WPP x InfoSum
'Playing Leapfrog' (or leaping the frog šø š¤ )
Friends of NFTBC.
Since starting NFTBC earlier in the year, my biggest flop so far is the WPP post - in terms of engagement and share price performance (both dismal!) My only regret however is the subtitle (āā¦Multibagger?ā) I had 15 subscribers at the time (200 now) and with hindsight I was trying a little too hard simply to get anyone to pay attention - anyone at all. To rub salt in the wound, I was shortly thereafter introduced to Betteridgeās Law of Headlines which states āAny headline that ends in a question mark can be answered by the word noā š¤¦āāļø.
I will continue to learn lessons and to evolve this blog, and in the coming weeks I intend to provide an update on future direction. But one thing I have taken away from the WPP post is that I was selling it too hard. I donāt really want NFTBC to be about me trying to persuade anyone about anything (or perish the thought - to buy anything). One of my favourite writers, the philosopher John Gray, often says that he doesnāt write to persuade people, but to get them to think - to engage with his ideas. In a similar vein, I want to write about stocks and related matters that I think are interesting, overlooked and worthy of some kind of reflection. If I find such things interesting, my hope is that there might be some sliver of audience out there who does too, whether they ultimately agree with me or not. As it happens, the stocks I write about will often be ones that I either currently own or am thinking about owning. Iāll write more about this at a later date.
Back to WPP. Outside of WPP itself, I donāt personally know anyone whoās bullish on the shares. Even inside the company, WPP is by no means universally popular. Iāll admit that WPP might very well be hated for good reason, but on the other hand I also note that for one of my best compounding stocks of the last five years (IRR >25%) there was a very similar backdrop initially - everyone seemingly hated this company. Investors, employees, customers - you name it. The company was accused of having an antiquated strategy that was no longer working - it was ex-growth and going extinct, a dinosaur. In reality, this company was trying something different, and it took a little time for the results to become fully manifest - today this business is a double-digit grower and still accelerating.
WPP, too, is trying something different. And that makes it interesting to me. But doing something different frequently comes across as weird, eccentric or just plain foolish, unless and until it starts showing results. In the latter part of 2024, it seemed to me that WPPās strategy was showing early signs of success and setting the scene for a modest acceleration this year. But then along came Q4 and we seemed to go back to square one. I wonāt revisit the full story here, but please do dive in if youāre intrigued.
The short version is that WPP has been reinventing itself around end-to-end marketing automation - essentially an internally-developed marketing software stack called āWPP Openā that can be used both internally and by clients. Everything that can be automated, has been or soon will be, while still leaving a central role for WPPās ācreativesā - genuine human creativity, being one of the few remaining things that cannot yet be automated (and perhaps never will be). If it works, then marketing becomes cheaper and more effective - and in a way thatās tough to replicate, lowers client churn and operates outside the walled gardens of the digital ad giants. Itās too soon to conclude whether itās working and to what extent, but investors are hardly looking at all - in the meantime I continue my investigationsā¦
The āMissing Pieceā
In my WPP post I discussed the companyās sophisticated data and AI strategy, all geared towards increasing the effectiveness and economics of advertising for clients. The perceived problem is that, relative to its major peers, WPPās own proprietary data assets are more modest in size. Iāll get into this a bit more later, but it really helps to explain why French rival Publicisā media business is very significantly outperforming WPPās, for example.
In early April WPP acquired InfoSum. Hereās how the press release put it:
WPP today announced the acquisition of InfoSum, the worldās leading data collaboration platform. InfoSum will join GroupM, WPPās media investment group, to power the creation of a new generation of AI-enhanced marketing solutions for clients, delivered through the industryās most powerful and secure data infrastructure.
The acquisition is a major strategic step forward for WPPās AI-driven data offer, giving WPP and its clients immediate access to the industryās largest cross-platform source of privacy-safe, actionable data for marketing intelligence, audience targeting, and AI model training.
āThe worldās leading data collaboration platformā. OK then. Thatās kind of a big claim, but also vague - so Iāve spent some time kicking the tires on InfoSum. Hereās how Mark Read weighed things up at Q1:
The InfoSum transaction is going to strengthen significantly our data proposition and we're seeing that with clients. So I think greater clarity on that is perhaps the thing that we have been missing a little bit.
And per Evan Hanlon, CEO of Choreograph (WPPās data business): āNow, we have all the piecesā. So we have here the prospect of a āmissing pieceā type acquisition. Iāve been wracking my brain to think of examples of genuinely transformational acquisitions where the insertion of that āfinal pieceā really did make all the difference. Itās perhaps not especially common, but there are such examples - Nvidiaās acquisition of Mellanox comes to mind. Which ones can you think of? I was initially tempted to include Publicisā acquisition of Epsilon, but Epsilon was perhaps less of a missing piece and more of a strategic change in direction.
The Debate
If you listen to WPPās earnings calls, there are frequently questions of the sort āwhen are you going to buy a database?ā The reason originates in Publicisā 2019 acquisition of Epsilon which brought with it hundreds of millions of IDs linked to individuals. Marketers use such IDs for targeting and measurement to make their campaigns more effective - and by bundling its data offering with its media offering, Publicis has been more competitive than peers at winning accounts. Hereās what Publicis CEO Arthur Sadoun has to say about it:
If you donāt add proprietary data that will bring to your clients something that they donāt have, they wonāt find new sources of growth. They wonāt be able to connect every individual to the entire media ecosystem.
Publicis recently doubled down on this strategy with the acquisition of Lotame. And itās also a strategy being pursued WPPās other largest peers - as exemplified by IPGās acquisition of Axciom and Omnicomās pending acquisition of IPG.
Hereās how Brian Lesser, GroupM CEO, framed WPPās side of the debate at Q4 2024:
The market is quickly moving from identity-based solutions to AI-driven connectivity. Hereās what that means:
We believe that data connectivity is more important than simply owning data, and the value of connectivity will only continue to grow
No matter how many traditional IDs you own, it will never be enough. The real power lies in connecting data across publishers, partners, retailers, platforms, and clients, and thoughtfully using our own first party data assets in conjunction with the available data in the world.
The future is about connecting disparate data sets to extract insights, create predictive models, and drive performance
Traditional ID solutions, like those grounded in email addresses, only learn from overlapping data points, relying on outdated lookalike models that limit insights. Using technologies, like federated learning, we can create shared knowledge and predictions across all our partners, without sharing raw data and activate via a simple connector. Itās a game-changer. And as Mark mentioned, this will drive significant efficiencies in audience targeting and real-time campaign optimisation, tailored to clients' unique needs based on their own data maturity.
By 2030, weāll be working with thousands of data partners to guide audience decisions. The only way to harness that scale is by leveraging AI at every step of the process. No singular, legacy database can manage that scale and complexity.
And he shared this slide:
And here is WPPās Alex Steer (chief data officer at Choreograph):
We reckon that only around 10 to 20 percent of data thatās useful for decisions that we make in marketing can be resolved back to an individual ID about an individual person. Most of the information that businesses own comes in different types and sits in different structures. You might have data thatās organised by store, or by postcode, or by keyword. AI is really good at translating between those things, and at learning and spotting patterns.
Which approach is better then - ID or āfrom ID to AIā?
Frankly the debate continues and there are both sceptics and optimists. But the thing is, the ID approach is clearly working for Publicis today and at scale. Whereas WPPās solution is earlier in its implementation, itās new, itās different and it raises challenges of its own - how do you deal with the trust and data privacy issues of bringing different parties together to collaborate on their proprietary data? How can you make it scaleable? Now bear in mind that WPP had not announced the InfoSum deal at the time Lesser produced his Q4 slide, but one way of looking at it is this - InfoSum is the infrastructure that allows WPP, brands, retailers, publishers, data providers and others to connect all of those ānodesā together to generate unique insights, and to do so securely, quickly and easily. If federated learning is WPPās strategy here, then InfoSum really is starting to look like a āmissing pieceā.
InfoSum
InfoSum was founded by Nick Halstead in the UK in 2016. A software engineer, serial entrepreneur and racing driver, heās a pretty interesting fellow and you can listen to his story here. Halsteadās prior two companies had been in social media, and he had spent years working in partnership with the usual suspects:
I was really thinking about what would level the playing field for everyone elseā¦ to think about about how do you solve for data, making every company be as data rich as Facebook and Googleā¦ Some of the largest public companies out there may still have access to a couple of hundred million people, but these days itās a global economy and people want to be able to market to literally everyone like you can on Facebook. And I was thinking about the technology that would break down those barriers of those big companies to be able to work with each other.
And the barriers he was talking about: trust, privacy, identity and integration. He looked at a number of approaches to integrating the first-party datasets of separate businesses but they all required centralisation, co-mingling, reversible encryption and such things. He raised venture funds and it took his team three years to build what would become InfoSum, launching in 2019. The approach itself was highly innovative, centring around what they call the ānon-movement of dataā. InfoSum clients never send each other their data - the system is completely decentralised, operating via patented mathematical models that never send any underlying identities.
By solving for ānon-movement of dataā InfoSum was able to break down Halsteadās four barriers. Customers can now collaborate without fear of data āleakageā or losing commercial value to other parties - instead the parties can focus their collaboration on the upside of bringing unique and disparate datasets together (non-zero sum). The same goes for privacy concerns - customers never send their data to anyone or hand over control. Moreover, customers are able to match their shared identity footprint but without ever exposing the underlying identities, via non-reversible synthetic identifiers. Having been built from the ground-up to address these barriers, InfoSumās argument is that most alternative data clean rooms have been designed as āadd-onsā. Per Halstead:
[What is] utterly unique to InfoSum is that we combine the clean room, the identity matching and the data processing all through the lens of non-movement.
And because data is never moved, it makes InfoSum much faster to onboard and to use both from the perspective of data extraction and upload and from a contracting perspective. According to InfoSum CEO Lauren Wetzel, clients save 156 days on average with InfoSum, seven times faster than other platforms - while I havenāt audited these claims, it is easy to recognise how a decentralised platform could be a lot faster logistically.
Once businesses are on the InfoSum platform, it then becomes very easy for them to collaborate - they just have to agree to:
InfoSumās extensive global data network represents hundreds of billions of data signals across multiple dimensions of data from media platforms including Channel 4, DIRECTV, ITV, Netflix, News Corp, and Samsung Ads, as well as major retailers around the world and identity and data partners including Experian, TransUnion, Circana, Dynata, and NCSolutions.
While we donāt know how big InfoSum currently is, media sources have suggested WPP paid around $150m. In any case, the deal was not a major surprise given the various hints that Lesser dropped at Q4 - moreover, Lesser himself was InfoSum CEO for four years prior to Wetzel. Cynics, of course, are saying that Lesser was self-dealing. Others have speculated that InfoSum was ātreading waterā with investors looking for a way out. For her part, Wetzel has stated on the record that InfoSum was doing a fundraise earlier this year (which included WPP) and during the process it became clear that the option of a takeover was compelling. With WPP already a client and familiar with the product, it was becoming clear that there was a clear strategic fit and shared mission.
Where Next?
In WPPās words, the acquisition allows them to:
leapfrog traditional identity-based solutions that depend on decades-old, deteriorating databases weakened by cookie deprecation, platform fragmentation, and splintering audience match ratesā¦
But the power of the transaction, is what it means in the context of WPP Open and GroupM more broadly and by integrating InfoSum's capabilities within WPP Open, our clients are able to unlock the full potential of their customer data and reach through AI in strong, quality media environments that's not always available to people using legacy first-party data systems. And using federated learning techniques, clients are able to build, train, and deploy custom AI models that they can use in a privacy compliant way. They can generate insights and audiences to create precise predictive models, optimized campaigns, and deliver measurable improvement in real time.
By integrating InfoSum directly into WPP Open, it should, in theory, make both propositions more attractive to customers. InfoSum customers can benefit from the application of WPPās AI/ML capabilities in the interrogation of their datasets, while WPP customers will get seamless access to InfoSumās enabling infrastructure to power their data collaborations. Moreover, one of the apparent reasons for the acquisition is to accelerate InfoSumās roadmap, with WPPās scale and resources - Wetzel (who is remaining as InfoSum CEO) believes theyāll be able to scale very quickly.
One outstanding question is whether InfoSum will lose its former neutrality. Will potential WPP competitors still want to partner with it? One has to assume that some degree of neutrality will be lost. But we have been told that InfoSum will continue āoperating under its own identity and with its own go-to-market, working directly with clients who want to be able to do data collaborationsā.
I guess weāll just have to see how things evolve from here. Per Mark Read at Q1 2025:
So the conversations we had with clients were very positive. Dozens of conversations over the last few weeks and the feedback very positive and then we're confident the deal alongside our broader simplification efforts will drive a step change in performance, particularly in competitive reviews.
Some degree of grace period must be allowed, after which Iāll be on the lookout for that step change. But it does seem to me that Mark Readās window of opportunity could be at risk and without signs of progress soon, there could be increased activism for change at the top. He has been CEO since 2018 and, despite profound change beneath the surface, thereās little to show in the financials or share price. If GroupM can accelerate to the sorts of growth rates achieved by Publicisā media business, at 40% of the group total it will be enough to underpin WPP as a whole and warrant a significant P/E re-rating. Perhaps weāll start to learn in the coming months if the puzzle is indeed complete now.
As always, get in touch if you have any comments or questions.