CRM consolidation is over and the record layer got its winners. The layer that decides what a visitor sees in the next 200 milliseconds is a different discipline — which is why suites keep buying it, renaming it, and losing it.
Some layers of the stack are supposed to consolidate. The system of record is one of them: customer truth wants to be centralized, governed, audited, and boring, and the twenty-year war to own that layer produced its winners. Salesforce won the CRM. The warehouse vendors won the data. Good — a record layer with three competing versions of the truth isn't a record layer.
The experience layer has the opposite physics. It runs per-visitor, in-session, in milliseconds, on the page — deciding what this human sees while they're still deciding whether to stay. It has to change weekly without a change-review board, work on the anonymous majority no profile will ever resolve, and prove its impact fast enough to be trusted. Centralized, governed, and quarterly is exactly what it cannot be. These are different disciplines the way accounting and point-of-sale are different disciplines: connected, and not interchangeable.
The record vendors know this, and the tell is in the M&A. Salesforce didn't build its personalization engine; it bought Evergage in 2020. Mastercard bought Dynamic Yield. Webflow bought Intellimize. The pattern repeats because runtime decisioning is a decision-layer problem, not a data-layer problem — and the muscles that win the record layer (schema design, governance, contract scope) are not the muscles that win the moment on the page.
Watch what happens after the acquisition, though. The engine gets renamed — sometimes twice. It gets re-platformed onto the acquirer's data layer, so its prerequisites grow to include the rest of the suite. And a few years in, the acquirer often builds a successor on its own architecture and sells both, leaving customers to map the product lineage themselves. The suite doesn't absorb the experience layer so much as metabolize it. We've argued the suite era's integration pitch is expiring anyway; the experience layer is where it expires first, because speed is the whole point of the layer.
None of this means the record was a bad investment. It means the two layers should be connected by an interface, not a merger. Keep the system of record central and governed. Let the experience layer be small, fast, and replaceable — an agentic experience layer your own tools can drive. The handshake between them is simple and directional: the experience layer reads what context exists, decides on the page in real time, and writes its outcomes back — the leads it captured, the segments it discovered, the holdout-proven revenue it produced. The record stays the record. It just stops pretending it can also be the reflexes.
And the seam between the layers has stopped being expensive. The old argument for buying both from one vendor was integration; when the layers speak through open tool surfaces and an agent can operate either side, a handshake between two sharp products beats a merger of two mediocre ones. You no longer need to buy the suite to get the seam.
AXO is deliberately the second layer, and only the second layer: the tag scores behavior and decides on the page, in the session, the always-on holdout keeps the receipts, and everything flows back into the record you already keep — leads and pipeline sync into Salesforce, audiences export to your ESP. We don't contest the system of record; we'd rather be the best possible thing to plug into it. If your record layer is settled and your website still greets every visitor identically, the layer that's missing is this one — and adding it doesn't require moving anything you've already won.
A system of record is the centralized, governed store of customer truth — the CRM or customer data platform — optimized for accuracy, auditability, and slow, deliberate change. An experience layer decides what each visitor sees in real time on the website, optimized for per-visitor decisions in milliseconds, coverage of anonymous traffic, and fast iteration with measurable proof. They are complementary disciplines: the experience layer reads context from the record and writes outcomes back into it.
Because real-time personalization is a decision-layer discipline the record vendors repeatedly find they cannot build with data-layer muscles. Salesforce acquired Evergage, Mastercard acquired Dynamic Yield, Webflow acquired Intellimize. Post-acquisition, the engines are typically renamed, re-platformed onto the acquirer’s data stack — growing prerequisites in the process — and sometimes joined by an in-house successor, leaving customers to navigate overlapping products.