Analytics Tools for Singapore Businesses: GA4, the Privacy-First Alternatives, and How to Choose

Last updated 30 August 2026 — by Adrian Tan, Singapore Digital Marketing

“Which analytics tool is best” is the wrong question, and it produces the wrong answer roughly every time. Every tool on this page will tell you how many people visited your website last month, and all of them will be approximately right. The question that actually separates them is narrower and more useful: what breaks first, and does that break matter to you?

For some Singapore businesses the answer is “nothing breaks, stay on the free one”. For others — a clinic, a law firm, anyone whose visitor list is itself sensitive — the first break happens on day one and has nothing to do with features. This guide walks the real ceilings, the published 2026 prices, and where the PDPA does and does not change the decision.

The four jobs an analytics tool has to do

Before comparing anything, be honest about which of these you actually need. Most SMEs need one and two, occasionally three, and almost never four.

1. Count and attribute traffic. How many people, from where, to which pages. Every tool here does this competently.
2. Measure outcomes. Form submissions, calls, purchases — and tie them back to the channel that produced them. This is where free tools start to differ.
3. Connect to advertising. Import conversions into Google Ads for bid optimisation. This is effectively a GA4-only capability, and it is the single strongest argument for staying on it.
4. Explore raw behaviour. Funnels, cohorts, path analysis, custom queries. Genuinely useful at scale, genuinely a distraction below it.

If your honest answer is “one and two”, the alternatives below are live options. If it is “and three”, the decision narrows sharply.

GA4: what you actually get free, and the four ceilings

Google Analytics 4 is free, capable, and the default for good reasons. It is also the tool people most often assume is telling them the whole truth. Four ceilings are worth knowing before you build reporting on it. Our GA4 setup guide for Singapore properties covers configuration; this is about limits.

Ceiling 1 — data retention is shorter than you think. On a standard GA4 property, user-level data can be retained for 2 or 14 months only. Event-level data can be set to 2, 14 or 26 months. Analytics 360 extends event data to 38 or 50 months, but user-level data stays capped at 14 months even there. Age, gender and interest data is held for two months regardless of your setting. Standard aggregated reports are unaffected — but the moment you want to explore behaviour beyond the retention window, the data is simply gone. Most Singapore SMEs we audit are still on the default rather than the maximum, which quietly costs them a year of comparison.

Ceiling 2 — thresholding hides small numbers, and you cannot turn it off. GA4 applies data thresholding to prevent viewers inferring the identity of individual users. It triggers on demographic data, on search query information with insufficient users, and on narrow date ranges with low counts. Google’s documentation is explicit that “data thresholds are system defined. You can’t adjust them.” For a small Singapore business looking at a single week of a single campaign, this is not an edge case — it is Tuesday. The workarounds are widening the date range or exporting to BigQuery.

Ceiling 3 — high-cardinality dimensions collapse into “(other)”. When a report has more unique dimension values than GA4’s row limits allow, the excess is aggregated into a single “(other)” row. Sites with heavily parameterised URLs or many custom values hit this and get a report that looks complete and is not.

Ceiling 4 — explorations sample. Standard reports do not, but explorations on large date ranges do, and the sampling notice is easy to miss at the top of the panel.

None of these make GA4 a bad tool. They make it a tool that needs to be read carefully — which is the argument of our guide to the six GA4 reports actually worth checking.

The AI-traffic question, and what GA4’s new channel really covers

This is the newest reason Singapore businesses reopen the analytics question, so it is worth being precise.

Google added an AI Assistant channel to GA4’s default channel group in 2026. Its documented rule is narrow: the channel matches where “the medium exactly matches ‘ai-assistant'”, and Google sets the medium to ai-assistant with campaign (ai-assistant) when the referrer matches its list of AI assistants. Google’s own documentation names ChatGPT, Gemini, Deepseek, Copilot and Grok — and states that the channel excludes Google’s own AI Overviews and AI Mode.

Three consequences follow, and they are the ones we walk clients through:

• Anything not on Google’s list still lands in Referral. An assistant that Google has not classified is not wrong in your data, it is just somewhere else.
• AI Overviews and AI Mode traffic is not in this channel at all. Those clicks arrive as Organic Search, because that is what they are. If you were hoping this channel would size your AI Overview exposure, it will not.
• Referrer-stripping apps show up as Direct. A native mobile app that sends no referrer produces a Direct session, and no channel grouping can recover it.

Notably, Plausible now surfaces AI-tool traffic as a first-class feature, naming ChatGPT, Perplexity and Claude on its own homepage. If measuring assistant referrals is a live priority rather than a curiosity, that is a genuine point of difference rather than marketing copy. Our guide to tracking AI search traffic goes into the measurement side properly.

The alternatives: published prices, 30 August 2026

All four below are privacy-first analytics products designed to run without cookies. Prices are as published on each vendor’s own pricing page on 30 August 2026.

Three observations a Singapore buyer should take from that table.

Matomo is the only one that publishes in Singapore dollars. SGD 29 per month excluding tax at the 50,000-hit tier, with overage at SGD 2.90 per additional 5,000 hits. Everyone else exposes you to the USD rate on every renewal. Over a three-year commitment that is a real variable, and it is invisible in every global comparison article.

Retention is where the alternatives beat GA4 outright. Matomo holds raw data for 24 months and report data indefinitely. Fathom keeps data for as long as you are a customer. Plausible holds three to five years depending on tier. GA4’s ceiling for event data is 26 months on a standard property. If year-on-year comparison matters to your business — and for seasonal Singapore retail and F&B it usually does — this is not a small difference.

Umami’s free tier is the genuinely underrated option. 100,000 events a month on one site, six months of retention, no cost. For a small brochure site or a single landing page, that is sufficient, and it is a real product rather than a trial.

The PDPA question, answered properly

This is where a great deal of nonsense circulates, so it is worth stating the position carefully.

Singapore does not have a GDPR-style cookie consent regime. What it has is the Personal Data Protection Act 2012, which applies when data is personal data — data about an individual who can be identified from that data, or from that data plus other information the organisation has or is likely to have access to. Aggregate visit counts are not that. A cookie-based identifier joined to a form submission may well be. Our guide to PDPA, cookies and marketing tracking works through where the line actually sits.

Two sections nonetheless shape the tool choice:

Section 25 — retention. You must cease to retain documents containing personal data, or anonymise them, as soon as it is reasonable to assume the collection purpose is no longer served and retention is no longer necessary for legal or business purposes. Cutting both ways, this means the tool with the longest retention is not automatically the best-behaved one — you need a defensible reason for the window you choose. A tool that lets you set retention explicitly is easier to defend than one that holds everything forever by default.

Section 26 — transfer outside Singapore. Personal data must not be transferred outside Singapore except in accordance with prescribed requirements ensuring a comparable standard of protection. Every tool on this page stores data outside Singapore, and the differences are worth knowing: Matomo Cloud lets you choose Europe as the storage location, and Plausible states that visitor data is processed exclusively on servers owned and operated by European companies and never leaves the EU. Self-hosting Matomo or Umami puts the question entirely in your hands — which is the strongest reason a Singapore clinic, law firm or financial adviser might choose it.

The honest summary: for a typical Singapore SME running a marketing website, GA4 is not a compliance problem and switching to Plausible is not a compliance fix. Where it genuinely changes is when the visitor list is itself sensitive, or when you have a contractual or sectoral obligation about where data sits.

Three-year cost, worked

Take a Singapore site doing roughly 50,000 pageviews a month — a fairly typical mid-sized SME site — and model three years at published rates, before tax.

Two cautions on that table. Matomo bills on hits, not pageviews — a hit includes events, so the 50,000-hit tier is not the same as 50,000 pageviews, and undersizing it is the most common Matomo billing surprise. And every USD figure is before the exchange rate and any foreign-currency card fee.

At this scale the money is not the deciding factor. Three years of any of these costs less than a fortnight of most Singapore SMEs’ ad budgets. Decide on retention, data location and whether you need the Google Ads link — not on price.

Running two tools: worth it, with one rule

Plenty of the businesses we work with run GA4 alongside a privacy-first tool, and it is a defensible setup: GA4 for the advertising integration and historical continuity, the alternative for a clean, fast, readable number that non-technical staff will actually open.

The rule, and it is not negotiable: nominate one as the source of truth for each metric before anyone reports on it. The two will not agree, and the reasons are structural rather than a bug — different bot filtering, different session definitions, different handling of consent-blocked traffic, different treatment of referrer-stripped visits. A team that has not decided in advance which number goes in the board pack will spend every month arguing about a discrepancy that was always going to exist. This is the same discipline our guide to vanity metrics argues for at the reporting layer, and the same reason consistent UTM tagging matters more than the tool you point it at.

The Singapore blind spot no tool on this page fixes

There is a measurement gap specific to how Singapore actually buys, and it is larger than any of the differences above. A meaningful share of enquiries here never touch a form. They arrive as a WhatsApp message, a phone call from a Google Business Profile listing, or a direct message on Instagram — and every tool on this page loses the thread at that moment.

Web analytics can tell you that someone clicked the WhatsApp button. It cannot tell you what they asked, whether they bought, or what that channel is worth. For a Singapore service business, that is frequently the majority of real demand sitting outside the dashboard entirely, which is why a “traffic is down but the phone is busier” month is so common and so confusing.

The fix is not a different analytics tool. It is instrumenting the handoff: a click event on every WhatsApp and call-to-action link, a distinct number or link per channel so calls are attributable, and a way of getting the outcome back into the same place the web data lives. We cover the mechanics in our guide to tracking calls and WhatsApp leads in Singapore, and the attribution consequences in choosing an attribution model. Do that work before you switch tools — it will change your numbers far more than the switch will.

Self-hosting: the cost that is not the licence

Matomo and Umami are both open source and both self-hostable, and for a business with a genuine data-residency requirement that is the cleanest answer available. It is not free.

Self-hosting means you own a server, a database, the upgrade path, the backups and the security posture of a system that by definition holds behavioural data about your visitors — which brings section 24 of the PDPA squarely into scope. In practice that is somewhere between a few hours a quarter and a genuine ongoing responsibility depending on traffic. The rule we apply: if there is nobody in the business who would notice the server going down over a weekend, do not self-host. A managed cloud plan with a documented data location is the better answer for most SMEs, and it is what we recommend far more often than self-hosting.

Frequently asked questions

Is Google Analytics legal to use in Singapore?

Yes. Singapore has no GDPR-style cookie consent regime, and no Singapore regulator has ruled against GA4 in the way several European authorities did. The PDPA applies where the data is personal data, and the obligations that matter are around retention (section 25) and transfer outside Singapore (section 26). Handle those and GA4 is a normal tool.

Do I need a cookie banner on a Singapore website?

Not as a blanket rule. What the PDPA requires is that you notify people of the purposes for which you collect, use and disclose personal data, and obtain consent where consent is the basis. Whether your analytics setup collects personal data at all depends on what you collect and what you can join it to. The privacy-first tools on this page are designed to avoid the question by not using cookies or persistent identifiers.

Can Plausible or Matomo replace GA4 entirely?

For traffic and outcome measurement, yes. For advertising, usually not. Importing conversions into Google Ads for automated bidding is a GA4 pathway, and if smart bidding is doing real work in your account, removing GA4 removes a signal the bidding depends on. Most businesses that switch fully are ones that do not run Google Ads.

Which analytics tool shows me traffic from ChatGPT?

GA4’s AI Assistant channel captures traffic where the medium is set to ai-assistant, based on Google’s list of recognised assistants, which its documentation gives as ChatGPT, Gemini, Deepseek, Copilot and Grok. It explicitly excludes Google’s AI Overviews and AI Mode, and anything not on the list stays in Referral. Plausible surfaces AI-tool traffic as a named feature, listing ChatGPT, Perplexity and Claude.

How much traffic can I run on a free plan?

GA4 is free at essentially any SME scale. Umami’s free Hobby tier covers 100,000 events a month on one site with six months of retention. Plausible, Fathom and Matomo Cloud are all paid from the start, with trials of 30, 7 and a free-trial period respectively.

Should I keep GA4 running if I switch?

Yes, for at least twelve months. It costs nothing, it preserves the historical comparison you would otherwise lose, and it gives you a second reading while you learn how the new tool counts. Just decide in advance which one is authoritative for reporting.

Where to start

Do this in order. Check your GA4 data retention setting today — if it is on the 2-month default, change it and you have just recovered two years of future comparison for free. Then write down which of the four jobs at the top of this page you actually need. If job three is on the list, stay on GA4 and stop shopping. If it is not, and either retention or data location is a live concern, price Matomo Cloud in SGD and Plausible in USD and pick on the constraint rather than the feature list.

And whichever you choose: the reporting habit matters far more than the tool. A simple dashboard that someone actually opens every Monday will beat a perfectly configured platform that nobody logs into.

If you would like a second opinion on your measurement setup, or on whether the numbers you are reporting are the ones worth acting on, get in touch. Our performance marketing guide covers the layer above this, and our case studies show how it plays out in practice.

Sources: Google’s own Analytics Help documentation on data retention, data thresholding and the default channel group; the published pricing pages of Matomo, Plausible, Fathom and Umami; and the Personal Data Protection Act 2012 on Singapore Statutes Online, current version as at 30 August 2026. All prices checked 30 August 2026 and subject to change.