Notes

Publicis agreed to buy LiveRamp at a 29.8% premium. It can buy the platform. It can't buy the trust.

Part of what made LiveRamp valuable: no agency group controlled it.

More than 25,000 publisher domains and over 500 technology and data partners connect through it.

LiveRamp owns the technology. It doesn't own the relationships that make the network worth joining.

That is a rake on loan.

Publicis already owns Epsilon and Lotame. Its promise: LiveRamp stays "a neutral, interoperable platform." No customer restricted.

I believe them. It doesn't matter.

A pledge is not an architecture. Whoever owns the platform sets its prices, its roadmap and its priorities.

Every rival already owns its own matcher. Omnicom has Acxiom. Dentsu has Merkury. WPP has InfoSum.
LiveRamp was where they could all meet.

My bet: the pledge is kept, and rivals steer their clients elsewhere anyway. Not because access is denied. Because a competitor now owns the room.

Neutrality is not a feature you can acquire. It's a fact about who owns you.

Infillion is buying Foursquare. The trade press read it as ad-tech consolidation.

I read it differently.

Foursquare spent a decade doing what every sensible data business does. It did the enrichment itself. It sold direct. It kept its share.

It still ended up one link short of the buyer's decision.

This week's newsletter is about that one link: where it sits, who is about to hold it, and the question every data business should ask of its own.

Spirit sold its customer data for ten million. Google kept the rest.

Spirit Airlines put its customer data up at auction and Google took it for ten million dollars, paid once, at closing, with no claim on anything built from it afterwards. Google handles the enrichment. Whatever that corpus turns out to be worth inside Google's models, it is worth it to Google.

Ten million looks like the price of a corpus. It is the price of a corpus somebody else still has to finish.

In the same week, CuriosityStream told its investors that licensing its library to the AI labs was running at around seventy-three percent gross margin, and that it intended to keep going. It clears the rights, builds the datasets and deals with the labs itself, so there is nobody in the middle taking a cut of the work.

Almost the same product, and the same kind of buyer. What separates ten million paid once from seventy-three percent paid over and over is which of the two had already done the work that makes a corpus sellable.

By August, Spirit had no move left. That was settled years earlier, in whatever meeting decided rights clearing and curation were somebody else's job, and a bankruptcy auction is not where you revisit it.

Value goes where the work is. The work gets done long before the buyer turns up.

Three prices for the same company in six years: 6.9 billion, 12.5, 7.7. Nobody did anything wrong. A sentence was deleted.

For twenty-four years, every company that wanted to be paid by the US government needed a number only Dun & Bradstreet could issue. The government paid about nineteen million a year for it. That was never the money.

Then the government started writing its own number. No revenue line disappeared that day. The price did, one renewal at a time.

Most data businesses run on a key they did not make. This edition is about how to find yours, and what it is worth to whoever holds it.

RBS owned the data. Its partner owned the enrichment and the customer. The partner just sold for $230 million, cash.

The data stayed exactly where it was. The price went to the layer built on top of it.

Reward spent more than ten years building card-linked rewards on RBS transaction data, and ran the same model across dozens of banks. The offers, the merchant deals, the programme the cardholder actually sees: all Reward's. In February Rezolve, a Nasdaq-listed AI commerce company, paid $230 million for it. Not for the data. Reward doesn't own the data, it owns everything that makes the data worth acting on.

I know this structure from the inside. I was COO at Reward for many years, when we launched the RBS programme, and later CPO at dunnhumby. The data owner watches a partner build on top of their substrate and thinks the arrangement is secure because they control the supply. It is secure until someone prices the enrichment.

The rake goes to the party that owns the enrichment and the customer relationship, not the party that owns the data underneath it. If a buyer priced your data partnership tomorrow, which side would they pay for?