Alembic

Category: Analytics/Attribution

Causal AI tying marketing spend to actual revenue.

Alembic applies causal inference to marketing measurement, which is a materially different proposition to the correlation-based analytics and media mix modelling most teams rely on. Rather than showing that spend and revenue moved together, it attempts to identify which activity actually caused revenue, including the hard cases: brand campaigns, PR, events and other activity that conventional attribution cannot see because it produces no click. It runs on its own dedicated Nvidia hardware rather than rented cloud compute, which is unusual and reflects how computationally heavy the approach is. The company raised a $145m round in late 2025 co-led by Prysm Capital and Accenture, with Silver Lake Waterman and others participating, and names Delta Air Lines, Mars and NVIDIA as customers.

Our verdict

The problem it addresses is the right one and most teams feel it acutely: last-click attribution systematically undervalues brand, PR and events, so those budgets get cut first in a downturn regardless of what they are actually worth. A causal approach, if it works, is the answer to that, and the backing is serious, with a large round co-led by Accenture and three named blue-chip customers. Running on dedicated hardware rather than rented cloud suggests the method is genuinely compute-heavy rather than a wrapper. Two honest caveats. Causal attribution claims are methodologically contested among people who do this for a living, and they are not verifiable from the outside, so treat the approach as promising rather than proven and ask hard questions in the demo. And this is enterprise only, with no published pricing, no trial, no self-serve route and no public documentation, so it is out of reach for almost all of our readers. All named customers are American, and there is no UK entity or data residency information published.

Best for: Large marketing organisations with substantial brand and non-clickable spend who need to defend that budget with something better than last-click.

Pros

• Addresses the real problem that last-click undervalues brand, PR and events
• Causal approach rather than correlation, which is the right direction
• Large funding round co-led by Accenture, so well resourced
• Delta Air Lines, Mars and NVIDIA named as customers
• Runs on dedicated hardware, suggesting genuine computational depth

Cons

• Causal attribution methods are contested and not verifiable from the outside
• Enterprise only, with no pricing, trial, self-serve route or public documentation
• All named customers are American, with no UK entity or data residency published
• Out of reach for almost all small and mid-size teams
• No MCP server, and documentation is thin generally

Pricing: Pricing on application, demo only (Enterprise)

Rating: 4/5