Earlier this year a procurement officer at a Dutch ministry called me with a question I’ve answered in various forms over the past ten years: “What should we use instead of Google Analytics?”
The question itself isn’t new. What is new is that she didn’t call because someone in IT had read a privacy blog. She called because her DPO had flagged it in an audit, the communications team was three months away from a NIS2 review, and the answer had become a procurement deadline rather than an architecture preference.
I started working in privacy-first analytics in 2016, when the question was still abstract — a handful of DPOs and academics worrying about data flows that most organisations were happily ignoring. Three legal earthquakes later (Schrems II, NIS2, the EU Data Act), the abstract has become operational. European public organisations are no longer asking whether they need to leave US-hosted analytics. They’re asking how, and what comes next, and they’re asking under time pressure.
A note on what I build
Before I continue: I’m co-founder of D8A.Tech (https://www.d8a.tech), which operates in the warehouse-native analytics category discussed below. Before that I spent eight years at Piwik PRO (https://www.piwik.pro). I’ve tried to write the comparison I would have wanted to read when I was at Piwik PRO and it didn’t exist — honest about what each option does well and poorly, including the category I’m now building in. Read this with that in mind.
Why this became urgent
The Schrems II ruling came in 2020. For three years, most public organisations dealt with it the way they deal with fire drill posters: present, acknowledged, ignored. That changed when regulators in multiple member states started enforcing.
Decisions and guidelines from supervisory authorities in Austria, France, Italy, and Norway, among others, established that Google Analytics in its default configuration was incompatible with GDPR. None of these individually was dramatic; together they sent the signal that the legal interpretation was settled and the transition period was over.
Then the NIS2 transposition deadlines came. The directive forces organisations within its scope — and that scope is much broader than NIS1, covering most public organisations, healthcare, education, water, energy — to actively demonstrate risk management of their digital supply chain. “We use the same analytics tool as everyone else” is no longer an answer. The data access provisions of the EU Data Act add another layer; the AI Act’s overlap with profiling-based processing adds a third.
The cumulative effect, for the procurement officer I described, is that the question is no longer theoretical. The cost of doing nothing is now a concrete enforcement risk on a concrete timeline. That’s why the calls have increased.
The four real options
There are dozens of analytics tools claiming to be GA alternatives. For the European public sector, four categories genuinely matter.
Piwik PRO
The closest thing to a direct GA replacement. Polish company, EU-hosted, with private-cloud and on-premise options. Functionally deep — campaign tracking, e-commerce, custom dimensions, the complete package.
- Sterk in
- Minimal disruption for marketing teams accustomed to GA's depth, strong out-of-the-box compliance position, mature procurement processes, good multilingual support.
- Levert in
- It's still a vendor holding your event data, just under EU jurisdiction. For organisations reading 'sovereignty' strictly — as in data ownership, not just data jurisdiction — that's a meaningful gap. Also paid; the budget impact is real.
- Beste fit
- Ministries, larger municipalities, and regulated enterprises that want low-friction migration. If your marketing team does serious work in GA and would revolt at a Plausible-like tool, this is the answer.
Matomo
The grandfather of the privacy analytics category. Open source, can be self-hosted or run via Matomo Cloud. The plugin ecosystem is the largest in this market; the community is mature.
- Sterk in
- Flexibility, ownership, cost (free when self-hosted, modest when cloud-hosted), and a strong fit for organisations that already have engineering capacity and want to manage their own infrastructure.
- Levert in
- Polish and time-to-value. Out of the box, Matomo demands more from you than Piwik PRO. The self-hosted version requires genuine long-term engineering ownership — patches, scaling, backups, the usual work.
- Beste fit
- Organisations with internal technical capacity that prioritise ownership over convenience. Universities often fit well; smaller municipalities with a capable IT team can make good use of it.
Plausible
The lightest option, in every sense. Estonian, open-source, deliberately simple, no cookies by default. The product vision: 80% of organisations need 20% of GA's features and are better served by a focused tool that does that 20% well.
- Sterk in
- Speed of implementation, auditability (the privacy position is so clean that DPO reviews are brief), low cost, low cognitive load. The fastest 'we're off GA' story I've seen from start to finish.
- Levert in
- Depth. If your stakeholders ask for funnel analysis, multi-touch attribution, cohort retention, or e-commerce purchase tracking, Plausible will frustrate them. Optimised for content sites, not marketing operations.
- Beste fit
- Content-driven organisations — government communications sites, public information portals, faculty pages, museums. If it comes down to 'the communications team wants to know what people are reading', the migration takes a week, not a quarter.
Warehouse-native
The newest category. Instead of sending event data to a vendor's analytics product, you send it to your own data warehouse in the EU — or better yet, your own ClickHouse environment in the Netherlands — and run analytics queries directly on the warehouse.
- Sterk in
- Maximum data ownership: the events live in infrastructure you manage yourself, in a jurisdiction you choose. Composability: analytics, BI, CDP, and ML pipeline all read from the same source of truth. Costs scale with infrastructure, not per event.
- Levert in
- Not a drop-in GA replacement. You need a warehouse, you need pipelines, you need someone who understands both. For organisations without a modern data stack, this is the heaviest option.
- Beste fit
- Organisations already running a data warehouse for other purposes, or organisations large enough that building one is justified. Several larger Dutch ministries and implementing organisations fall into this category.
A decision framework that actually works
Vendor comparisons tend to end in a four-quadrant chart that nobody uses. After having this conversation hundreds of times, here are the four questions that actually determine which option fits.
1. Do you have a data warehouse, and is it staffed? If yes, warehouse-native is a serious option worth taking far into your evaluation. If no, it isn’t — not because it’s bad, but because you’re not in a position to manage it.
2. What does your marketing team actually need? Be specific. If “GA-equivalent depth on campaigns, attribution, and e-commerce” is non-negotiable, then you’re choosing between Piwik PRO and warehouse-native. If “we want to know which content is being read” is the truth, then Plausible is probably the right answer, and you save time and money by admitting that.
3. Who is going to manage it? Self-hosting Matomo is great until the person who set it up changes jobs. Be honest about whether you have sustained engineering ownership, or whether you’re better off with a vendor relationship that includes operations.
4. What does procurement actually allow? Some organisations have framework agreements that include Piwik PRO or Matomo Cloud and exclude others. Some have policies that effectively require open source. Some have no constraints. Find this out before drawing up a shortlist, not after.
If you answer those four questions honestly, the right option usually narrows down to one or two. The remaining choice becomes a matter of values rather than features — how strictly you read “sovereignty”, how much you trust your engineering capacity, how much you’re willing to invest now to build on later.
Where this is heading
The category is in motion. The interesting movement in the next 12 to 16 months is warehouse-native: not because the other three options are bad (they’re not), but because the underlying architecture of analytics is converging towards the same pattern as the rest of the modern data stack.
There’s no good reason for analytics to be the only workload where a vendor still owns your raw events while everything else runs on infrastructure you manage yourself. The market will figure this out, and Piwik PRO and Matomo will move in that direction — Piwik PRO already has an export-to-warehouse story; Matomo’s roadmap points in similar directions.
That’s the bet I’m making at govanalytics & D8A.Tech, but it’s a bet on the category, not just on our product. The smart move for organisations evaluating today is to choose the option that fits their current operational reality — with one eye on data portability. Make sure that whatever you choose enables you to extract your raw data and load it into a warehouse later, because that’s where the second migration will land.
Schrems II won’t be reversed. NIS2 enforcement won’t ease. The cost of doing nothing keeps rising. The cost of getting it right the first time, not so much.
Get in touch? Want to know more?
We discuss in 30 minutes which variant fits — Cloud, On-Premise, or building on the open-source core yourself. No sales pitch, no obligations.
Click here to book an online meeting →