GA4 vs Google Tag Manager — How to Choose (2026)

Colour key: GA4 GTM gtag.js alone

TL;DR

What they are. Google Analytics 4 (GA4) is a reporting product — it receives events, stores them, and gives you reports on traffic, funnels, attribution and revenue. Google Tag Manager (GTM) is a delivery product — a container you load once on the page that decides which tracking scripts fire, when, and with what data, without a code deploy. Both are free from Google, both install as one snippet in <head>, and both are configured in a browser UI by non-developers. They are not substitutes: GA4 is the destination, GTM is the pipe.

Is one better, or does it depend? They aren't rivals, so the real question is "if I could only install one, which?" — and there the answer is unambiguous: GTM, 8–4 on the scorecard, because GTM can deploy GA4 (plus Meta, Ads, TikTok, Hotjar) while GA4 can deploy nothing but itself. But that's a thought experiment. The correct answer for ~99% of sites is both: install GTM, deploy GA4 through it. Since 20 Aug 2026 Google has been merging the two — standalone Google tags are being upgraded into full GTM containers — so "GTM or gtag?" is a question Google itself is retiring.

	GTM — reach for it when	GA4 — reach for it when
Advantage	Unlimited 3rd-party tags · publish changes in ~1 min, 0 deploys · versioned with 1-click rollback · consent mode + server-side tagging live here	The only one that actually reports · 0 JS knowledge needed · ~2–4 h to productive · can't break your site
Don't use it if you need…	❌ a report, a chart, or a number — GTM stores 0 data❌ zero risk of a bad tag breaking the page❌ to avoid learning dataLayer/triggers (~1–2 days)❌ one fewer request on a hard perf budget	❌ to fire a Meta/TikTok/LinkedIn pixel❌ to change tracking without a code deploy (on raw gtag.js)❌ >14 months of retained data on the free tier❌ rollback of a bad config change

If you're already on hardcoded gtag.js and only ever send data to Google, migrating to GTM buys you less than it used to — the Aug 2026 update gives Google-tag users container features anyway. New builds should still start with GTM.

---

Status right now — 8 Sep 2026

	GA4	GTM
Share of all websites	47.6%	45.5%
Share of its own category	83.2% of sites with a known analytics tool	99.6% of sites with a known tag manager
Nearest rival	Yandex.Metrica 6.1%, Jetpack 6.0%, Matomo 2.5%	Adobe DTM 0.5%, Tealium 0.2%
11-year category trend	81.6% (2015) → 83.2% — flat, +1.6 pts	88.9% (2015) → 99.6% — +10.7 pts
Still shipping?	✅ cross-channel budgeting beta (16 Jan 2026), Measurement Protocol update (Feb 2026), Search Console latency 48 h → 12–24 h	✅ Google-tag/GTM merge (20 Aug 2026), Visual Tagging beta, redesigned Overview + Settings UI
Next major in flight	No GA5 signalled — GA4 is the platform	Containers absorbing Google tags; rollout is progressive, not automatic

GTM is the more entrenched of the two — a 199× lead over its closest competitor is effectively a monopoly, and it's still growing. GA4 has the larger absolute footprint but a softer category position: its share dipped to 79.0% in Jan 2026 before recovering to 83.2%, and privacy-first rivals (Matomo, Plausible, Fathom) chip at the edges in the EU. GTM is the more durable bet of the pair.

---

Summary table — the factors that decide it

Score = which tool owns that job. It measures role coverage, not quality — a low GA4 score on "fires Meta pixel" is by design, not a defect.

#	Factor	GA4	GTM	Gap	Winner
1	Produces reports / stores data	✅ full reporting, explorations, attribution	❌ 0 reports, 0 storage	absolute	GA4
2	Deploys 3rd-party tags (Meta, TikTok, LinkedIn, Hotjar)	❌ 0	✅ unlimited, 100s of templates	absolute	GTM
3	Change tracking without a deploy	⚠️ only events already arriving	✅ publish in ~1 min, 0 dev tickets	GTM only	GTM
4	Version control + rollback	❌ change-history log, no rollback	✅ versions + 1-click revert	GTM only	GTM
5	Share of all websites	47.6%	45.5%	+2.1 pts	GA4
6	Category dominance	83.2%	99.6%	rival at 0.5% vs 6.1%	GTM
7	Free-tier ceiling	14 mo retention · 30 key events · 50 event-scoped dims · 50k cardinality	500 containers/acct · 3 workspaces · 1,000 triggers · 200 KB container	GTM's limits bite far later	GTM
8	Cost to exceed it	GA4 360 ≈ $50k/yr entry (25M events/mo), typical $100k–175k	GTM 360 bundled in GMP; sGTM on Cloud Run $50–500+/mo	~100–1000× cheaper	GTM
9	Learning curve ⁽¹⁾	~2–4 h to read reports, 0 JS	~1–2 days — dataLayer, triggers, JS variables	~4× faster ramp	GA4
10	Blast radius if misconfigured	❌ bad data only, site keeps working	⚠️ Custom HTML injects live JS — can break the page	GA4 is safe	GA4
11	Ad-blocker / ITP resistance	⚠️ depends entirely on the transport	✅ server-side GTM + free Tag Gateway (Google-reported median +11% conversion signal)	GTM only	GTM
12	Consent Mode v2 governance	⚠️ consumes signals	✅ central place to wire consent for every vendor at once	GTM only	GTM
=	Total	4	🏆 8		GTM

⁽¹⁾ Row 9 is indicative, not measured — no vendor publishes time-to-productive.

---

Main features

Capability	GA4	GTM
Reports, funnels, explorations	✅	❌
Attribution + conversion modelling	✅	❌
Audience building → Google Ads	✅	❌
BigQuery export	✅ free, 1M events/day cap	❌ (server container can stream)
Predictive metrics / AI insights	✅	❌
Auto-collected events (scroll, outbound, file download)	✅ Enhanced Measurement	✅ built-in triggers
Fire non-Google pixels	❌	✅
Trigger logic (clicks, forms, visibility, timers, custom JS)	❌	✅
dataLayer / custom variables	❌	✅
Preview & Debug before publish	⚠️ DebugView (post-hoc)	✅ Preview mode (pre-publish)
Environments (dev/staging/prod)	❌	✅
User permissions per container/property	✅	✅
Server-side container	❌	✅
Data retention	2 mo default, 14 mo max free	n/a — stores nothing
Custom dimensions	50 event / 25 user / 10 item	n/a
Key events (conversions)	30 max	n/a
API	Data API, Admin API, Measurement Protocol	Tag Manager API — 10,000 req/project/day, 0.25 QPS
Price	Free · 360 from ~$50k/yr	Free · 360 via GMP

---

How to choose

Hard blockers — no further thought needed:

If…	Then
You need a chart, a number, or a funnel	GA4 — GTM cannot produce one
You need to fire a Meta / TikTok / LinkedIn pixel	GTM — GA4 cannot
Marketing must change tracking without a dev ticket	GTM
You have exactly one Google-only tag and a static site	gtag.js alone is fine; skip GTM
You're on a hard Core Web Vitals budget and Google-only	gtag.js alone — one fewer request
Regulated / EU consent obligations across many vendors	GTM + Consent Mode v2 as the control point
>14 months of history, or >30 conversions	GA4 360, or GA4 free + BigQuery export

For everyone else — install both. Points quiz, if you're deciding what to build first:

• More than one tracking vendor now or within 12 months → +2 GTM
• Non-developers own tracking changes → +2 GTM
• Ad-blocker loss is a known revenue problem → +2 GTM (server-side)
• You have no data at all yet and need a baseline this week → +2 GA4
• Nobody on the team writes JavaScript and never will → +1 GA4

That order is reversible cheaply; the reverse (hardcoded gtag.js everywhere, then retrofitting GTM) means a migration.

---

Pros

Dimension	GA4	GTM
Core job	✅ The only one that answers "how many, from where, worth what"	✅ One snippet becomes an unlimited tag surface
Speed of change	Enhanced Measurement toggles need 0 code	✅ ~1 min from edit to live, 0 deploys
Safety	✅ Cannot break the site	✅ Preview mode + versioned 1-click rollback
Cost	✅ Free at a scale most sites never exceed	✅ Free tier so generous the 360 upsell rarely triggers
Ecosystem	✅ 47.6% of all websites — every guide assumes it	✅ 99.6% category share — every vendor ships a GTM template
Privacy posture	Consent Mode–aware, EU-mode modelling	✅ Central Consent Mode v2 + free Tag Gateway (+11% median signal)

Cons

Dimension	GA4	GTM
Core job	❌ Ships 0 ability to deploy anything but itself	❌ Produces 0 reports — useless on its own
Speed of change	❌ New event params still need dev work upstream	Requires a dataLayer contract with engineering
Safety	Bad config silently corrupts data with no rollback	❌ Custom HTML runs arbitrary JS — can break the page
Cost	❌ 360 jumps to ~$50k/yr; no mid-tier	Server-side adds $50–500+/mo Cloud Run
Limits	❌ 14 mo retention, 30 key events, 50k cardinality → (other)	3 workspaces free, 200 KB container, API at 0.25 QPS
Learning curve	Explorations/attribution confuse most first-timers	❌ ~1–2 days before you're safe to publish
Performance	—	❌ One extra async request; container up to 200 KB

---

Similarities

• Both free, both from Google, both installed as one <head> snippet.
• Both browser-configured by non-developers, with granular user permissions.
• Both organised the same way: GA4 account → property → data stream; GTM account → container → workspace.
• Both had a big 2026 release and neither shows any sign of being wound down.
• Both are the runaway category leader — GA4 at 83.2% of known analytics installs, GTM at 99.6% of known tag managers.
• Both are converging: as of 20 Aug 2026 new snippets are identical and Google tags are being upgraded into full GTM containers.

GTM decides what data gets sent and when; GA4 decides what the data means — which is why every scorecard row about flexibility goes to GTM and every row about insight goes to GA4.

---

Trend

Category share, W3Techs, January of each year (Sep for 2026):

Year	GA (analytics category)	GTM (tag mgr category)	GTM's nearest rival	Ratio
2015	81.6%	88.9%	Yahoo 3.3%	27×
2017	83.4%	92.6%	Adobe DTM 3.4%	27×
2019	85.8%	98.3%	Adobe DTM 1.1%	89×
2021	84.0%	99.4%	Adobe DTM 0.4%	249×
2023	86.1%	99.5%	Adobe DTM 0.6%	166×
2025	82.2%	99.7%	Adobe DTM 0.5%	199×
2026 (Jan)	79.0% ⚠️	99.7%	Adobe DTM 0.4%	249×
2026 (Sep)	83.2%	99.6%	Adobe DTM 0.5%	199×

No crossover — these never competed. What the curves show:

1. 2019–2021: GTM's category closes. Adobe DTM's end-of-life and Yahoo Tag Manager's shutdown handed GTM a near-monopoly it has held for five years.
2. 2023–2025: GA's soft patch. The forced Universal Analytics sunset (1 Jul 2023 free / 1 Jul 2024 for 360) was the moment sites re-evaluated, and some left for Matomo/Plausible — GA slid 86.1% → 79.0%. It has since recovered to 83.2%.
3. Aug 2026: the merge. Google tags becoming full GTM containers means GTM's already-dominant share now absorbs the gtag.js-only population too.

---

Links

	GA4	GTM
Docs	GA4 help	GTM help
Get started	Set up GA4	GTM quick start
Console	analytics.google.com	tagmanager.google.com
What's new	GA changelog	GTM/Google tag updates, Aug 2026
Limits	GA4 property limits	Tag Manager API quotas
The decision doc	GTM vs gtag.js — Google's own comparison	Server-side tagging · Tag Gateway
Adoption data	W3Techs — GA	W3Techs — GTM

If the reason you're comparing these is privacy or ad-block loss rather than capability, Matomo (2.5% and rising, self-hostable, no sampling, unlimited retention) or Plausible/Fathom replace GA4 outright — but none of them replace GTM, and all three deploy through it.

---

Background

Skip unless you're writing an ADR.

	GA4	GTM
First shipped	Google Analytics Nov 2005 (from the Apr 2005 Urchin acquisition); GA4 as "App+Web" beta Jul 2019, GA in Oct 2020	1 Oct 2012
Predecessor	Universal Analytics — sunset 1 Jul 2023 (free) / 1 Jul 2024 (360)	none
Model	Event-based (UA was session/hit-based — the reason migration hurt)	Container of tags + triggers + variables, versioned
Business model	Free tier funnels data into Google Ads; 360 sold via GMP sales/partners	Free; 360 bundled into Google Marketing Platform
Current state	GA4 is the platform — no successor announced	Absorbing Google tags as of 20 Aug 2026

---

Caveats

1. These are complements, not competitors. The 8–4 scorecard answers "which single tool covers more jobs", not "which is better". Deploying GA4 through GTM scores 12/12 and is what almost everyone should do.
2. The scorecard flatters GTM by construction. GA4's single row-1 win is worth more in practice than four of GTM's — nobody buys a pipe for its own sake.
3. Row 9 (learning curve) is indicative, not measured. No vendor publishes time-to-productive figures.
4. W3Techs measures the top 10M sites, skewing toward larger, English-language, CMS-driven sites. Percentages of all websites (47.6% / 45.5%) are not directly comparable to per-category shares (83.2% / 99.6%).
5. GA's Jan 2026 dip to 79.0% may be a measurement artifact rather than real churn — the Sep 2026 recovery to 83.2% is a larger swing than the underlying market plausibly moved. Treat the 2026 rows as noisy.
6. GA4 360 pricing is not published. The $50k entry / $100k–175k typical figures come from third-party reporting and vary by contract, region and event volume.
7. The Aug 2026 merge is rolling out progressively. Your container may still show the old UI; nothing changes automatically, and existing installs keep working unchanged.
8. sGTM cost is highly traffic-dependent. $50–500+/mo covers typical mid-market Cloud Run usage; high-volume sites pay considerably more.

Sources

W3Techs — Google Analytics · W3Techs — Google Tag Manager · W3Techs — tag manager yearly trends · W3Techs — analytics yearly trends · Google — Updates to Google tag and GTM · Google — GTM vs gtag.js · Google — Tag Gateway · Google — Tag Manager API quotas · Google — GA changelog · Analytics Mania — GTM limits · Analytics Mania — GTM vs GA · XPON — Tag Gateway vs sGTM · GA4 Auditor — GA4 cost · mbadv — sGTM architecture & cost · GA4 Optimizer — GTM 2026 update · APIs and sites queried 8 Sep 2026.