My journey into this started with something Professor Sunil Gupta said on The Parlor Room: stop treating data like oil. That one thought stuck, and it’s been rattling around in my head ever since.

“Data is the new oil” might be the most over-sold slogan in marketing decks since someone first discovered clip art. Catchy, yes. But oil is static, finite, and only valuable once it’s drilled and refined. Data doesn’t play by those rules.

Data is a garden. Seeds scatter everywhere—some bloom into real value, most just rot or sprout weeds. It’s alive, unpredictable, and demanding. Ignore it, and the weeds win. Abuse it, and you strangle whatever growth you had.

Here’s the part brands hate saying out loud: most don’t even know what they’re cultivating. They plaster “data-driven” on their slides while their analytics resemble an abandoned lot—overgrown, tangled, and clearly missing a gardener.

I’ve skimmed through a handful of podcasts lately circling this theme. Not all of them hit home, but a few cracked open useful starting points. Some sharp, some questionable, but enough to help frame where this conversation should actually begin—and which voices are worth bringing into it.

If we’re serious about the “value of data,” the question isn’t what multiple you can cook up for your next investor pitch. It’s whether you’ve grown a garden that can actually feed you—or just another patch of weeds you’ll keep mowing down every quarter.

Think of data as ingredients. Event logs from every click, tap, and binge session. Frankenstein user tables full of hashed emails and stitched-together attributes that marketers squint at and insist are “people.” Product catalogs with SKUs and prices. And the datasets your CFO actually respects—sales, churn, lifetime value.

That’s the pantry. And here’s the irony: most companies don’t realize how much is already sitting on those shelves.

Joel Acheson, founder of Catalytics, pointed out to me that the raw logs from web servers and ad servers provide more than enough information to build an efficient advertising operation. Every time a company drops a pixel onto a site, that pixel is quietly generating its own set of logs.

In other words, the pantry keeps stocking itself.

Here’s the kicker: brands already own this data—often for free—yet they spend billions outsourcing the privilege of having it parsed and then sold back to them in glossy dashboards. Acheson doesn’t sugarcoat it. The so-called “value of data” may be less about the numbers themselves and more about the collective revenue of vendors repackaging logs the brands already control.

So maybe the punchline isn’t that data is the new oil—it’s that the real profit center is the toll booths vendors set up to sell you access to your own pantry. Every pixel, every log, every line of server data is yours. Yet somehow, entire budgets get burned buying back the same numbers in shinier wrappers.

And even then, owning the pantry isn’t enough. Gary Herman, co-founder of madSense, makes the point that’s often skipped: raw data isn’t the oil—it’s the labeling and structuring. Without a disciplined taxonomy, all those shelves of ingredients just collapse into chaos. Companies end up drowning in scattered, constantly changing inputs that are functionally useless. Even modest campaigns spanning three devices and half a dozen platforms can generate billions of potential outcomes. At enterprise scale, the numbers explode into the hundreds of billions. Without order, the pantry doesn’t feed anyone—it just becomes sludge.

So the paradox is simple: you already own the ingredients, and they’re more than enough to cook with. But without structure, discipline, and the willingness to use what’s right in front of you, you’re not running a kitchen—you’re just paying someone else to walk through your pantry and hand you back your own food in a prettier bag.

Roger Vasquez, VP Technology Solutions at Clinch, offers another perspective over email: data isn’t oil, and maybe it’s not even a pantry. It’s sourdough.

As a Portlander, he picked up baking during COVID. Sometimes he sourced better flour, sometimes he experimented with yeasts, sometimes he kneaded and rested endlessly. Results varied. The lesson: better ingredients without process fail. A great process with fourth-hand ingredients also fails.

Predictability requires both. A good bake is about process + inputs, not possession. And that’s the real endgame of data: validation, predictability, and outcomes. Not just claiming to “own” the flour.

In Jewish law, reshus means rightful ownership or control. In data, the same principle applies: if you can’t prove reshus, you don’t own anything—you own a liability.

Oliver Gwynne, data strategist at 173tech, says the industry doesn’t even talk about it: “I don’t see chain custody being a hot topic, and no one has really asked me about it. Most data capture is done automatically—through forms, into CRM systems—and if it’s for analytics, personal details rarely even surface. The more stipulations are put around consent, the more companies work around it.”

Translation: ownership isn’t the problem. Usability is. And that’s a rabbit hole worth going down.

Every deck in adtech promises a “moat.” But let’s be honest—most of those moats are puddles. If a competitor can replicate your dataset in twelve months, it’s not a moat. It’s a commodity.

Gwynne has seen the hype fade: pandemic Slack groups? Dead. WhatsApp groups? Chaos. Facebook groups? Full-time unpaid jobs. “Inevitably some moats will survive,” he admits, “but I doubt the juice is worth the squeeze.”

Mattia Fosci, founder of Anonymised.io, sharpens the knife: “The most valuable data is also the least scalable—purchase intent, conversions, financial transactions. Open web data, with a few exceptions, is a big bucket of manure… It’s worse than a coin toss.”

Once upon a time, plumbing was the differentiator.

Get your pipes right and you could out-activate competitors. Today, everyone can move data anywhere. The new edge is whether it’s structured and annotated for AI.

Gwynne doesn’t mince words: “If you have centralised your data, then you can pipe it any which way… having the data ready for AI, with row-level descriptors and notes, is activation that is worth £££.”

Gary Herman adds the enterprise perspective: extracting insights during the campaign’s lifecycle to adapt and change tactics based on the data turns the sludge into gas at the pump.

Taxonomy is the silent killer. Without consistent schemas and definitions, data is a bonfire of wasted resources.

Gwynne recalls Fyxer AI, a client that spent three days just defining MRR as their “north star” metric. Painful, but worth it—they’re now one of the UK’s fastest-growing companies. “This is the exception to the rule,” he admits. “Most companies want their entire business to revolve around data but they don’t care about the definitions—or want to outsource the pain.”

Herman has seen the same thing: plenty of first-party data, zero labeling discipline. What should be an asset ends up as chaos. And that chaos is a terrible business strategy.

Buyers used to pay for promises. Now they demand evidence.

But Gwynne is blunt: “The real problem here is that people don’t know how to use data—even if they have the perfect data and the answer is staring them in the face. And it’s virtually impossible for a third-party to regulate or certify the quality of data, because you can just fabricate it.”

Fabricate it? That doesn’t sound good at all.

Subscription models should reward quality. In practice, they don’t. “Most consumers just want as much data as they can get, for the lowest cost,” Gwynne says.

Fosci warns the same is true for measurement: if we measure outcomes in the wrong way—by selecting bullshit KPIs—then the value of data is distorted. Scale beats quality every time.

And Judy Shapiro, CEO of EngageSimply, delivers the knockout: “Nearly half the data used for ad targeting is wrong.”

Profiles attach intent to accidental behavior, fail to update after life changes, and mistake bots for people. “The fixation on data—who owns it, how to mine it, how to manage it—was born of an unhealthy obsession with surveillance marketing… Here we are, with a system designed less to help advertisers than to bilk them, while intermediaries rake in billions in ill-gotten gains.”

So what’s the value of data? It isn’t in possession. It isn’t in puddle moats, vanity dashboards, or leaky consent forms.

Gwynne’s verdict: buyers still chase cheap bulk.
Fosci’s warning: bad KPIs let scale masquerade as success.
Shapiro’s reality check: half the signals are garbage.
Acheson’s frustration: you already own the logs but pay billions to buy them back.
Herman’s reminder: without taxonomy, everything collapses into sludge.

Most so-called “data assets” are moldy sourdough starters at the back of a fridge—once full of potential, now rancid.

The survivors won’t be the loudest “data-driven” evangelists. They’ll be the ones willing to rethink the model itself: away from fake moats, away from vanity KPIs, away from profile manure, and toward something provable, actionable, and genuinely scarce.

Because at the end of the day, the market doesn’t pay for potential. It pays for proof.

And right now, proof is in dangerously short supply.

If defining your data was about knowing what’s in the pantry, then measuring its health is about asking whether it’s actually safe to eat. The market is full of vendors parading around their so-called “assets,” all of them sounding impressive until you peek under the hood and discover duplicated IDs, stale signals, missing fields, or datasets so old they belong in a museum exhibit.

Oliver Gwynne of 173tech put it plainly to me: “The real problem here is that people don’t know how to use data—even if they have the perfect data and the answer is staring them in the face.”

That ignorance is why health scores matter. Without health, there is no value. Data health becomes the litmus test between fiction and fact, between marketers who sell dreams and those who actually deliver outcomes.

A health score moves the conversation away from gut feelings and into the realm of proof. Engineers don’t have to mumble about null values or mysterious “pipeline issues.” They can point to a number—this dataset is ninety-two percent complete, reliable despite a few gaps—and suddenly the conversation shifts.

Numbers speak a universal language. Once quality is quantifiable, suspicion fades, collaboration improves, and prioritization stops being a fight.

This is where Clinch fits neatly into the frame. Roger Vasquez, their VP of Tech Solutions, has argued that data is never the end in itself—it’s only as valuable as the process that makes it predictable. A health score, in Clinch’s view, isn’t marketing fluff. It’s the bridge between raw ingredients and the validated predictions marketers actually use to justify spend. Without that process discipline, you’re not baking bread—you’re hoarding flour.

And yet again, Gwynne reminds us of the elephant in the room: “Most consumers just want as much data as they can get, for the lowest cost.” Health scores can’t cure the obsession with volume, but they can force the uncomfortable truth into the open—cheap isn’t the same thing as good.

Data health, like a physical exam, is never just one metric. It is coverage, accuracy, freshness, actionability, and uniqueness all at once.

Clinch often frames this as the difference between prediction as theory and prediction as product. If health fails on even one of these dimensions, you’re not fueling campaigns—you’re selling fertilizer.

Gwynne adds a sharp caution here: “It’s virtually impossible for a third-party to regulate or certify the quality of data, because you can just fabricate it.”

That still bothers me. Does it bother you?

Which means a health score isn’t just a number to wave around—it has to be part of a cultural discipline of transparency. Otherwise it becomes just another marketing trick.

What makes the idea of a health score powerful is not just that it measures hygiene but that it makes hygiene financial. Coverage, accuracy, and freshness aren’t only technical checks—they flow directly into pricing and margin allocation.

This reframes the exercise as buyer-first. The criteria reflect how real buyers make decisions: deduplication, clean-room readiness, endpoint acceptance. Leakage and collision are no longer ignored but treated as first-class issues. Freshness is finally recognized as dynamic—intent data losing value in hours while demographics decay over months.

Clinch leans into this framing because it aligns directly with activation. If the data isn’t portable, annotated, and fresh enough to drive campaigns, it doesn’t matter how “big” the rows are. Health is the only thing that transforms signals into spend.

And incentives shift with this framing. Those who curate and maintain quality are rewarded with premiums, while those who dump raw signals into the market are treated like commodity suppliers. Gwynne’s own verdict fits neatly here: the juice is rarely worth the squeeze unless it proves both scarcity and freshness.

The breakthrough comes when health stops being internal housekeeping and becomes a visible price signal. A dataset scoring in the nineties commands premium margins. One in the seventies still moves, but without rarity. Anything in the fifties or below is nothing more than bulk wheat—stripped of differentiation and sold at base yield.

By tying health directly to pricing, the market gains a discipline mechanism. Sellers have proof of value, buyers can defend their spend, and the theater of vague negotiations starts to collapse.

Clinch’s position is clear: the only way to restore trust is to make quality a price signal. If your dataset is alive, structured, and ready for activation, it deserves a premium. If not, it belongs in the commodity bin.

Gwynne is realistic enough to know that buyers won’t abandon their appetite for cheap bulk, but a clear score at least forces the tradeoff into daylight. If you still buy the stale loaf, you can’t pretend you didn’t see the mold.

Data health isn’t just QA—it’s currency. It transforms datasets from vague “assets” into priced commodities. It rewards curation, punishes sloppiness, and disciplines a marketplace that has thrived for too long on opacity and promises.

The sourdough analogy still holds. It’s not enough to own the starter—you have to keep it alive, feed it, and prove it’s worth baking with. Otherwise, all you’ve got is a jar of moldy goo.

And as Oliver Gwynne reminds us, most buyers are still happy to pay for jars of mold, as long as they’re cheap.

The challenge now is whether health scores—and the companies like Clinch operationalizing them—can finally force the market to tell the difference.

This isn’t another acronym explainer — it’s a field manual for the post-cookie ad economy. Inside ADOTAT+, we take ArcSpan’s AAA Machine and peel it open like a circuit board.

We map the Unified Event Model. We show how Consent Ledgers make regulators sweat less and investors smile more. We pull receipts on Synthetic Scarcity, why it’s the next premium lever, and how the smartest publishers are packaging data like it’s 1985 Bordeaux.

Then we follow the trail to incrementality’s dirty little secret — where experiments rot from post-bid contamination, how “scientific rigor” became a party trick, and what the real math warriors (not the deck jockeys) are building to fix it.

Subscribers get the blueprint — not the buzzwords.
👉 The data pipelines.
👉 The testing frameworks.
👉 The companies quietly using AAA discipline to outgrow the noise.

ADOTAT+ is where the puppet strings get cut.
Stay Bold, Stay Curious, and Know More than You Did Yesterday.