The 24 Carat Data Standard: How Pure Is Your Business Data?
Everyone says data is the new gold — nobody asks what carat. This is the test. Told through King Croesus, who minted the world’s first pure gold coins on a river that ran with it, the 24 Carat Data Standard turns “how good is my data?” into five purity tests — can you take it with you, is it real, can you ask it anything, will it be there next year, does it show you everything — scored across 25 sub-criteria out of 25. GA4 scores 9 carat (mostly bot traffic and sampled guesses). HubSpot scores 18 (real, but rented). Only owned, server-side, AI-queryable data is 24 carat. Score yours, and see the path from filler to pure gold.
An hour east of Izmir, in western Turkey, a small river called the Sart Çayı winds through olive groves and farmland before passing through the ruins of an ancient city most tourists have never heard of. Unlike Ephesus, there are no crowds. No selfie sticks. Some days, archaeologists from Harvard and Cornell — who have been excavating here since 1958 — outnumber the visitors.
The locals call it the Sardis stream. The Greeks had an older name for it: Chrysorrhoas — “streaming with gold.”
Two and a half thousand years ago, this river ran gold.
Not metaphorically. The River Pactolus, as the ancients knew it, carried gold dust in its current — electrum, a natural alloy of gold and silver fused together by the earth itself, tumbling down from Mount Tmolus and settling in the sands where workers sieved it by hand. Legend said King Midas had washed his hands in these waters to rid himself of the golden touch, and in doing so, turned the riverbed into a mine.
Around 560 BCE, a man named Croesus stood on these banks and watched his workers pull pale, heavy lumps from the sand. It looked like gold. It felt like gold.
It wasn’t gold.
It was electrum — and every piece was a guess. Some nuggets were 70% gold. Some were 55%. Some were less. You couldn’t tell by looking. You couldn’t tell by weighing. The gold and the silver were so thoroughly mixed that every coin his father Alyattes had minted from it carried the same uncertainty. Lydia’s currency — the world’s first coins — looked valuable. They were valuable. But nobody could say exactly how valuable, because nobody could separate what was real from what was filler.
Croesus looked at the electrum in his hand and saw the problem that would define his reign.
He didn’t want coins that were probably gold. He wanted coins that were provably gold. Pure. Standardised. Unmistakable.
So he did something no one had done before. He hired metallurgists to develop a refining process that separated gold from silver completely — heat, salt, acid, patience. When they finished, Croesus minted the world’s first pure gold coins. And the world’s first pure silver coins beside them. No more guessing. A gold stater was gold. A silver stater was silver. The purity was guaranteed by the king’s stamp: a lion and a bull, facing each other.
Within a decade, “rich as Croesus” wasn’t a compliment. It was a fact. His treasury was the largest in the known world. His currency was trusted from Athens to Babylon. He funded the construction of the Temple of Artemis at Ephesus — one of the Seven Wonders of the ancient world. He had solved the oldest problem in commerce: how do you know what something is actually worth?
He considered himself the happiest man alive.
He was wrong about that. But that part of the story comes later.
Data = Gold. But What Carat?
Everyone says “data is the new gold.” Nobody asks the follow-up: what carat?
You’re standing on the banks of your own River Pactolus right now. Your analytics — the numbers in your GA4 dashboard, the contacts in your CRM, the traffic reports you present to your board — look like gold. They feel like gold.
But how much of what you’re holding is real?
9 carat gold is only 37.5% pure. The other 62.5% is silver, copper, and zinc — filler metals that tarnish over time. Your Google Analytics data is about the same: roughly 40% real human activity, the rest bot traffic, sampled guesses, and invisible AI crawlers. You’re wearing a 9 carat ring and calling it gold.
HubSpot is 18 carat — 75% pure. The CRM data is genuine. The contacts are real. But you don’t fully own it. The exports are incomplete. The behavioural trails stay behind when you leave. And on July 1, 2026, HubSpot updated its terms of service to allow contact enrichment data to be shared across customer accounts — opt-out by default. By July 5, after four days of backlash on LinkedIn, they reversed it entirely. Duncan Lennox, HubSpot’s Chief Product and Technology Officer, posted “We Got This Wrong. And We Are Fixing It.” They reversed it. But the tarnish showed through.
24 carat is 99.9% pure. No filler. No tarnish. No pretence. 24 carat data is the same: complete, real, owned, portable, and untouched.
This framework gives you five tests to measure the carat of your data — and a clear path from 9 to 24.
The 5 Purity Tests
Each test has sub-criteria. Score each sub-criterion: Yes = 1 point, Partially = 0.5, No = 0. Add them up at the end.
Test 1: Can You Take It With You?
If you cancelled every tool tomorrow, what would you walk away with?
| # | Sub-criterion | What it means |
|---|---|---|
| 1.1 | Can you export ALL your raw event data — every visit, every click, every form submission? | Not summaries. Not reports. The raw records. |
| 1.2 | Can you export it in a standard, open format (CSV, JSON, database dump)? | Not a proprietary format that only works inside one tool. |
| 1.3 | Is there a limit on how far back you can export? | GA4 only retains detailed data for 2 or 14 months by default. |
| 1.4 | Can you export it without paying extra? | Some platforms charge for data exports or API access above a threshold. |
| 1.5 | Would the exported data make sense to another system — or is it full of internal IDs that mean nothing outside? | Portable data works anywhere. Platform-locked data works nowhere else. |
Test 2: Is It Real?
How much of what your tools call “traffic” is actual humans doing actual things?
| # | Sub-criterion | What it means |
|---|---|---|
| 2.1 | Does your tracking filter out bot traffic before it hits your reports? | GA4 claims to filter bots. Thales’ 2026 Bad Bot Report found 53% of all web traffic in 2025 was automated. GA4’s filtering catches only a fraction of it. |
| 2.2 | Does it capture AI crawler and agent visits separately? | Claude, ChatGPT, Perplexity and other AI tools visit your site constantly. Ahrefs found AI sends just 0.1% of total referral traffic — yet McKinsey reports half of consumers now use AI as their primary source for product research. Over 70% of AI-driven visits arrive without referrer headers, landing as invisible “Direct” traffic in GA4. |
| 2.3 | Is the data collected from every visitor, or sampled from a percentage? | GA4 samples high-traffic sites. You’re making business decisions on a survey, not a census. |
| 2.4 | Does it work when visitors use ad blockers or privacy browsers? | Client-side tracking (like GA4) is invisible to roughly 30% of visitors using ad blockers — up to 40% on desktop. |
| 2.5 | Can you verify the data independently — compare it against server logs, sales records, or another source? | If you can’t cross-check it, you’re trusting one system’s word. |
Test 3: Can You Ask It Anything?
Can you ask your data any question — or only the questions your tool lets you ask?
| # | Sub-criterion | What it means |
|---|---|---|
| 3.1 | Can you query the data with your own questions, in your own way (SQL, AI, custom reports)? | GA4 lets you ask what Google decided you should ask. Your own database lets you ask anything. |
| 3.2 | Can you combine it with data from other sources (sales, email, support, social)? | Siloed data answers siloed questions. Connected data answers business questions. |
| 3.3 | Can you feed it to an AI assistant and get answers about your business? | If your data lives in someone else’s system, your AI can’t see it. |
| 3.4 | Can you build custom reports without hitting feature limits or upgrade paywalls? | Free tiers and starter plans restrict what you can ask. Your data shouldn’t have a paywall. |
| 3.5 | Can you reshape the data — merge fields, create custom dimensions, calculate new metrics — without the platform’s permission? | Your data should bend to your questions, not the other way around. |
Test 4: Will It Be There Next Year?
If the platform changes, shuts down, or raises prices — what happens to your data?
| # | Sub-criterion | What it means |
|---|---|---|
| 4.1 | If the platform shuts down or changes its product, do you lose your historical data? | Google announced the Universal Analytics sunset in March 2022 with a July 2023 deadline. Most businesses still lost years of historical data because the migration path was incomplete. |
| 4.2 | If you stop paying, does your data stay accessible? | Some platforms delete data after cancellation. Others keep it but lock you out. |
| 4.3 | Does the data live on infrastructure you control (your server, your database)? | If it lives on theirs, every answer above is “at their discretion.” |
| 4.4 | Can you back it up independently? | A backup you control is the only backup that counts. |
| 4.5 | Has the platform ever changed how it collects or reports data in a way that broke your historical comparisons? | GA4 did this with the UA migration. HubSpot does this with property changes. Owned data doesn’t change its definition under you. |
Test 5: Does It Show You Everything?
Are you seeing the full picture — or looking through a keyhole?
| # | Sub-criterion | What it means |
|---|---|---|
| 5.1 | Does it capture every visitor, including those using ad blockers, privacy browsers, and VPNs? | Client-side tracking misses roughly 30% of visitors. Server-side captures all of them. |
| 5.2 | Does it track the full journey — from first visit through every return visit to conversion — for each person? | Most tools show you sessions, not people. You see fragments, not stories. |
| 5.3 | Does it show you where visitors actually came from, including dark traffic (direct, email, messaging apps)? | “Direct” traffic in GA4 is often unattributed — it came from somewhere, the tool just doesn’t know where. |
| 5.4 | Does it show you what AI systems are doing with your content? | AI agents, chatbots, and search engines visit and use your content. If your tracking doesn’t see them, you’re blind to a growing share of how people find you. |
| 5.5 | Does it give you the same numbers every time you run the same report? | Sampled data gives different results depending on when you look. Pure data is deterministic — same question, same answer, every time. |
The Scoring System
25 sub-criteria. Each scores 0, 0.5, or 1. Maximum score: 25 points.
| Score | Carat Rating | What It Means |
|---|---|---|
| 0–9 | 9 Carat (37.5% pure) | Your data is mostly filler. You’re making decisions on bot traffic, sampled numbers, and borrowed access. The ring looks gold, but it tarnishes. |
| 10–18 | 18 Carat (75% pure) | Your data is genuinely better. Real contacts, real CRM records. But you’re renting it, the exports are incomplete, and there’s a governance layer hiding the gaps. |
| 19–25 | 24 Carat (99.9% pure) | Your data is yours. Every visitor, every event, server-side, portable, queryable by AI, untouched. No filler. No tarnish. No pretence. |
Most small businesses score 4–7. They don’t know.
Now you do.
Renting Gold vs Owning Gold
Here’s the distinction most businesses miss: there’s nothing wrong with using GA4. There’s nothing wrong with VidIQ, or HubSpot, or any platform that gives you data inside its walls. These tools are genuinely useful. The problem isn’t using them. The problem is only using them — and assuming the data inside someone else’s system is yours.
It’s not. You’re renting it.
When your data lives inside GA4, you see what Google lets you see. When it lives inside VidIQ, you see what VidIQ lets you see. When it lives inside HubSpot, you see what your pricing tier lets you see. The insights are real — but they’re furnished apartments. You can rearrange the furniture. You can’t knock down a wall. And if you stop paying rent, you lose the apartment and everything in it.
Owning your data means having a copy that lives in your warehouse, in a format you control, queryable by any tool you choose — including AI. You can still use GA4. You can still use VidIQ. But alongside them, your server-side layer captures the same events into a database you own. The rented view and the owned record sit side by side — and when they disagree, you know which one to trust.
This isn’t an either/or. It’s a both/and. Use the tools. Rent the dashboards. But own the gold.
When You Rent the Refinery, Someone Else Decides What’s Pure
There’s a cost to renting that doesn’t appear on any invoice: compliance inheritance.
When you use GA4, you inherit Google’s compliance posture — which defaults to the strictest privacy jurisdiction on earth. GA4’s Consent Mode means that every visitor who declines a cookie banner becomes invisible or “modelled.” In European markets, opt-out rates average 42% and individual sites regularly see 50–61% of visitors decline. Those aren’t bots. Those aren’t sampling errors. Those are real humans, actively choosing to become invisible — and your tool making them disappear.
The numbers are staggering. After Google silently enforced Consent Mode V2 for EEA and UK traffic in July 2025, some WordPress stores saw GA4 traffic drop 50–95% overnight. No warning. No deprecation notice. Just a dashboard that went dark. And if your site doesn’t get 1,000+ daily events from consenting users, Google’s behavioural modelling — its attempt to estimate the missing data — won’t even activate. For small businesses, there is no safety net.
Here’s what nobody tells you: if your business operates from Singapore, or the Philippines, or Australia, and your customers are in the same region, GDPR doesn’t apply to you. Singapore’s PDPA, the Philippines’ Data Privacy Act, Australia’s Privacy Act — these have their own rules, and they’re different from Europe’s. But because you’re renting Google’s system, you’re following Google’s rules. You’re applying European consent requirements to Southeast Asian visitors who never asked for them — and losing a third or more of your data in the process.
Someone else is forcing you to have low quality gold — and you don’t realise it.
That’s not compliance. That’s collateral damage from someone else’s compliance.
Server-side tracking on your own infrastructure puts you back in control. You comply with your jurisdiction’s actual legal requirements — not with the broadest interpretation a platform vendor applies to protect itself globally. The visitors you have a legal right to see become visible again. The data you’re legally permitted to collect actually gets collected.
This isn’t about sidestepping privacy law. It’s about following your privacy law instead of someone else’s.
Croesus didn’t send his electrum to someone else’s furnace and hope they gave him back pure gold. He built his own refinery. He controlled the process. He set the standard. He stamped the coins himself.
The moment you hand your data collection to someone else’s infrastructure, they set the standard. And their standard is designed to protect them — not to give you the purest data possible.
What Happened to the Happiest Man Alive
We left Croesus at the peak. The richest king in the known world, currency trusted everywhere, a treasury overflowing with pure gold. He had solved the purity problem. He could measure exactly what he had.
But pure gold and good judgement are different metals entirely.
Before the fall, a visitor arrived at Sardis — the Athenian statesman and lawgiver Solon, one of the wisest men in Greece. Croesus showed him everything: the vaults, the gold, the treasury that made kingdoms envious. Then he asked the question he was sure he already knew the answer to.
“Who is the happiest man you have ever seen?”
Solon didn’t say Croesus. He named Tellus — an ordinary Athenian who lived in a well-governed city, raised children who gave him grandchildren, and died honourably in battle defending his people.
Croesus, offended, pressed further. Surely he ranked second? Solon named Cleobis and Biton — two young men who yoked themselves to a cart to pull their mother to a temple, and died peacefully in their sleep that night.
Croesus was furious. He was the richest man in the world. How could a dead farmer and two dead boys be happier than him?
Solon’s reply has echoed for twenty-six centuries: “Count no man happy until he is dead. For no man can know what the gods may have in store for him.”
Croesus dismissed him. A fool philosopher. What would a lawmaker know about wealth?
Then Persia rose.
Cyrus the Great had overthrown the Median kingdom and was expanding west with alarming speed. Croesus, confident in his treasury and his alliances, consulted the Oracle at Delphi — the most revered source of prophecy in the ancient world. He asked whether he should go to war against Persia.
The Oracle’s reply is the most famous prophecy in history:
“If you cross the river Halys, you will destroy a great empire.”
Croesus heard what he wanted to hear. He assumed the great empire to be destroyed was Persia’s. He crossed the river. He attacked.
The great empire that fell was his own.
Cyrus captured Sardis in 546 BCE. Croesus was taken prisoner and, according to Herodotus, placed on a funeral pyre. As the flames rose around him, he cried out a single name — not a god, not a general, not a treasurer.
“Solon.”
He remembered the warning he had dismissed. The ordinary farmer who died well was happier than the richest king who misread his own data.
Cyrus, hearing the cry and learning the story, was so moved that he ordered the fire extinguished and spared Croesus’ life.
Here’s the thing.
Croesus didn’t fall because his gold was impure. He had the purest gold in the ancient world — he invented it. He heard what he wanted to hear. He saw what he wanted to see.
The Oracle didn’t lie. The data was accurate. “You will destroy a great empire” was perfectly true. But the data was ambiguous, and Croesus — blinded by his wealth, his confidence, his certainty that the answer would favour him — never considered the other interpretation.
That’s what happens when businesses read their GA4 dashboards. The data isn’t lying. It’s just ambiguous, incomplete, and layered with assumptions. And most businesses — like Croesus — read it in the way that confirms what they already believe.
Traffic is up. (Is it? Or is 53% of it bots?)
Direct traffic is growing. (Is it? Or is AI-referred traffic arriving without referrer headers?)
Our conversion rate is stable. (Is it? Or are 30% of your real visitors invisible to client-side tracking?)
Pure data is only half the equation. You still have to be able to ask it the right questions — and hear the answer it actually gives, not the one you want.
That’s what the 24 Carat Data Standard tests. Not just purity. Queryability. The ability to ask your data anything and get a true answer — same question, same answer, every time. No Oracle required.
The Metallurgy of Data
Gold alloys use specific metals for specific reasons. Each one maps to a data concept — but the mapping is more nuanced than you might expect.
Because here’s what most people get wrong: silver and copper aren’t villains. They’re heroes that degrade.
Silver — The Metal That Purifies
Silver has been used to purify water for thousands of years. The Romans listed it in their pharmacopoeia in 69 BC. Ancient mariners stored water in silver vessels because silver ions kill bacteria — they disrupt cellular respiration in microorganisms and prevent contamination. The Phoenicians kept their water, wine, and vinegar in silver bottles. NASA used silver ion purification on the Apollo spacecraft and still uses it on the International Space Station today.
Silver does real work. It fights disease. It protects people.
But it tarnishes. The very reactivity that makes silver antimicrobial — the release of ions — is what destroys it over time. The thing that makes it useful is the thing that degrades it.
| Metal | Role in Gold Alloys | % in 9ct | % in 18ct | % in 24ct | Data Equivalent |
|---|---|---|---|---|---|
| Gold | The valuable core | 37.5% | 75% | 99.9% | Real data — genuine human visitors, actual clicks, verified conversions, true attribution. The signal your business decisions should rest on. |
| Silver | Adds strength, looks valuable, but it’s not gold | 42.5% | 16% | 0% | Bot traffic and AI crawlers — they do real work (indexing your site, feeding the AI systems people use to find you), but they contaminate your analytics. Silver purifies the water but tarnishes the vessel. |
| Copper | Adds hardness, but corrodes over time | 20% | 7% | 0% | Sampled, duplicated, and dark traffic data — it built the age of digital marketing, but it corrodes under modern conditions. Copper protects the water but corrodes the pipe. |
| Zinc | Stops the copper from tarnishing — a stabiliser, not a purifier | 0% | 2% | 0% | Governance layers, polished dashboards, data validation rules — they make the surface look clean without removing the impurities underneath. |
Copper — The Metal That Protects
Copper’s antimicrobial history is even older than silver’s. The Smith Papyrus from 1600 BCE — one of the oldest medical documents in human history — records copper being used to sterilise drinking water and disinfect wounds. The Egyptians, Greeks, Romans, and Aztecs all used it medicinally. The Romans built their aqueducts and plumbing from copper specifically because of its sanitation properties.
During the cholera epidemic in 19th-century Paris, physicians discovered that workers in copper smelters were almost immune to the disease while the surrounding population died in thousands. In 2008, the US Environmental Protection Agency registered copper as the first antimicrobial metal surface.
Copper defined an entire age of civilisation. The Copper Age preceded the Bronze Age. The word “Cyprus” comes from copper — the island was named for the metal. But copper corrodes. The Statue of Liberty is green because copper can’t resist the elements. Over time, copper became invisible infrastructure — wiring, plumbing, pipes. Essential. Everywhere. Taken for granted.
That’s your GA4 data. Client-side JavaScript tracking, session-based measurement, sampled reports — this infrastructure built digital marketing. It worked when the internet was simpler, when there were fewer bots, no ad blockers, no AI crawlers, and everyone was on desktop. It protected businesses from total blindness. But it corrodes under modern conditions. The longer you rely on it, the greener it gets.
Gold — The Metal That Does Nothing
Here’s the twist that makes gold extraordinary: it doesn’t purify. It doesn’t protect. It doesn’t fight bacteria. It doesn’t react with anything.
Gold is chemically inert. It is the only metal that doesn’t tarnish, doesn’t corrode, and doesn’t need coating. You can pull a gold coin out of a 400-year-old shipwreck and it looks the same as the day it sank. That’s not a metaphor — that’s literally what happens.
Silver and copper are heroes that degrade. They serve, but the serving destroys them. Gold’s power isn’t what it does — it’s what it doesn’t do. It doesn’t interfere. It preserves. Indefinitely. Without changing.
That’s your 24 carat data. A data warehouse that doesn’t sample, doesn’t filter, doesn’t touch the records. Same query, same answer, every time. No degradation. No tarnish. No zinc required.
Old Coppernose
There’s a historical moment that ties all of this together.
In the 1540s, Henry VIII of England debased his own currency. Short on funds for his military campaigns, he ordered the silver coins of the realm to be mixed with copper — stretching the treasury by diluting the standard. The coins looked silver when they were freshly minted. But as they circulated and the silver wore off the high points of the design, the copper showed through — particularly on the raised portrait of the king’s nose.
Henry earned the nickname “Old Coppernose.”
The debasement showed on his own face.
Every business running GA4 is wearing a Coppernose coin right now. The reports look polished at first glance. The dashboards are well-designed. But the more you handle the data — the more you rub against the high points, the decisions that actually matter — the more the copper shows through. The bot traffic. The sampled numbers. The invisible visitors. The AI-referred traffic buried in “Direct.”
And the dashboards — those beautiful HubSpot dashboards — are the zinc. They galvanise the surface. They stop the copper from showing its corrosion. But they don’t remove the copper. They just make it look silver a little longer.
What the composition tells you
9 carat has no zinc. The impurities show openly — tarnished reports, obviously wrong numbers, traffic that doesn’t match reality. GA4 doesn’t pretend its data is complete. It just gives you what it has and lets you assume it’s gold.
18 carat adds zinc — a stabiliser that prevents the copper from showing its tarnish. HubSpot’s polished dashboards, structured pipelines, and clean CRM interface are the zinc. The surface is beautiful. The data looks reliable. But 25% of it is still filler — silver (bot-influenced), copper (incomplete behavioural data), and the zinc itself (governance that masks gaps rather than fixing them).
24 carat needs no zinc because there’s no copper to protect against. Pure gold doesn’t tarnish. Pure data doesn’t need governance layers to hide its problems — because it doesn’t have problems to hide. Server-side collection, owned infrastructure, complete records, no intermediary touching or sampling or filtering.
GA4 Scorecard: 9 Carat
| Test | Sub-criterion | Score | Why |
|---|---|---|---|
| 1. Take it with you | 1.1 Export all raw data | 0 | GA4 gives aggregated reports. Raw event data requires BigQuery Export setup — most small businesses never do this. |
| 1.2 Standard format | 0.5 | BigQuery export is standard, but the default UI exports are PDF/CSV summaries. | |
| 1.3 No time limit | 0 | Detailed data retained 2 or 14 months depending on settings. | |
| 1.4 No extra cost | 0.5 | GA4 is free, but BigQuery export incurs Google Cloud costs. | |
| 1.5 Portable data | 0 | GA4 dimensions and metrics are Google-specific. Meaningless outside the ecosystem. | |
| 2. Is it real | 2.1 Bot filtering | 0.5 | Claims to filter known bots. Thales’ 2026 Bad Bot Report found 53% of all web traffic is automated — GA4’s filtering catches only a fraction. |
| 2.2 AI crawler tracking | 0 | AI sends just 0.1% of referral traffic according to Ahrefs — yet over 70% of AI-driven visits arrive without referrer headers, invisible to GA4. | |
| 2.3 Not sampled | 0 | Samples data on high-traffic sites. Thresholding hides low-volume segments entirely. | |
| 2.4 Ad blocker proof | 0 | Client-side JavaScript tracking. Invisible to roughly 30% of visitors using ad blockers — up to 40% on desktop. | |
| 2.5 Independently verifiable | 0.5 | Can compare against server logs, but most businesses lack the skills to do this. | |
| 3. Ask anything | 3.1 Custom queries | 0 | Limited to pre-built reports and Explorations within Google’s UI. |
| 3.2 Combine with other sources | 0 | Siloed. Combining with CRM or email data requires complex integration. | |
| 3.3 Feed to AI | 0 | Data lives in Google. Your AI assistant can’t query it directly. | |
| 3.4 No paywalls on reports | 0.5 | Free, but advanced features (like longer retention) require paid Google Cloud. | |
| 3.5 Reshape freely | 0 | Custom dimensions exist but are limited in number and locked to Google’s schema. | |
| 4. There next year | 4.1 Survives platform changes | 0 | Google announced the UA sunset in March 2022 and shut it down in July 2023. Most businesses lost years of historical data because the migration path was incomplete. |
| 4.2 Survives non-payment | 1 | Free product, so no cancellation risk — but Google can change terms anytime. | |
| 4.3 Your infrastructure | 0 | Google’s servers, Google’s terms, Google’s decisions. | |
| 4.4 Independent backup | 0 | No native backup. BigQuery export is the closest, but requires active setup. | |
| 4.5 Stable definitions | 0 | The UA to GA4 migration broke every historical comparison. | |
| 5. Full picture | 5.1 Every visitor | 0 | Client-side. Roughly 30% of visitors invisible due to ad blockers. |
| 5.2 Full journey | 0 | Session-based, not person-based. You see fragments, not stories. | |
| 5.3 True attribution | 0 | “Direct” traffic is a catch-all for everything GA4 can’t attribute. | |
| 5.4 AI visibility | 0 | AI traffic is effectively invisible in GA4 reporting. | |
| 5.5 Deterministic | 0 | Sampled reports give different numbers depending on when you look. |
GA4 Total: 3.5 out of 25. Rating: 9 Carat.
14% of the possible score. The ring is mostly silver and copper.
HubSpot Scorecard: 18 Carat
| Test | Sub-criterion | Score | Why |
|---|---|---|---|
| 1. Take it with you | 1.1 Export all raw data | 0.5 | CRM contacts and deals export well. Behavioural timelines, engagement sequences, and workflow history are partial or lost on export. |
| 1.2 Standard format | 1 | CSV and API exports in standard formats. | |
| 1.3 No time limit | 0.5 | CRM data persists. But website analytics data tied to HubSpot tracking has retention limits depending on plan tier. | |
| 1.4 No extra cost | 0.5 | Free CRM exports, but full API access and bulk export features require paid tiers. | |
| 1.5 Portable data | 0.5 | CRM records are portable. But custom properties, lifecycle stages, and internal pipeline logic are HubSpot-specific. | |
| 2. Is it real | 2.1 Bot filtering | 1 | CRM data is genuine — real people who opted in, filled forms, or were manually entered. |
| 2.2 AI crawler tracking | 0 | HubSpot’s website tracking has the same client-side limitations as GA4 for AI crawler visibility. | |
| 2.3 Not sampled | 1 | CRM data is complete, not sampled. Every contact, every deal, every interaction recorded. | |
| 2.4 Ad blocker proof | 0.5 | CRM data bypasses ad blockers (it’s form-submitted). Website tracking is still client-side and blocked by ad blockers. | |
| 2.5 Independently verifiable | 1 | CRM records can be cross-checked against email, sales records, invoicing. | |
| 3. Ask anything | 3.1 Custom queries | 0.5 | Better reporting than GA4, but still limited to HubSpot’s report builder. No raw SQL access. |
| 3.2 Combine with other sources | 0.5 | Integrations exist (Zapier, native connectors), but combining data requires paid tiers and technical setup. | |
| 3.3 Feed to AI | 0.5 | HubSpot has AI features, but feeding raw HubSpot data to your own AI requires API extraction. | |
| 3.4 No paywalls on reports | 0 | Custom reports, dashboards, and advanced analytics are locked behind Professional and Enterprise tiers ($800–$3,600/month). | |
| 3.5 Reshape freely | 0.5 | Custom properties and calculated fields exist but are constrained by HubSpot’s data model. | |
| 4. There next year | 4.1 Survives platform changes | 0.5 | HubSpot is stable and well-funded. But platform changes (property deprecations, feature moves between tiers) affect historical data. |
| 4.2 Survives non-payment | 0 | Cancel HubSpot, lose access to everything above free-tier limits. Historical data behind the paywall. | |
| 4.3 Your infrastructure | 0 | HubSpot’s servers, HubSpot’s rules. | |
| 4.4 Independent backup | 0.5 | API exports allow backup, but no native one-click full backup. Relational context is lost in flat exports. | |
| 4.5 Stable definitions | 0.5 | More stable than GA4, but property changes and tier restructuring have broken workflows historically. | |
| 5. Full picture | 5.1 Every visitor | 0.5 | CRM contacts: yes. Anonymous website visitors: same client-side limitations as GA4. |
| 5.2 Full journey | 0.5 | Timeline shows contact activity post-identification. Pre-identification journey is fragmentary. | |
| 5.3 True attribution | 0.5 | Multi-touch attribution exists on Enterprise tier. Starter/Pro users get last-touch only. | |
| 5.4 AI visibility | 0 | Same blind spot as GA4 for AI agent traffic. | |
| 5.5 Deterministic | 1 | CRM data is deterministic — same query, same answer. No sampling. |
HubSpot Total: 12.5 out of 25. Rating: 18 Carat.
50% of the possible score. The zinc keeps the surface polished — but half the ring is still filler.
24 Carat: What Pure Data Looks Like
A 24 carat data system scores 25/25. Here’s what that means in practice.
| Test | What 24 Carat Delivers |
|---|---|
| 1. Take it with you | Every event lives in a database you control. Export the whole thing anytime — JSON, CSV, SQL dump, whatever you need. No retention limits. No export fees. No proprietary formats. Walk away from any tool and your data walks with you. |
| 2. Is it real | Server-side collection captures every visitor. No client-side JavaScript for ad blockers to kill. No sampling. No thresholding. AI crawlers and agents are tracked and attributed properly. What the data says happened actually happened. |
| 3. Ask it anything | Your data sits in a standard database (like BigQuery, PostgreSQL, or a warehouse). Query it with SQL. Feed it to Claude or any AI. Join it with sales data, email data, support tickets — any source, any question, any time. No paywalls. No report builder limits. |
| 4. There next year | The data lives on your infrastructure or a warehouse you pay for directly. No vendor can kill it, lock it, or change how it’s defined. Cancel any tool and the data stays. Back it up, replicate it, move it between clouds. It’s property, not access. |
| 5. Full picture | Server-side sees 100% of visitors. The journey from first anonymous visit through conversion to repeat customer is one continuous record. Attribution is real — you know where every person came from, including dark traffic, AI referrals, and sources GA4 calls “Direct.” Same query, same answer, every time. |
What this looks like for a small business: a server-side tracking layer (like Transmute Engine™) collecting events into a data warehouse (like BigQuery), queryable by AI, owned by you, independent of any single platform. GA4 and HubSpot can still be part of your stack — but they feed the warehouse, not the other way around. The warehouse is the gold. Everything else is alloy.
Your Data Purity Test
Score each question: Yes = 1, Partially = 0.5, No = 0. Add them up.
Test 1: Can You Take It With You?
- 1.1 Can you export ALL your raw event data?
- 1.2 Is the export in a standard, open format?
- 1.3 Can you export data from any time period with no retention limit?
- 1.4 Can you export without paying extra?
- 1.5 Would the exported data make sense to another system?
Test 2: Is It Real?
- 2.1 Does your tracking filter out bot traffic before it hits your reports?
- 2.2 Does it capture AI crawler visits separately?
- 2.3 Is all data collected — not sampled from a percentage?
- 2.4 Does it work when visitors use ad blockers?
- 2.5 Can you verify the data against another source?
Test 3: Can You Ask It Anything?
- 3.1 Can you query the data with your own questions (SQL, AI, custom)?
- 3.2 Can you combine it with data from other sources?
- 3.3 Can you feed it to an AI assistant?
- 3.4 Can you build custom reports without hitting upgrade paywalls?
- 3.5 Can you reshape the data without the platform’s permission?
Test 4: Will It Be There Next Year?
- 4.1 Would your data survive if the platform shut down?
- 4.2 Does your data stay accessible if you stop paying?
- 4.3 Does the data live on infrastructure you control?
- 4.4 Can you back it up independently?
- 4.5 Has the platform kept stable definitions without breaking your history?
Test 5: Does It Show You Everything?
- 5.1 Does it capture every visitor including those using ad blockers?
- 5.2 Does it track the full journey per person, not per session?
- 5.3 Does it show true attribution including dark traffic?
- 5.4 Does it show what AI systems are doing with your content?
- 5.5 Does it give you the same numbers every time you run the same report?
Your score: ___ / 25
| Score | Your data carat |
|---|---|
| 0–9 | 9 Carat — mostly filler. Time to refine. |
| 10–18 | 18 Carat — better than most. But you’re renting, not owning. |
| 19–25 | 24 Carat — pure. You own your data future. |
The Pyre and the Farmer
Croesus had the purest gold in the ancient world. He invented standardised currency. He solved the electrum problem that had plagued his father’s kingdom. He could measure exactly what every coin was worth.
And he lost everything — not because the gold was debased, but because he asked an ambiguous source a loaded question and heard the answer he wanted.
On the pyre, he didn’t call out for his gold. He called out for Solon — the man who told him that an ordinary farmer who lived well and died honourably was happier than the richest king in the world.
Solon’s lesson wasn’t about wealth. It was about what you measure.
Croesus measured gold. Solon measured a life.
Your GA4 dashboard measures sessions. Your business needs to measure people.
Your CRM measures contacts. Your business needs to measure journeys.
Your reports measure what your tools can see. Your business needs to measure what’s actually happening.
The 24 Carat Data Standard isn’t just a scoring framework. It’s a question Croesus should have asked twenty-six centuries ago: how pure is the thing I’m measuring with?
Don’t wait for the pyre. Test the gold now.
Take the test. Find out what carat ring your business is wearing.
Then plant a Data Tree — coined by Seresa — that compounds in value every month it grows. Because you can’t harvest fruit from trees you didn’t plant, and you can’t train AI on data you don’t own.
P.S. The River Pactolus still flows through Sardis. The gold ran out centuries ago. But the ruins of Croesus’ refinery — the place where he first separated gold from silver — have been excavated. You can visit them. An hour east of Izmir, past the olive groves, where the archaeologists outnumber the tourists. The stream they call Sart Çayı doesn’t carry gold anymore. But the question Croesus asked on its banks — how do you know what something is actually worth? — hasn’t been answered by most businesses yet.