On our radar: Alembic wants to go beyond MMM to show how campaigns actually drive sales

No time to dig into every new tool on the market? In “On our radar,” Open Garden breaks down the companies worth knowing: what they offer, where they stand and what they still have to prove.

Your TV campaign drove sales. Great. But did it get shoppers into stores, spark Google searches or give your e-commerce site a lift? And how much of the sales bump came from the promotion that ran the same week?

Those are the questions Alembic wants to answer. The platform ties marketing data to business results and factors in what else can move them: pricing, promotions, product availability and market conditions.

Stick with that TV campaign. A standard MMM mainly estimates the incremental sales it drove. Alembic says it also breaks down what happened between the spot airing and the purchase: more branded searches on Google, then site visits and, finally, sales.

Alembic does all of this with aggregate data, without tracking individual consumers. In its methodology overview, the company says it spots changes in that data and tries to work out the chains of cause and effect behind them.

The goal is to help marketing teams see how their spend produces results, then simulate what would happen if they made different calls.

In November 2025, Alembic announced $145 million in funding at a $645 million valuation, according to The Wall Street Journal. Prysm Capital and Accenture led the Series B. The company counts Nvidia, Delta Air Lines and North Sails among its clients.

Tomás Puig runs the company, and his résumé goes well beyond marketing measurement. He started at NASA at 15 and moved into film and advertising. He later became a CMO and helped drive the acquisitions of WP Engine and Emarsys, as well as Cambium Networks' IPO. That career put him in the shoes of the teams Alembic now sells to. Alembic's board includes Laura Maness, CEO of Havas New York, according to CB Insights.

Accenture's role goes beyond writing a check. The firm also struck a strategic partnership to roll Alembic out to its clients, and said at the time that it was testing the platform on its own campaigns. Nvidia is billed as Alembic's first enterprise client and its exclusive supercomputing partner.

Having Accenture in its corner could help Alembic sell to large accounts. Alembic has to win over marketing, but it also needs data from sales, operations and finance teams. A consultancy already embedded with the advertiser can help bring those people to the table.

For all the “causal AI” talk, the business model is fairly standard. Alembic sells an annual license, priced deal by deal depending on the markets, connected data sources and use cases covered. It doesn't publish a list price.

That pricing gives Alembic every incentive to expand gradually within each account. It can come in through marketing measurement, expand to more countries, then help with pricing, promotions or some operational decisions. That opens the door to bigger contracts than a tool limited to media budgets could command.

Deployment means connecting the data sources, configuring the models and testing their output against past decisions where the teams already know the outcome. Alembic says implementation takes a few weeks in most cases. Its public materials don't say whether all of that onboarding is included in the license. Buyers should pin that down when they compare quotes.

Alembic is going after spend that marketing teams often struggle to back up with numbers: sponsorships, events, brand campaigns and organic exposure.

The problem isn't always a lack of data. Teams know how many people saw a campaign, engaged with content or showed up at an event. Things get harder with the CFO when they have to say how much in incremental sales all that drove.

Alembic cites a Fortune 500 airline that wanted to measure the return on its main sports sponsorship. The company says it connected marketing and sales data in real time to estimate the sales the sponsorship generated and let the airline shift spend mid-campaign.

That case leaves a lot unanswered. The public summary doesn't give the amounts shifted, the performance gains or the method used to verify the results. That makes it hard to judge how much the platform actually changed the advertiser's decisions.

The platform can also simulate how a change in price, promotion or budget would affect revenue and margins. That helps when weighing options before spending, but Alembic isn't the only vendor offering that capability. Buyers will need to look at which decisions it can handle and how reliably.

Before we go on. On the morning of Nov. 25, Open Garden is hosting Marketing Effectiveness Innovators, an invitation-only event in Paris that brings together advertisers, agencies and media sales teams around one question: how to measure what media spend actually returns. Two hours of talks and discussion among decision-makers, followed by networking.

It's probably the first question an advertiser with an MMM will ask: why add Alembic if the MMM already says what TV returns?

With Meridian, Google's open-source MMM framework, teams define the channels to measure and the factors to control for. They supply the matching data: spend, sales, promotions, seasonality and so on. The model estimates each channel's contribution and what additional spend could return.

Back to the TV example. An MMM can estimate TV's total effect on sales, including when that effect runs through Google searches. It doesn't need to break out every step to credit that contribution to TV.

Alembic says it makes those steps more visible and does more of the legwork of identifying them automatically. How much did TV lift branded searches? How did that demand then turn into visits and sales?

For a CMO, the payoff is concrete. If part of search's strong performance comes from demand that TV created, cutting TV to move the whole budget to Google could hurt the very results the shift was meant to improve. Putting a number on that link would help justify the call to the leadership team.

Still, the contrast shouldn't be overstated. Some advanced MMMs also model intermediate effects, and Meridian offers a full-funnel approach. Alembic has no monopoly here.

And forget the convenient framing of MMM as correlation and Alembic as causation. Both try to estimate what would have happened without the spend. If your sales and your ad spend both rise around Christmas, you still have to separate the campaign's effect from the seasonal lift. Adding Google searches and site visits to the analysis doesn't solve that on its own.

So the comparison comes down to three things: what Alembic explains that the MMM doesn't, how much work it takes and how solid the results are. One product difference matters: Meridian is a framework that teams have to set up and run, in-house or with a vendor. Alembic sells a proprietary platform that promises to take on more of that work.

The first is convincing teams that the relationships Alembic finds hold up. Branded searches rising after a TV campaign is encouraging. But if a promotion or a news event happened at the same time, how does the model tell their effects apart?

That isn't just a debate for data scientists. A media lead asked to move budget will want to know what the recommendation is based on. If it comes down to trusting a black box, adoption could stall.

Documentation on one of Alembic's uncertainty metrics adds a caveat. The intervals it shows reflect how well the model fits the data, not the full uncertainty around causal effects. In other words, the model can track sales closely without having pinned down what drove them. The caveat is specific to that metric, not the platform as a whole.

The second challenge is data. Separating a campaign's effect from a price cut or a product coming back in stock only works if that information is available at the right level of detail. If a promotion isn't logged properly, no amount of computing power will fill the gap.

Finally, Alembic will have to show what teams do with the extra analysis. Protecting a TV budget, rethinking the media spend behind a sponsorship or fixing an allocation that overweights search: those are the decisions that can justify the purchase. If the output stops at a more detailed map of the customer journey, renewing the license will be a harder sell.

LiftLab is a competitor to watch closely. It combines MMM, budget simulation, daily signals and incrementality testing. Those tests typically involve changing spend in some regions, comparing results against control regions and feeding the findings back into the model.

LiftLab leans on experimentation to check its estimates. Alembic puts more weight on discovering causal relationships through its proprietary models. The sources we reviewed say less about how much Alembic relies on experiments to validate its results.

Measured also combines MMM, incrementality testing and media plan optimization. Fospha, which is more commerce-focused, uses attribution, modeling, forecasting and testing to guide spend.

So should a brand that already runs an MMM sign up for Alembic? There's no obvious reason to if the MMM answers the team's budget questions. A good place to start a pilot is a question the current setup handles poorly, such as what a sponsorship returns or how TV feeds search.

Buyers should then ask the vendor for a concrete test case. It should show what decision the advertiser would have made with Alembic, how it differs from what the MMM recommended and how to verify it would have produced better results. That's how to decide whether an extra license is worth paying for.