Source of Truth is back - with a new focus
Back on air, with max headroom to speculate
Hi all,
It’s been almost two years since I last published here. A lot has changed in that time - at Snowplow, in the industry, and in what I think is actually worth writing about.
Quick housekeeping first: if you signed up years ago and have since lost interest, I won’t be offended if you hit unsubscribe! I’d much prefer to have a smaller list of readers who genuinely want this in their inbox, than a bigger one that doesn’t.
Cyberpunk homage to Wanderer Above the Sea of Fog, courtesy of Midjourney
I’m keeping the name Source of Truth for now, but that won’t stick long term… The original framing has done its job (see below), and the new direction deserves a banner that is a better fit. I want to get a couple of new long-form pieces out first, then I’ll come back and ask you all what the right name should be. More on that in a future post.
For those sticking around, here’s what’s changing...
What this Substack was originally about
When I started this newsletter back in February 2023, the original framing was the “Great Reset” in adtech and martech, driven by data warehouses, first-party data, and the early stirrings of AI.
When the first post came out, there was something in the air - and yet, ChatGPT had only just launched and it was unclear just how fast everything would change:
Things changed at pace, and validated the founding thesis for this blog - certainly faster and arguably more dramatically than I expected:
Composable, warehouse-native architectures have gone from contrarian to mainstream, and are now the default for AI-forward enterprises.
Scott Brinker’s recent “composable canvas” report is, I think, the clearest articulation yet of where this is all going. My reaction piece is now on the Databricks blog:
🔗 Real-Time Decisioning for AI Agents: Why you Need a Customer Context Layer First
I want to come back to this - but first, let’s recap how we got here:
Composability won the war
At Snowplow we have been championing the composability story for over three years - here is an early piece (October 2023) from ourselves, Databricks and Hightouch:
🔗 Building a complete and composable CDP on the Lakehouse
But what our early positioning around composability couldn’t have foreseen was quite how aggressive the adoption curve (and hype cycle!) around GenerativeAI / LLMs would be.
No packaged ‘walled garden’ martech vendor could ever compete with this Cambrian explosion in data and AI tooling. A product squad shipping a “now comes with AI!” feature can’t compete with each brand’s in-house AI engineers being let loose on the company’s own customer data.
Tom Affinito of LiveRamp and I called this early, on this blog:
Fast forward to today, and many of the traditional players in CDP have been acquired (mParticle, ActionIQ), have pivoted (Treasure Data AI) or have made architectural concessions (the ‘zero copy’ narratives of Adobe and Salesforce). And in a kind of Revenge of the Composable Nerds, last month saw Hightouch’s Series D, raising $150m at a $2.7bn valuation - higher than Braze, a public company ($2.4bn market cap).
So, composability won the war. Of course, there are still laggard enterprises to be transformed, and stray vendors to mop up. But if our focus is on where things are headed over the next 18-36 months - then, well, we get to be speculators again…
Max headroom to speculate
With the architecture question largely settled, the interesting frontier is no longer just about where customer data lives. It’s about what AI agents do with that data — in real time, in the moment, on behalf of customers and businesses alike.
This decisioning layer around customer data is suddenly a Wild West. There are new entrants like Aampe, Uniphore and Hightouch, incumbents such as Adobe and Pega who are innovating again - and, most excitingly, there are opportunities for ambitious technology teams to completely re-imagine the end-to-end customer journey.
At the same time, at Snowplow we are witness to a tsunami of disruption coming to the live digital experience for consumers. We started writing about this last year with pieces like Yali’s Is Your Website Ready for AI Agents and Adam’s Agentic Browsing Is Here. Is Your Analytics Stack Ready?
Major digital brands used to be able to just blanket block bots. That’s now untenable, with all manner of Non-Human Traffic (NHT) from LLMs, consumer agents and foundation labs. Much of this NHT is a net positive - from brand and product discovery to content consumption and even transactions. So, NHT needs to be let through onto your site, and the question shifts to what is the optimal experience for a brand to serve these agents and other bots.
After the relative stability of the mobile app era, suddenly the web is exciting again, and every decided question about customer journey and digital experience is re-opened for debate.
Put all this together and almost every aspect of the customer journey is being disrupted, at the same time. That makes it an incredible time to be exploring the future and coming up with new products and ideas.
Where this newsletter is headed
So, where are we headed? Going forward, this newsletter will focus on:
The AI agent–enabled future of customer data — collection, governance, identity, semantic ‘shift left’ coherence at the source
Real-time customer decisioning — what changes when agents, not humans, are making the call
The next generation of customer experience — where agents act for customers, and businesses respond in kind
The architectural & strategic implications for CDPs, CRMs, martech suites, and the data platforms underneath them
To kick things off, as mentioned above I just published a reaction piece on the Databricks blog responding to Scott Brinker’s report:
🔗 Real-Time Decisioning for AI Agents: Why you Need a Customer Context Layer First
The argument: the composable canvas needs a layer Scott’s framing doesn’t fully develop — the real-time customer context layer — and getting it right is the difference between AI agents that learn and AI agents that drift.
Signing off
The piece above is a fair preview of what I’ll be writing about here over the coming months. The cadence will be roughly monthly, and as before I’ll commit to producing no filler, no news updates, no commentary, just long-form original content.
And of course, let me know what you would like me to write about next:
Thanks for sticking around — or for cleanly unsubscribing! Stay tuned for another piece from me soon.
— Alex
Good to read you back!