What you need to know

• When Gartner released its first CDP Magic Quadrant last week, it hinted strongly that the category is peaking and consolidation might be next.
• Then a key Gartner analyst, Lizzy Foo Kune went further, saying CDPs have arrived in the analyst firm's infamous "trough of disillusionment".
• Jeff Lunsford, the CEO of Tealium - one of four companies in the leadership quadrant - gave Mi3 a very different take. He says AI will drive further adoption at least in Tealium's "category of one", a comment suggesting the martech vendor is in its own differing league when it comes to CDPs.
• He says companies have been solving problems around data collection, enrichment and distribution to improve customer experiences. They have now realised investments into those efforts solves the same kinds of problems they will face getting data ready for AI models.
• And he says, Tealium can help with the next step - activation.

It’s a potentially confusing time in the world of Customer Data Platforms (CDPs). Last week, Gartner came out with its first CDP Magic Quadrant, but the report also flagged a view within the research outfit that the market may be peaking and about to consolidate.

A few days later, Lizzy Foo Kune, the Gartner VP who authored the report, went further, telling Adexchanger CDPs are now deep in another Gartner-inspired meme – the Trough of Disillusionment.

Rival Forrester meanwhile, says the party is just getting started, at least in Asia-Pacific. 

To get a better read on what is really happening, we’ve gone back to the CDP source, and one of the sector's pioneers, Jeff Lunsford, CEO of Tealium since 2013, and an investor in the business prior to that. Tealium was one of the four companies listed in the recently released Magic Quadrant for CDPs. Mi3 spoke with Lunsford the day after we broke details of the Magic Quadrant, but before Foo Kune’s comments about the sector languishing in the Trough of Disillusionment.

Lunsford says he isn’t fussed by Gartner’s characterisation of a maturing market, and one ripe for consolidation. Instead he claims Tealium is in a category of one when it comes to CDPs. But perhaps more importantly, he describes how the rise of AI creates renewed impetus behind the adoption of the technology due to the primacy of data.

He believes the company's heritage leaves it centrally placed to deal with the emerging era of AI.

When Tealium was founded in 2008, the problem it wanted to solve was to create a neutral layer atop the enterprise data architecture in global enterprises who, at the time, were grappling with the problem of connecting multiple sources of data across multiple geographies in highly scaled environments.

Those early use cases looked a lot different to the complex world of enterprise IT that has subsequently emerged. When Lunsford first became a board member at Tealium 13 years ago, martech was in its infancy. The giant marketing clouds from companies like Adobe, Salesforce and Oracle were yet to emerge, and Scott Brinker's famous marketing landscape had barely 150 martech tools in it, not the more than 11,000 he tracks today.

For many marketers and IT managers, their first encounter with Tealium likely revolved around tag management – and what the company used to call its digital wallet – effectively collating all the digital fingerprints a consumer left behind to determine and validate as precisely as possible, the identity of a consumer. That challenge gave the company the incentive to master data complexity.

“We've always run towards complexity, the complexity of data collection, enrichment, and distribution and activation," says Lunsford.

Lunsford also has the benefit of having built the business working with some fairly unforgiving customers. He describes Tealium's almost 800 enterprise customers around the world (including almost 70 in Australia) as operating in the most hardened and audited industries on the planet including finance, telco and healthcare.

“That’s where all of this stuff really, really matters - security matters, privacy matters and compliance matters,” he says.

Next wave with AI

Solving those problems for its clients also positions Tealium well for the next wave of corporate data demands, AI and especially generative AI. As we wrote in Part Two of our investigation of CDP markets, all the effort companies invest to implement CDPs as part of their plans to ensure world-class customer experiences in a more privacy-demanding world is also necessary if they're to leverage the emerging world of AI.

As the Customer Data Institute's Simon Periera told us at the time, the role CDPs will play in getting brands AI-ready reflects a major challenge everyone must face sooner or later.

“AI does not provide you a competitive advantage, it is an equaliser, it helps you catch up. If you don’t have the creative resources that a big tech company has, you don't have the wealth of knowledge around marketing processes it helps you catch up, but it does not deliver a competitive advantage," Periera said.

By contrast, Periera saw the only way of driving competitive advantage from AI is by utilising your own data. “All of the work required around a CDP is the same foundational work you need to do to get AI ready for the AI projects you will have to execute in the next sort of five to 10 years." 

That’s a message Lunsford is also now proselytizing. “For this massive AI wave, AI needs data to be successful [because] without data you have no AI.”

According to Lunsford, “As every company was thinking about collecting customer behavioural data, enriching it, distributing it and activating it, they were thinking about it in the context of the customer experience, really up until about a year and a half ago.”

Then came ChatGPT, the breakthrough AI platform which has popularised a technology that had been bubbling away in labs for 40 years.

“Now everyone is realising that ‘not only do I need all this data to create personalised experiences, I also need all this data to train my AI models’," Lunsford says. "And they may be chatbots or pricing algorithms, they may be next-best action systems. But without the data, the algorithm is just an empty algorithm. The model is just an untrained model.”

According to Lunsford, Tealium architecturally helps deliver real customer outcomes, and it does so in a way that derisks investments in AI models.

“Companies need to track consent all the way through. If you take the data from one customer who then asks to opt-out, and you accidentally train your model with it, you just spent $10 million training your model and that entire model is tainted," he says.

That’s because an AI model is not a database. Think of it instead as set matrix multiplications that get stored in a chip, says Lunsford.

"You can't just extract that one person from a model." he says. "Instead, you would have to build a whole new model based on the way regulation is emerging."

This is one of the key reasons the Tealium boss thinks the CDP model is just getting started, Gartner notwithstanding.

There’s another point he is keen to stress. In addition to playing in the data collection, enrichment and distribution space, Tealium helps with activation.

“You’ve got all these AI models and AI companies that take data in and they give you some output, but that they are not connected to the customer experience layer the way that Tealium is," Lunsford says. “With Tealium, you can activate the output of your model. You can orchestrate putting someone in or out of an audience, or enhance their prediction scores, or enhance a pricing algorithm. We are an AI enabler.”

While Lunsford is aware of some of the work his customers are doing in AI, the vendor lacks visibility over a lot of those efforts.

“We are aware of a couple of our customers using Tealium data for AI, but there's likely many, many more. We don't have visibility. Typically, a lot of Tealium data gets collected by the marketing team and used for marketing campaigns, content management and email. And it also goes into a back-end data warehouse, where [the IT department] then does whatever they're going to do with it," he says.

“There is a 100 per cent guarantee there's a bunch of activity training AI models with Tealium collected data. We don’t have visibility because we are not selling to the IT folks today who are managing those AI models.”

That may be about to change though. Last month, Tealium launched Tealium for AI.

“There are two ways to get to Tealium-derived data. You can let it flow back into your data warehouse, and then you can pull it out of there and all your other silos, and you can do what's called data wrangling. That's a batch process. It's slow and expensive because you have to wrangle the data together, which requires processing and queries. And then you send that to your model. Or you can just plug right into the source, plug into Tealium, and just go direct in real-time, and stream that data into your AI model and bypass all that back-end stuff," Lunsford says.