The trouble with a big sponsorship is that everybody can see it and almost nobody can price it. A logo appears during the Olympics. Search traffic rises. Tickets sell. Which of those sales did the logo cause? For years, the usual answer came in a parade of impressions, surveys and carefully worded confidence. Delta Air Lines wanted something closer to a receipt. During its Paris 2024 Olympics sponsorship, it ran a pilot with Alembic, a San Francisco company that makes software for tracing business outcomes back to the things that might have caused them.
- Alembic joins marketing and sales signals to model incremental impact.
- Delta used it to examine an Olympics sponsorship and ticket sales.
- North Sails changed its social content after Alembic's analysis.
- It sells custom annual enterprise licenses; pricing depends on scope.
This is an expensive question to answer well. It asks what would have happened without the sponsorship, the campaign or the viral moment. A marketing dashboard can show two rising lines. It cannot, by itself, tell a finance chief whether one line pulled the other up. Alembic's commercial proposition is that its causal models can. The company takes signals from media, social channels, site traffic, customer systems and sales, then looks for relationships that survive a more demanding test than simple coincidence.
The bill for applause
Delta's pilot is a tidy illustration because a major sports sponsorship lives far from the last-click world. A viewer may see a broadcast placement, search for a flight days later and book through a different channel. According to Alembic's account of the work, its system connected brand activations and exposure during the Paris Games to ticket sales. At a 2025 marketing panel, then-Delta CMO Alicia Tillman described the exercise as a way to measure sponsorship effectiveness. Alembic says the analysis helped Delta defend continued sponsorship investment through 2028. That is a reported customer result, rather than a public release of the underlying model or ticket-level calculations.
The point is practical: if marketing can make a credible case for incremental revenue, a CFO can discuss it as an investment. If it cannot, a beautiful campaign still arrives at budget season dressed as an expense. Alembic founder and CEO Tomás Puig put the complaint neatly in Accenture's 2025 partnership announcement: “Most companies are not short on data. They are short on answers.”
Puig founded Alembic in 2018 with John Adams, its chief information officer, and Seth Little, its chief commercial officer. The company's name is borrowed from the vessel alchemists used to distill a mixture. The metaphor is a little theatrical for a software firm, but fitting: Alembic says it separates the drivers of an outcome from the surrounding noise. Its early focus was marketing measurement. Its current pitch reaches into pricing, sales, supply and operations as well.
A sailboat is a better test than a banner ad
North Sails, the sailing equipment brand, supplies a smaller and more visible example of a decision changing. Its social feed leaned toward polished catalog-style content. Alembic's analysis pointed the company toward more authentic posts and a younger audience. In Alembic's published case study, average post impressions went from 24,000 to 56,000, and the audience age moved down by 10 to 15 years. The company attributes the shift to the new content strategy. The figures are useful as a reported before-and-after, while leaving room for the usual questions about seasonality, creative changes and other forces.
That example makes the product less mystical. A company can use Alembic to find a change worth making, make it, and watch whether the outcome improves. Another public case study describes a Fortune 500 technology company's fragmented Salesforce data being rebuilt as a time series. Alembic reports that the customer expanded its sales pipeline by 37%. In both cases, the hard labor starts before a model gives an answer: data from different systems has to be cleaned, timed and connected to an outcome that matters.
A very costly way to ask why
Alembic's technical answer uses causal graphs, time-series reconstruction and a proprietary spiking neural network. Its platform page describes data activation, observability, causality, prediction and an “Intelligence Brief” that turns model output into an executive report. The company says it processes roughly 500 million data points from 2.5 million sources in a day. That figure is a measure of claimed throughput, not a guarantee that every recommendation will be right. A causal estimate is only as good as the data, assumptions and validation behind it.
The machinery is unusually literal. Alembic operates private NVIDIA-based supercomputing infrastructure and says NVIDIA was its founding enterprise customer and exclusive supercomputing partner. In November 2025, Prysm Capital and Accenture led a $145 million Series B at a reported $645 million post-money valuation. Accenture also entered a strategic partnership to offer Alembic's measurement capabilities to clients; its marketing arm, Accenture Song, said it was piloting the system on its own campaigns. The round followed a $14 million Series A in 2024.
A buyer does not purchase a $145 million supercomputer. Alembic's FAQ describes a custom annual enterprise license, with price shaped by the number of markets, data sources and use cases. There is no useful single sticker price for the kind of integration Delta needed. The true project cost also includes the unglamorous work of agreeing on revenue definitions, getting CRM and media histories into usable shape, and deciding what evidence would be strong enough to move a budget.
“Most companies are not short on data. They are short on answers.”Tomás Puig, founder and CEO
From campaigns to company decisions
Alembic's March 2026 release, version 3.0, pushed its ambition beyond marketing. It introduced continuously recomputed causal models and scenario simulations meant to show how a change in pricing, promotion, supplier or budget might affect revenue and margin. That puts Alembic in a broader contest with marketing mix models, multi-touch attribution, brand trackers, consulting studies and internal analytics teams. Its claimed difference is a living model that spans more variables and returns a decision while the decision is still available to make.
The expansion raises the standard of proof. It is one thing to say which posts appeared near a lift in impressions; it is another to say what a pricing change will do to margin amid supply constraints and shifting demand. Alembic's own technical paper discusses cross-validation and checking its models against other statistical methods. Those are the right questions for a customer to bring to a pilot. A team with thin historical data, little variation in spending, or no agreed outcome will have a harder time separating cause from a persuasive story.
A sensible way to borrow the method begins without buying the platform: pick a consequential decision, specify the outcome, gather the relevant history, and write down what would count as a counterfactual. Run a contained pilot alongside the existing measurement system. Ask which recommendation changed, what it cost to act on it, and whether the result held after the next cycle of data arrived. The glamorous part is a causal graph. The valuable part is discovering that a budget line deserves to be bigger, smaller, or gone.
That is why Alembic's story is interesting beyond marketing technology. Modern companies can count almost anything and still struggle to explain a choice. The Olympic logo and the airline ticket are easy to see. The missing link between them is where a great deal of money has been spent, defended and sometimes wasted. Alembic has built a business around making that link visible. The spreadsheet, at last, gets to ask the sponsorship a rude question.