GA4 Cost Calculator: When You’ll Actually Pay (And How Much)

“Google Analytics is free.” That’s been the marketing line for two decades, and it’s still mostly true. But “mostly” hides a lot of asterisks, and in 2026 the asterisks have grown teeth. GA4 has hard limits, soft limits, sampling thresholds, and a quietly expanding set of features that only work on GA4 360 — Google’s six-figure annual licence. Most teams find out about these limits the day a critical report stops working.

This article is the calculator I wish I’d had three years ago: when do you actually start paying for GA4, what does that look like, and at what point does the maths start tipping toward a paid privacy-first alternative anyway. I’ll work through three honest scenarios — small business, mid-market, enterprise — and show the inflection point in each.

What “free” actually means in GA4

The free tier of GA4 ships with the following caps, all of which are real and several of which I’ve watched teams hit:

• 10 million events per month before sampling kicks in. Above this, BigQuery export and standard reports start sampling.
• 50 custom dimensions and 50 custom metrics per property. Heavy event-tracking setups burn through these quickly.
• 10 conversions per property. Reduced from earlier limits in 2024.
• 14-month retention by default, 26 months max. No way to extend on the free tier.
• BigQuery free tier exports up to 1M events/day. Above that, you’re paying BigQuery costs.
• 2 million events/day in the data API. Extracts beyond this rate-limit out.

And the ones that are harder to put a number on:

• Sampling. Standard reports show “this report is based on 100% of your data” until they don’t. Once you cross thresholds, sampling activates without warning, and the threshold isn’t published in pageviews terms — it’s based on session count, query complexity, and time range.
• Data freshness. The free tier shows yesterday’s data within 24-48 hours. GA4 360 ships intraday updates.
• Service-level commitments. Free tier has none. 360 has 99.9%.
• Support. Free tier has documentation. 360 has dedicated reps and SLAs on response time.

When you’ll actually start paying

The transition from “GA4 is free” to “we need GA4 360” tends to happen at one of three triggers. None are pageview-based; that’s the misleading frame.

Trigger 1: Sampling makes a critical report unreliable. A finance team needs to pull last month’s traffic by channel for the board pack. The report shows “based on 4.2% of sessions”. The number could be off by 15%. The CFO is unhappy. Now the team needs unsampled data, which means either BigQuery (engineering work + cloud bill) or 360 (six-figure licence).

Trigger 2: A custom-event tracking setup hits the dimension cap. The product team wired up 60+ custom events for a beta launch, only to find out 50 is the limit. Existing dimensions can’t be deleted without losing historical data. The team faces either dropping events (which loses analytical depth) or upgrading to 360 (which raises the cap to 125).

Trigger 3: Compliance demands data residency or stricter retention. Healthcare, government, defence, and German Bundesländer privacy regulations have all explicitly stated GA4 free tier doesn’t satisfy their requirements. 360 offers EU-residency commitments and BAA-eligible processing. Free tier doesn’t.

For teams at any of these triggers, the GA4 360 quote starts at $50,000 per year and routinely lands in the $100k-$200k range for medium-traffic enterprise sites. The pricing isn’t published; you negotiate it through a Google reseller. The pricing isn’t a flat fee — it’s traffic-based, with multiple tiers, and renewable annually.

The hidden BigQuery cost that catches teams out

The free tier’s BigQuery export is the recommended workaround for sampling concerns. Pipe events to BigQuery, run unsampled queries against the raw data, problem solved. Except.

BigQuery itself has a free tier (1TB query / 10GB storage per month) that most analytics workloads exceed within 60-90 days. Once you’re past that, you’re paying:

• Storage: $0.02/GB/month. A medium ecommerce site generates 50-200GB/month.
• Queries: $5 per TB scanned. A poorly-optimised dashboard scans 50-100GB per refresh, multiple times per day.
• Streaming inserts: $0.05 per GB. Live-data dashboards add up.

The real BigQuery bill for a moderately data-keen team often lands at $300-$1500/month — which is 10-50x what you’d pay for a privacy-first SaaS analytics tool that doesn’t sample in the first place. The point isn’t that BigQuery is expensive in absolute terms; it’s that the path of “GA4 free + BigQuery for the gaps” is more costly than its proponents admit.

Scenario 1: Small business (under 100k pageviews/month)

You’re a one-person consultancy or a small ecommerce shop. Traffic is steady at 50-80k pageviews per month. You want to know which marketing channels work, which pages convert, and whether the recent SEO push paid off.

What you’d pay for GA4: $0.00. You’re well below all caps. Sampling won’t kick in. The free tier is genuinely free for you.

What you’d pay for a privacy-first alternative: $9-19/month. Plausible Cloud Growth at $9/month for up to 10k pageviews, or Pirsch Hobby at $6/month for 100k pageviews. Both ship the standard reports plus the privacy posture that lets you drop the cookie banner.

Inflection point: The financial cost of a privacy-first alternative is real but small. The non-financial cost of GA4 is the cookie banner (which costs you 7-12% of bounce rate per our conversion analysis) and the GA4 learning curve. For most small businesses, the maths tips toward the alternative on UX grounds before it tips on financial grounds.

What I’d do: Switch to Pirsch or Plausible. The $6-19/month bill is recovered within weeks via the bounce-rate improvement from dropping the cookie banner.

Scenario 2: Mid-market SaaS (1-5M pageviews/month)

You’re a 50-person SaaS business. Marketing site plus app traffic. You have a marketing team that needs attribution, a product team that needs feature-usage analytics, and a finance team that wants the monthly board pack.

What you’d pay for GA4 free: Notionally $0. In practice, you’ll start hitting the sampling threshold around 3M pageviews/month for complex reports — which is when the marketing team starts complaining that the channel breakdown doesn’t match the ad-platform numbers. The fix is BigQuery export.

BigQuery hidden cost: $300-$800/month for the typical mid-market setup. That’s $4-10k/year.

Plus the engineering cost. Setting up BigQuery export, building the unsampled dashboards (Looker Studio or similar), maintaining the schema as GA4 ships breaking changes, training the marketing team to query BigQuery for the sampled-elsewhere data. Conservatively 1-2 senior-engineer days per month, ongoing.

What you’d pay for a privacy-first alternative: Plausible Business at $59/month for 1M pageviews, or Matomo Cloud Essential at $39/month for 100k visits. Either ships unsampled data by default. Both have published APIs. Neither needs BigQuery.

Inflection point: Mid-market is where the maths actively flips against GA4. The “free” version is paying $5-10k/year in BigQuery + engineering, plus the cookie-banner conversion drag. The privacy-first alternative is paying $700-2000/year and shipping the data unsampled.

What I’d do: Run a 90-day evaluation of Matomo Cloud or Plausible Business. Compare the channel-attribution numbers against GA4 (they’ll be within 5-15%, with the alternative usually higher because no banner = more data captured). If the comparison passes the marketing team’s smell test, decommission GA4 and reinvest the BigQuery savings.

Scenario 3: Enterprise (10M+ pageviews/month, multiple properties)

You’re a media company, multi-brand ecommerce, or large SaaS. Multiple GA4 properties. A marketing team that runs paid acquisition. A product team that runs A/B tests. A data team that joins analytics with CRM and revenue data.

What GA4 free actually costs you:

• Sampling on most useful reports unless you’re piping to BigQuery.
• BigQuery costs in the $1500-5000/month range. That’s $20-60k/year.
• Engineering team to maintain the data layer: at least 0.5 FTE, probably 1 FTE. That’s $80-150k/year fully loaded.
• 14-month retention cap means historical YoY analysis stops at 14 months unless you’ve been archiving to BigQuery from day one.
• No data residency commitment. EU compliance starts to wobble.

What GA4 360 costs you: $100-200k/year in licence. Plus the engineering team is still needed for BigQuery, just slightly smaller.

What a privacy-first enterprise alternative costs you: Self-hosted Matomo on your own infrastructure: $10-30k/year in cloud + 0.3 FTE engineering. Or Matomo Cloud Business plan: $40-150/month per property, varies. Plausible Business at high traffic tiers: $300-800/month per property.

Inflection point: At enterprise scale, the comparison isn’t GA4-free vs alternative. It’s GA4-360 vs alternative. And the alternative is usually 30-50% cheaper in total cost of ownership, including the engineering work, with better data residency and unsampled data by default.

What I’d do: Self-host Matomo, with a managed-self-host provider if you don’t want full DevOps responsibility. Use Plausible or Fathom Cloud for the secondary marketing properties where the SQL access isn’t needed. Total spend ends up in the $40-80k/year range, with a cleaner privacy posture and unsampled data.

The decision framework

Total cost of GA4 ownership is rarely zero. Even on the free tier, you’re paying in cookie-banner conversion drag and in the engineering work needed to fight sampling. The question isn’t “free vs paid” — it’s “which paid setup gives you the data you actually need at the lowest TCO”.

The framework I use:

1. Estimate your real GA4 bill: BigQuery costs + engineering FTE + cookie-banner conversion drag (in revenue terms).
2. Compare against a privacy-first alternative at your traffic tier.
3. If the alternative is cheaper or even within 20%, the privacy posture and unsampled data tip the decision.
4. If the alternative is more expensive but you’re hitting compliance triggers (residency, retention, custom-event caps), the alternative still wins because GA4 will need to upgrade to 360 anyway.

For traffic-tier-by-tier comparisons, the dashboard at analytics-alternatives.com/dashboard shows live pricing and feature comparison across the seven tools we cover. For the head-to-head against GA4 specifically, our GA4-to-Plausible migration guide walks through the technical work involved.

What I see teams getting wrong

Treating GA4 as the default and alternatives as “investigations”. The default in 2026 should be a privacy-first tool. GA4 is the special case that needs justification — usually because of an existing GTM stack or a marketing team that’s been trained on GA UX. The investigation should run the other way.

Underestimating the BigQuery bill. The bill grows. Storage compounds. Queries multiply as the team builds dashboards. The first-month bill of $80 turns into the twelfth-month bill of $1200. Project the 12-month total before you commit to BigQuery as a sampling workaround.

Overestimating the cost of switching. Migrating from GA4 to a privacy-first tool is usually 1-3 weeks of one engineer’s time, plus a 30-day overlap period. The work is the snippet swap, the consent-banner removal, the CSV export of historical aggregates, and re-pointing dashboards in Google Looker / Looker Studio. It’s not a year-long migration.

Forgetting the cookie-banner cost. The 7-12% bounce-rate impact from the GA4 cookie banner is a real revenue cost. On a moderately-trafficked site, that’s often $10-50k/year in lost conversion. The privacy-first alternative recovers that revenue and pays for itself before you account for any other savings.

The cross-vendor cost comparison

If the framework says it’s time to switch, the next question is which alternative. Here’s the head-to-head against GA4 on the dimensions that matter for cost-conscious teams:

Plausible is the closest UX match to GA4 for marketing teams used to the GA layout. Fathom is the cleanest UX but has fewer dimensions. Matomo is the most feature-rich and the most engineering-heavy. All three undercut GA4 360 by an order of magnitude at typical mid-market traffic levels, and all three ship unsampled data by default.

Frequently asked questions

How do I know if my GA4 reports are being sampled?

Each report header in the GA4 UI shows a “based on X% of sessions” indicator. If it’s anything below 100%, you’re sampled. The threshold isn’t a fixed pageview count — it depends on session count, query complexity, and time range. Long-range queries (12+ months) sample more aggressively than 30-day queries.

Is GA4 360 worth it?

For sites in regulated industries with no realistic alternative path, sometimes yes. For everyone else, no — the same outcomes (unsampled data, EU residency, longer retention) are achievable on a privacy-first alternative at 5-20% of the licence cost.

What does a Google reseller charge for GA4 360?

The starting quote is usually $150,000/year, but it’s heavily traffic-tier dependent and negotiable. I’ve seen sites in the 5M pageviews/month range pay $80-120k. I’ve seen sites in the 50M+ range pay $300-500k. The pricing is opaque by design, which is itself a red flag for procurement teams.

Can I run GA4 free + a privacy-first tool side by side?

Yes, and many teams do during the migration period. The catch: if both tools are loading on the same page, you still have the cookies from GA4 and you still need a consent banner. The “side by side for comparison” approach loses the privacy benefit until you fully cut over.

What if my team is locked into GTM for tag management?

GTM works fine with privacy-first analytics. Plausible, Fathom, and Matomo all have GTM templates. The migration is the GA4 tag swap, not a tooling change. You can keep the rest of your GTM stack intact.

Does Plausible or Fathom support multi-property setups like GA4 properties?

Both treat each domain as a separate site, with shared user-management at the team level. The Plausible Business plan includes 50 sites; Fathom has unlimited on the higher tiers. For most multi-brand setups, this maps cleanly onto the GA4-property model.

What’s the realistic timeline for an enterprise migration?

For a single property with one marketing team: 2-4 weeks. For multi-property enterprise with several teams and downstream BI dashboards: 2-4 months. The bulk of the time is dashboard re-pointing and team training, not the tracking-snippet swap.