Cassandra vs Sellforte: two Bayesian marketing mix modeling platforms, compared on grain, scope and pricing
Sellforte and Cassandra both publish a Bayesian marketing mix model with geo experiments. Sellforte states it delivers incremental ROAS at campaign and ad-set level every day, for retail and ecommerce, and publishes prices and a trial. Cassandra, the marketing mix modeling platform, reads at campaign level for budget planning.
On method the two are close. The main differences are the reporting grain, the scope of what each platform covers, and the sectors each one names. Sellforte includes attribution and names retail and ecommerce, and we do neither. This page sets out what each says it is, why we read at campaign level, what a finer grain offers, both vendors' published pricing, where the products differ, and when Sellforte is the better choice. Everything on this page about Sellforte comes from its own published pages.
What Cassandra is
Cassandra is a marketing mix modeling platform. Its production engine is a Bayesian model built on PyMC, so every channel effect comes back as a distribution with its uncertainty stated. Geo experiments are designed and run in the same system, using the same method family as GeoLift, and each result enters the model as a constraint rather than as a separate report.
Reads sit at campaign level, in the planning layer, and nothing changes in an ad account unless a person confirms the exact figure. The platform does not do multi-touch attribution.
It is not built on Google Meridian. Meridian is one of three model engines a project can choose, it is being retired, and it is not offered on new projects.
How this comparison was made
Everything below about Sellforte comes from what Sellforte publishes. We read their homepage, pricing page and the product pages their navigation links to on 14 September 2026, and quoted what we found.
We have not tested their product. Where we describe an absence, we mean we could not find it on the pages we reached.
If something here is out of date, tell us and we will correct it.
When Sellforte is the better choice
Of the ten vendors that passed our screening, Sellforte is the only one publishing both a price and a trial.
You are a retail or ecommerce brand and want one system covering MMM, incrementality and attribution. Their own description is "the Measurement and Optimization OS for Retail and Ecommerce", and they state that they unify all three.
You need a number at ad-set level, daily. They publish that claim. We read at campaign level, for the reasons below. If your workflow depends on daily ad-set readings, their product is built for it and ours is not.
You want to see the price before talking to anyone. Of the twenty-one vendors we checked, seven publish a price, and Sellforte publishes theirs in four tiers.
What each one says it is
Sellforte describes itself as "the Measurement and Optimization OS for Retail and Ecommerce", and states that it delivers "true incremental ROAS at campaign and ad-set level, every day", unifying marketing mix modeling, incrementality testing and attribution. Captured 14 September 2026.
Cassandra, the marketing mix modeling platform, runs a Bayesian model of the whole mix calibrated by geo experiments in the same system. Reads sit at campaign level, in the planning layer.
Both publish a Bayesian claim. Of the ten vendors that passed our screening, three publish that claim, and Sellforte is one of them.
What Sellforte publishes about its method
Across the pages we reached, Sellforte publishes claims covering Bayesian modelling, marketing mix modeling, incrementality, geo experiments, holdouts, multi-touch attribution, causal inference and calibration, and it mentions a trial.
On method family it overlaps with ours on the Bayesian model and the geo experiments, and it adds attribution, which we do not offer. How their experiments connect to their model is not something we could establish from their published pages.
Why we read at campaign level
A marketing mix model reads spend against outcomes over time. What lets it separate one channel from another is variation: spend moved, the outcome moved differently, and the model reads the difference.
That variation is usually present at channel and campaign level over weeks. At ad-set level over a single day there tends to be much less of it. Ad sets are small, their audiences overlap, and their budgets are often adjusted together, so a daily ad-set series carries less separable signal than its number of rows suggests.
Our judgement is that daily ad-set readings go beyond what the evidence in a model typically supports, so we do not produce them and report at campaign level with the uncertainty stated. That is a view about the method. We have not tested Sellforte's model, and they may address this in ways we cannot see from outside.
What a finer grain offers
The case for a finer grain is real. Performance teams adjust spend daily. If the only measurement arrives on a planning cycle at campaign level, it sits outside the loop where many decisions are made.
A daily ad-set reading of the kind Sellforte publishes, where it holds, brings incremental measurement into that loop. That is a genuine product choice, and for teams that trade daily it is an important one.
The practical test is to ask how their model separates two ad sets whose spend moved together all quarter, and to judge the answer. That question applies to any model offering readings at that grain.
A second test is stability. Refit the model with two more weeks of data and compare the ad-set readings with the previous run. Readings that move a lot without any change in spend are a sign the model is fitting noise at that level. The same check is worth running on campaign-level readings, where the movement between refits should be small.
What both of us publish about pricing
Sellforte publishes prices in four tiers. Captured 14 September 2026: Incrementality Testing $1,900 per month, Incremental Attribution $2,500, Full-Funnel MMM $4,500 plus customisations, and Enterprise $8,500 plus customisations, each stated against an average monthly media spend of $300,000, in four currencies. They also mention a trial.
For a reader comparing marketing mix modeling platforms, the comparable tier is Full-Funnel MMM. The Incrementality Testing tier covers experiments on their own.
We publish a price as well. Of the ten vendors that passed our screening, one other publishes a price, and none of the remaining eight mentions a trial on the pages we reached. On pricing transparency, the two are close.
Where the products differ
Reporting grain. Sellforte states incremental ROAS at campaign and ad-set level, daily. We report at campaign level on a planning cycle.
Scope. Sellforte states it unifies MMM, incrementality and attribution. We run a model and geo experiments, and we treat attribution as answering a different question.
Sector. Sellforte's positioning names retail and ecommerce. Ours does not name a sector. A platform built around retail calendars and promotions has an advantage for those businesses.
Where experiments sit. We run geo experiments in the same system as the model, using the same method family as GeoLift, and the result constrains the model. Sellforte publishes geo experiment, holdout and calibration claims.
Access. Both publish a price. Sellforte also mentions a trial.
Uncertainty. Both publish a Bayesian claim, so both can in principle report each estimate with a range. How each presents that range in reports and budget plans is worth checking in a demo, because it decides whether the uncertainty reaches the people making the decision.
Questions to ask both vendors
How does each model separate two channels whose spend moved together? The answer shows what grain the evidence supports.
Show me the priors on a live model, and change one. Both publish a Bayesian claim, so both should be able to show their priors.
Export the output behind this chart. The numbers, in a file you keep.
How do experiment results enter the model? As constraints on the fit, or as a separate report.
Which model version produced last quarter's result, and can you rerun it? A result that cannot be reproduced cannot be checked when it is questioned later.
What we do not do
Reads sit at campaign level. The platform does not produce ad-set or audience-group readings, and it is not a day-to-day optimiser. For daily decisions inside a channel, an ad platform's own tools or an attribution tool are the right instruments.
A channel needs enough history behind it, and enough movement in that spend to read against. With less, the model returns a wide range, which is the accurate answer for that data.
Geo experiments need regional structure, so a single-market business with no way to build a comparison group cannot run one. And the platform does not offer attribution.
Cassandra, the marketing mix modeling platform, is the campaign-level option: a Bayesian model and geo experiments calibrated against each other, read on a planning cycle, with priors your team helps set and output you can export.
A price is published, and so is theirs, in four tiers. Both sides of this comparison disclose what they cost, so the cost question can be settled without talking to either of us, and the grain question is the one that decides it.
What moving between platforms involves
A cleaned, reconciled history of spend and outcomes is what either platform needs, and assembling it is most of the work. It moves with you, and so do experiment results, which remain valid evidence in any model.
Moving from the daily ad-set readings Sellforte publishes to campaign-level planning changes the reporting rhythm more than the method. Teams that relied on daily readings need another source for in-channel optimisation, usually the ad platforms' own tools.
Moving the other way adds a daily reporting layer. Run both on the same period first, so any difference between the two views of the same campaign can be explained.
In either direction, agree in advance which figure decides which kind of decision. Daily readings and a planning model can coexist, as long as each team knows which one governs the budget split and which one governs day-to-day changes inside a channel.
How to choose
Retail or ecommerce, daily trading decisions, one system for MMM, incrementality and attribution: Sellforte.
Budget planning across channels, with each estimate's uncertainty stated and experiments calibrating the model, in any sector: Cassandra, the marketing mix modeling platform.
Unsure which grain you need: ask both vendors how their model separates two ad sets whose spend moved together, and choose the answer you can defend.
Questions
What is the difference between Cassandra and Sellforte for marketing mix modeling?
Both publish a Bayesian model with geo experiments. Sellforte states incremental ROAS at campaign and ad-set level daily, for retail and ecommerce, and includes attribution. Cassandra, a marketing mix modeling platform, reads at campaign level for budget planning and does not offer attribution.
Does Sellforte publish pricing?
Yes, in four tiers, captured 14 September 2026: Incrementality Testing $1,900 per month, Incremental Attribution $2,500, Full-Funnel MMM $4,500 and Enterprise $8,500, the last two plus customisations. They also mention a trial.
Is Sellforte Bayesian?
They publish a Bayesian modelling claim. Of the ten vendors that passed our screening, three do, and Sellforte is one of them.
Can a marketing mix model read at ad-set level daily?
Some vendors publish that claim. Our view is that daily ad-set series usually carry too little separable variation to support it, so we report at campaign level with the uncertainty stated.
What are alternatives to Sellforte?
Vendors publishing both Bayesian and geo experiment claims include Recast and Cassandra, a marketing mix modeling platform. Our list of marketing mix modeling software sets out who each suits.