PageSight vs Google Analytics

SIGNALS CAPTURED PER VISITOR

GA4 captures around fourteen useful fields per hit. We capture 274.

That's the difference, geometrically. Everything downstream — the heatmaps, the bot detection, the experiments, the recommendations — runs on signal volume. When the inputs are sparse, the answers are vague. When the inputs are rich, the math becomes specific.

14 GA4 · USEFUL FIELDS

vs.

274 PAGESIGHT · CAPTURED

EIGHT SIGNALS GA4 CAN'T COMPUTE

Things you've been flying blind on, organized into eight little tragedies.

Each of these is a field PageSight captures and GA4 doesn't. Each is a question your competitor can answer that you currently can't.

01

hardware.tier · cpu.cores · device.memory

GA4 knows browser and OS. It doesn't know if your visitor is on a 2-core Android Go phone or an M3 Max. So it can't tell you that your checkout's slowness is killing the low-end cohort exclusively. PageSight · captured · GA4 · absent

02

network.ect · rtt · downlink · saveData

The Network Information API tells you a visitor's connection type (4G / 3G / 2G / slow-2G) and Save-Data preference. GA4 ignores it. PageSight surfaces "your /pricing page loses 47% of saveData=true visitors." PageSight · captured · GA4 · absent

03

vitals.{lcp,inp,cls}.p75 × cohort

GA4 reports a single average for Web Vitals. PageSight reports P75 LCP/INP/CLS at cohort granularity — by device tier × connection type × country. So "the site is fast" becomes "the site is fast for desktop fiber and 1.8s too slow for Android 4G." PageSight · cohort-bucketed · GA4 · global average

04

referrer.ai_engine

PageSight detects ChatGPT, Perplexity, Claude, and Gemini referrals (both their crawlers and human visitors arriving from chat sessions). GA4 lumps these into "Direct" or attributes them to the public homepage URL of the engine. PageSight · 4 engines · GA4 · "direct"

05

dark_social.class · whatsapp · slack · imessage

"Direct" traffic is a wastebasket. PageSight breaks it apart using in-app browser detection × arrival pattern recognition: WhatsApp shares look different from Slack shares which look different from iMessage forwards. PageSight · sub-classified · GA4 · "(direct)"

06

paragraph.dwell · paragraph.read.ai_engine

IntersectionObserver on every <p> tells you which paragraphs readers dwell on, which they skip, which AI engines quote. GA4 gives you a pageview. PageSight · per-paragraph · GA4 · per-page

07

bio.fitts.residual · bio.tremor.spectral_band

Behavioral biometrics — Fitts's Law residual against expected target-acquisition time, tremor frequency in the 4–8 Hz physiological band — distinguish humans from bots that scripted libraries cannot fake without modeling neuromuscular noise. PageSight · joint-distribution · GA4 · IP heuristic

08

committee.role · committee.size · committee.asn

Three visitors from the same ASN view your /pricing page in seven days. Role detection identifies an engineer, a procurement analyst, and an executive. A buying committee is forming. GA4 will not tell you this exists. PageSight · reconstructed · GA4 · invisible

FEATURE MATRIX · APPLES TO ORANGES, COMPARED ANYWAY

What each tool was built to do.

Capability	Google Analytics 4	PageSight
Pageview counting	✓ · industry standard	✓ · plus dollar-value per page
Acquisition / channel reports	✓	✓ · with dark-social sub-class
UTM tracking	✓	✓ · plus all 10 click-IDs
Conversion goals	✓	✓ · revenue-weighted
Cohort analysis	limited · 12 dims	274 dims · auto-discovered
Web Vitals (LCP / INP / CLS)	average	P75 by hw × net × geo
Heatmaps / scroll depth	—	✓ · cohort-segmented
Session replay	—	✓ · metadata + selective frames
A/B testing	—	✓ · edge · cohort-stratified
Error tracking	—	✓ · revenue-weighted
Uptime monitoring	—	✓ · real-visitor absence
Bot scoring	basic IP heuristic	Bayesian · 30+ signal classes
B2B identity / account detection	—	✓ · Fellegi-Sunter
Buying committee detection	—	✓
AI citation tracking	—	✓ · closed loop, 4 engines
Prescriptive recommendations	—	✓ · dollar-denominated
Verification loop on actions	—	✓ · 14-day automatic
Data retention	14 months (2 with consent)	30d – 7y · by tier
Data export	BigQuery (paid · 1M ev cap free)	Parquet · JSON · CSV · anytime
Price	free · GA360 from $50K/yr	$0 – $399/mo · Enterprise custom

THE HONEST PARAGRAPH WE WROTE FIRST

Most teams should keep GA4 running alongside PageSight. GA4 is free, ubiquitous, and the lingua franca of marketing — your CMO will want it for cross-channel reporting. PageSight is the tool that tells you what to fix. Use GA4 to count. Use PageSight to act.

The exception: if your team has stopped opening GA4 because nobody can find the report they need, cancel it. If GA4 is the only analytics you have and your conversion has been flat for three quarters, the issue isn't motivation — you're missing signals. That's what we're for.

PROOF · NOT VIBES

A 14-day side-by-side. We pay for it.

Run GA4 and PageSight on the same site for two weeks. At the end, we'll show you, in writing: every insight GA4 produced that we missed, every insight we produced that GA4 missed, and the dollar value of each. If GA4 wins on net, we cancel your trial and refund nothing — because you didn't pay anything.

Start the side-by-side Read a sample report (PDF)

ALSO COMPARE

Or read the other matchups.

vs HOTJAR Cohort-aware heatmaps Heatmaps that change when you change cohorts. Hotjar can't. vs SENTRY Revenue-weighted errors Sentry ranks by frequency. We rank by dollars blocked. vs SEMRUSH Your visitors, not competitors' Semrush sees other people's sites. We see yours. vs HUBSPOT Anonymous traffic too HubSpot tracks contacts. We track everyone who lands.

GA4 shows you traffic. PageSight shows you what's happening to your revenue.

GA4 captures around fourteen useful fields per hit. We capture 274.

Things you've been flying blind on, organized into eight little tragedies.

What each tool was built to do.

A 14-day side-by-side. We pay for it.

Or read the other matchups.

Keep GA4. Add what's missing.