I set the "no not track" flag in my browser, that means: do not tell Google, Facebook, Twitter or any of these that I am currently visiting this page.

Is that so difficult to understand?

So, when you see the "do not track" header, just do not include any request whatsoever that lets Google know this PC with this IP address just visited this page. Any piece of code that GA asks you to include in every page, don't include it.

The GA cookie was recategorized as a “Performance Cookie” to allow people to opt out. When you see a drop in reported page views, that means people succeeded in regaining some privacy. What you are doing by upgrading GA is to help Google to revert the situation and keep doing what people request them not to do.

When we say "no", that means "no".

You will say my data is anonymized. But If you send any request to Google, Google can identify you by fingerprinting your browser. And if they can identify you, they will. Because that's what they do.

7
  • 4
    While true, knowing amount of visitors is crucial for websites, it's a business after all. We can't ignore it. And we want the business to be successful, or at least survive. So it all comes down to trust, at some point we have to trust a business it won't use our private data. Commented Sep 4, 2022 at 10:13
  • 19
    @ShadowTheKidWizard Stack Exchange already has an in-house view tracker, though; the view count of every question is known. They already have enough information to know the number of visitors. If they need more information, there are ways of getting it other than Google Analytics.
    – wizzwizz4
    Commented Sep 4, 2022 at 10:53
  • 2
    @wizzwizz4 views count is ancient relic, it's not reliable enough, and doesn't count unique visitors which is the most crucial info, I believe. Of course there are other ways, but who says they're more secure than Google? If you mean build in-house the full capabilities of GA, well, guess it's possible but requires way too much development efforts. Commented Sep 4, 2022 at 10:58
  • 24
    @ShadowTheKidWizard Fathom, Plausible and Matomo have not been repeatedly fined for GDPR violations, and the use of the softwares have not been ruled unlawful. All three can be self-hosted, gratis, no development effort needed (except re-doing the integration). Matomo has most of the features that Google Analytics has.
    – wizzwizz4
    Commented Sep 4, 2022 at 11:01
  • 1
    @wizzwizz4 Thanks for sharing the alternatives to GA4. It would be great if you can post a feature-request regarding asking SE to shift to privacy-friendly web analytics application :) Commented Sep 4, 2022 at 16:09
  • 5
    @RandomPerson I posted a feature request. I will be seeing marketing copy when I dream, tonight.
    – wizzwizz4
    Commented Sep 5, 2022 at 19:32
  • 13
    @ShadowTheKidWizard: "it's a business after all..." Likewise for child labor, and toxic meat-packing, and... Time for some regulation, if not doing away with business after all. Business as usual - the wild west if left to its own anarchic devices. After all, if SE doesn't do it, its competitors will. Age-old wail.
    – Drew
    Commented Sep 6, 2022 at 21:35
11

Neat stuff! Thanks for giving us the timeline for these changes as well as the flowchart that describes your current and future setups.

Out of curiosity, for what reason will Super User have these changes rolled out first?

Also:

What this means for us is we will be able to better understand user behavior and any other aggregate traffic data, without collecting any identifying information.

Do you have any plans for this data? I.e. are you intent on making any internal decisions based on the results of the data you collect following this change?

4
  • 15
    Presumably Super User is big enough to be a meaningful test but small enough not to be a really big problem if it goes wrong. Commented Aug 31, 2022 at 15:21
  • 10
    Your theory is correct @RobertLongson! The sample size on Super User was perfect to start this rollout before we introduced it to the network.
    – Bella_Blue StaffMod
    Commented Aug 31, 2022 at 15:32
  • 4
    Spevacus, regarding your second question about the data that Google collects from the cookieless pings, we expect to have more visibility into the overall traffic with Consent Mode. It's a bit early to tell what product decisions will be made yet until we start collecting and analyzing data in our new GA4 properties.
    – tanj92 Staff
    Commented Aug 31, 2022 at 15:55
  • 10
    I'd assume that there is also a chance for better bug reports on Super User than, say, on cooking or biblical hermeneutics. Commented Sep 1, 2022 at 11:04
8

A few questions arise naturally:

  1. Alexa ranking was closed recently (1st May 2022). Does SE envision that using GA4 will be important for any upcoming ranking that will substitute Alexa ranking? (SE was ranked in the TOP 40 previously.)

  2. SE consistently ranks high on Google searches, is there any positive known correlation between using Google analytics and the SEO ranking?

10
  • 5
    Google claims that Google Analytics doesn't affect SEO. It's conceivable that it does, in subtle ways (e.g. pages known to GA are added to the "pages to index" list) or directly (like how Google treated AMP), but even if it does, I don't honestly think Stack Overflow needs to care about that kind of thing. It's Stack Overflow; it's up there with Wikipedia (a site known to be special-cased by Google).
    – wizzwizz4
    Commented Sep 5, 2022 at 14:59
  • 7
    @wizzwizz4 then the remaining question is why does SE consider analytics more important than adding value for the userbase by hardening their privacy? (I fail to find any answer for this, 25k tools were considered a marginal convenience adding little real value.) Why detriment the content creators/curators privacy (that's the real question)?
    – bad_coder
    Commented Sep 5, 2022 at 15:26
  • 1
    They don't, necessarily. Stack Overflow has used Google Analytics since forever; they could easily have just not thought about it, not considered alternatives, and taken Google's marketing copy as valid and accurate claims about Google Analytics' legality. Maybe they don't even know about alternatives! Google has a large advertising budget, which most privacy-friendly alternatives don't have.
    – wizzwizz4
    Commented Sep 5, 2022 at 15:30
  • 1
    @wizzwizz4 saying SE is special cased doesn't mean it will remain special cased if they don't deploy GA4.
    – bad_coder
    Commented Sep 5, 2022 at 15:30
  • 3
    I understand distrusting Google, but we have no reason to believe Google would do something like that. If word got out that Google was doing that kind of thing… Well, I'd expect it to have already leaked, if they were, given what else we know about what Google does. We have no evidence that Google Analytics affects Google search results in any way; seeing as we've had that evidence about many other things (e.g. Google AMP), absence of evidence is evidence of absence.
    – wizzwizz4
    Commented Sep 5, 2022 at 15:37
  • 2
  • 2
    @wizzwizz4 it's called "non disclosure" clause in the work contract. That's why you never hear about bankruptcies or ethically questionable business practices until long after they're done. (In fact, reading the excellent above answer there's not a single reason to trust Google with anything, a choice that should me mine and not a default site option.)
    – bad_coder
    Commented Sep 5, 2022 at 17:15
  • 6
    @wizzwizz4 Google has a long trackrecord of illegal business activities
    – bad_coder
    Commented Sep 5, 2022 at 17:40
  • 3
    I'm aware of Google's record. I wrote the "excellent above answer". I don't think Google currently favours Google Analytics users in their search rankings.
    – wizzwizz4
    Commented Sep 5, 2022 at 17:43
  • 4
    Google has not made any announcements about creating any type of site ranking similar to Alexa ranking. If they were going to create such a system, I wouldn't expect them to use Google Analytics data to produce it because they would want to include sites that don't have GA installed. Alexa rank data was gathered from users with a browser extension installed that collected stats about sites they visited. They also wouldn't want sites to be able to game their rankings easily by sending junk data to Google Analytics. Commented Sep 6, 2022 at 17:59
4

I believe users will consent to different levels of behavior analytics depending on some variables:

  • where is my data stored?
    • if it is stored inside stackexchange infrastructure, there is higher trust
    • if it is stored in a big tech like Google, there comes a concern on what they will do with it
  • what is collected?
    • page views, clicks and read duration: anyone can understand that will help enhance the websites
    • user identification: that may start creating doubt

Have you considered using a self-hosted service? Tuned to collect what you really need? I think that would positively change the users' perspective on web analytics.


See also: Study of analytics alternatives by @wizzwizz4

1
0

I really love this new upgrade rolling out, especially the flowchart and timeline given! :P

However I have a few questions:

Why do you think ad_storage and analytics_storage is better than the beforehand Universal Analytics?

Also, why does data have to be sent to GA3 property and GA4 property in the GA4 flowchart? Why not just GA4?

1
  • 6
    I think the question covers all that - GA3/UA is going away, so they need to switch no matter what, and they'd run them simultaniously so they can figure out any glitches in the new system before completely switching over Commented Sep 4, 2022 at 8:17

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .