whatismmm

Open-source vs managed MMM

Open-source MMM is free the way a free puppy is free. The license costs nothing; the data scientist who runs it costs $175,852 a year fully loaded on BLS numbers. Managed vendors that publish pricing span $35,880 to $220,000+ per year. The real decision is where you want the model's assumptions to live.

By Oliver Wakefield-Smith · Updated 2 August 2026 · Every number on this page resolves to a primary source in the ledger below.

Pricing the free option honestly

Meridian, Robyn, and PyMC-Marketing all carry a $0 license.34 The operating cost is a person: the US median data-scientist salary is $120,2301 (BLS OEWS, May 2025), and employer compensation data puts total cost at about 1.46x2 wages once benefits load in, giving roughly $175,852 fully loaded. A first build realistically consumes most of an FTE for a year; thereafter perhaps half, split with other work. Add compute (small) and opportunity cost (not small; that person is not doing other analytics).

Pricing the managed option honestly

Published pricing exists for exactly three of the twelve managed vendors in our registry: Sellforte from $2,990/mo5 ($35,880/yr), Cassandra from EUR 2,700/mo6, and Paramark at $100k to $220k+/yr7, all verified 2 August 2026. The other nine quote. The published spread is the honest bracket this site can defend; your quote will land where your spend, channel count, and negotiating leverage put it.

Where each path breaks

In-house breaks at the person: one data scientist is a bus factor of one, and when they leave, the model's judgment leaves in their head. It also breaks quietly at validation; without external pressure, in-house models are rarely stress-tested until a board meeting does it involuntarily. Managed breaks at the black box: assumptions you cannot inspect, a vendor marking its own homework unless you force experiment calibration, and data-egress friction at offboarding, which is why data ownership is question 10 of the vendor questions.

Verdict by profile, and the hybrids

Data scientist on staff and geo-level data: build; the calculator will show the math favoring you. Neither: buy, or take the middle path of self-serve SaaS, which prices below managed programs and above zero. Fifteen-plus million in spend with a contested mix: consider the audit hybrid, a managed vendor plus one in-house analyst who re-derives the big numbers in open source; expensive, and cheaper than a wrong $3M reallocation. Run your own inputs on the build-vs-buy calculator.

Source ledger

Every numeric claim on this page resolves to one of these primary sources.

  1. [1]BLS OEWS, Data Scientists (SOC 15-2051), May 2025 national estimateshttps://www.bls.gov/oes/current/oes152051.htmretrieved 2026-08-02
  2. [2]BLS Employer Costs for Employee Compensation, March 2026https://www.bls.gov/news.release/ecec.nr0.htmretrieved 2026-08-02
  3. [3]Google Meridian documentationhttps://developers.google.com/meridianretrieved 2026-08-02
  4. [4]Meta Robyn repository (GitHub)https://github.com/facebookexperimental/Robynretrieved 2026-08-02
  5. [5]Sellforte pricing pagehttps://sellforte.com/pricingretrieved 2026-08-02
  6. [6]Cassandra (cassandra.app) pricing pagehttps://cassandra.app/pricingretrieved 2026-08-02
  7. [7]Paramark pricing pagehttps://paramark.com/pricingretrieved 2026-08-02