Identity Graph Race: TransUnion, Experian, and Zeta Build Proprietary Stacks

Full analysis

Two identity giants are spending billions to build "proprietary" graphs that increasingly aren't. TransUnion assembled its stack through TruSignal, Signal, Tru Optik, and the $3.1 billion Neustar buy. Experian bought Tapad, then Audigent. Zeta grabbed LiveIntent, ID5 grabbed TrueData. The question for every buyer and every publisher: if two graphs draw from the same licensed feedstock, what exactly are you paying twice for?

This is a Type 1 call for the acquirers, hard to reverse, they've already spent the money. It's a Type 2 call for the buyers, easy to reverse, they can switch vendors or renegotiate any quarter. What's actually being decided isn't who has the biggest graph. It's whether "proprietary" survives contact with the fact that everyone licenses from the same suppliers. Forcing function: audience-overlap reports and renewal cycles, which land quarterly, not eventually.

The Market Analyst: TransUnion and Experian are priced today as if they own durable data assets. If the differentiation thesis holds, that's fine. If it collapses, their identity revenue gets repriced from "proprietary asset" to "data reseller," and that's a real haircut in the multiple, meaning investors pay less per dollar of that revenue. Zeta is the more interesting position: LiveIntent's email-based signal is structurally harder to copy than Neustar-style assembly. In plain terms, a database everyone can rent is worth less than one you built and nobody else can. The market still hasn't made that distinction.

The Skeptic: Steelman the incumbents. This convergence critique isn't new. Analysts have written this exact piece since 2019, and TransUnion is still standing. Cookie deprecation kept slipping, so the urgency that justified the Neustar price softened, yet the graph still gets bought. For the thesis to bite, buyers must notice the overlap, act on it, and move faster than the incumbents can acquire the next scarce asset. None of that is fast. And TransUnion and Experian have balance sheets. They can just buy the differentiated signal too. The plain version: being big and boring has outlasted a lot of clever critiques.

The Operator: The pain shows up in procurement, not the strategy deck. Buyers running TransUnion-Neustar and Experian-Tapad side by side are matching near-duplicate tables and paying two vendors for one answer. Campaign teams will see it in overlap reports this quarter, if they bother to run them. The second-order effect matters more: any publisher with a logged-in audience, even mid-sized, suddenly has leverage it didn't have six months ago, because buyers need a first-party anchor to get off the licensed-data treadmill. Plain version: if two suppliers hand you the same phone book, you stop paying for the second one.

The Customer / End User: The customer here is the DSP and brand buyer, and they've been sold "proprietary resolution" twice over. What they actually want is reach they can trust and a match rate they can't get elsewhere. Convergence means the premium they pay for a second graph buys almost nothing. That's the moment publishers and retail media networks get to charge for direct access to their own logged-in users, because that signal genuinely isn't for sale from the graph vendors. The buyer's rational move is to route more budget toward owned-data partners and treat the graphs as commodity backfill.

The CFO: Look at the sunk cost honestly. North of $5 billion combined went into stacks that share feedstock. That money's spent; the question is what the next dollar buys. Buying more graph scale compounds the convergence problem. Buying owned, consent-based signal, newsletters, loyalty programs, portals, is the only spend that widens the gap. The trap is defending the asset already on the balance sheet at its acquisition price long after the differentiation eroded. The $3.1 billion Neustar tag anchors perceived value; it shouldn't anchor the next capital-allocation decision.

The tensions. The Skeptic and the Strategist-minded reads part ways on speed: does convergence bite on a quarterly renewal cycle, or does it stay a slow-burn critique the incumbents outrun with their next acquisition? That's the whole disagreement. The Operator says buyers will see the overlap in this quarter's reports; the Skeptic says seeing it and acting on it are years apart. Second tension: the CFO and the Market Analyst agree first-party signal is the scarce asset, but the incumbents can buy that too. So is Zeta's LiveIntent edge durable, or just the next thing TransUnion writes a check for?

What it hinges on. Two beliefs. First, whether buyers actually consolidate vendors when they notice duplicate match tables, or keep paying for redundancy out of inertia. Second, whether owned first-party signal stays scarce, or gets rolled up the same way the graphs did. If both cut against the incumbents, the graph business reprices. If buyers stay lazy and the incumbents keep acquiring, the moat holds long enough to not matter. The council leans toward the convergence thesis being right on direction and slow on timing. The scarce asset is the signal you make yourself. The mistake is assuming the market prices that in this year.

To de-risk: run the overlap report. If your two graph vendors match the same 80% of your file, you have a negotiation, not a data strategy. And watch where the next M&A dollar goes. Graph scale, or owned-data feedstock.

Prediction: By the end of Q1 2027 (covering Q4 2026 earnings and renewals), at least one of TransUnion or Experian will announce an acquisition of a first-party or consent-based owned-data source, newsletter, loyalty, or portal data, rather than another graph-assembly asset.

Confidence: Medium. Incumbents with balance sheets buy their way out of commoditization.

Why: The article's own logic is that scale on licensed feedstock converges, so the only spend that widens the gap is owned, unlicensable signal, and TransUnion and Experian have both the cash and the pattern of buying whatever the scarce asset is. Zeta already moved on LiveIntent's email signal and ID5 grabbed TrueData, so the pivot toward owned data is already visible across the peer set. The opposite outcome, another pure graph roll-up, is less likely precisely because both firms have publicly leaned into differentiation messaging and know a duplicate graph adds no defensible value. The main risk to the call is timing, not direction: deals slip, and a quiet quarter is possible.

Revisit by 2027-03-31: We're right if TransUnion or Experian announces a first-party/consent-based owned-data acquisition. We're wrong if the only identity M&A from either is more licensed-graph scale, or nothing at all.