Mixpanel added 150% overage fees and removed the AI data opt-in

The change

On June 2, 2026, Mixpanel updated its Supplemental Terms, consolidating what used to be separate addenda into a single product-specific document. There was no announcement and no changelog entry beyond a shifted revision date.

Venpo detected the update the next morning and identified two material shifts buried inside what looks like a routine cleanup: new premium overage fees for Experimentation and Feature Flagging, and a rewritten AI section that changes how customer inputs and outputs may be used.

Mixpanel sits in the analytics layer of a large share of SaaS products, so a quiet change to its billing and AI data rules can affect a lot of teams at once.

What changed

The rewrite folded the old Generative AI Service Addendum and the Mixpanel Agent Beta Service Addendum into one consolidated set of Supplemental Terms, then added new commercial and AI clauses on top.

Three changes stand out:

• premium overage fees for Experimentation and Feature Flagging

• a standing right to use AI inputs and outputs without a separate opt-in

• reduced public detail about how AI data is handled

New 150% overage fees on Experimentation and Feature Flagging

The Supplemental Terms now define an a la carte fee model for two product areas.

For Experimentation, if the number of experiments per Monthly Experiment User exceeds thirty in a calendar month, every additional thirty experiments count as another Monthly Experiment User. If a customer then exceeds twelve times its purchased Monthly Experiment User quantity over a twelve-month period, Mixpanel may charge additional fees at 150% of the per unit price and bill them at the end of the month.

For Feature Flagging, exceeding the purchased quantity of Feature Management API Requests over a twelve-month period triggers the same structure: additional fees at 150% of the per unit price, billed monthly.

In both cases the premium rate and the monthly billing cadence are new. Teams that scale usage of these features can now incur charges above their committed pricing without a separate purchase decision.

AI inputs and outputs can now be used without an opt-in

The previous Generative AI Service Addendum stated that, by default, Mixpanel would not use any input or output to train or improve the AI features, and that improvement use required customers to affirmatively opt in through the Admin console.

The new terms replace that with a Non-Training Improvement clause. Customers now grant Mixpanel a standing right to use inputs and outputs to improve the AI features, provided the data is used only in de-identified form and the usage does not involve training a large language model. There is no separate opt-in step.

Mixpanel still commits not to train or fine-tune its models on customer data except for that customer's benefit, and its AI service providers remain contractually barred from training on inputs or outputs. The shift is in the default: improvement use that previously required an explicit choice now happens unless the structure of the de-identified clause is understood and managed.

Less transparency around AI data handling

The AI provider is no longer named as a single company. The terms now point to whatever providers appear on Mixpanel's subprocessor list, and customers must comply with whichever provider usage policies apply at the time.

In parallel, Mixpanel's Generative AI features page dropped its feature-by-feature table describing each AI tool and the data it sends to outside models, and the related Data Processing Addendum removed the schedule that specifically documented Spark AI handling. The net effect is a more flexible framework for Mixpanel and less concrete, written detail for customers reviewing how AI data flows.

The same rewrite also added an Agents section making customers responsible for supervising any AI features that take actions on their behalf, broadened the output disclaimers to cover legality and third-party infringement, and prohibited certain healthcare uses.

Why this matters

Analytics platforms like Mixpanel are embedded deep in most SaaS products, and their terms quietly govern both spend and data handling.

A consolidation that reads like housekeeping can carry two very different kinds of risk at once: a commercial one, where overage fees create unbudgeted cost as usage scales, and a data one, where the default for AI input and output usage moves from opt-in to standing permission. This is exactly the kind of silent drift covered in our analysis of the hidden risk of vendor legal changes.

Because there was no announcement, a team would only notice either change by reading a long, restructured legal document line by line.

Potential impact for SaaS companies

Companies using Mixpanel for analytics, experimentation, or feature flagging may want to review whether:

• their committed Experimentation and Feature Flagging volumes leave headroom before the 150% overage tiers apply

• finance and procurement are aware that overage can now be billed monthly at a premium rate

• their internal AI governance accounts for inputs and outputs being used in de-identified form by default

• their vendor and data-processing documentation still matches Mixpanel's new, more flexible AI provider framework

Even when individual changes are defensible, the combination of new cost exposure and a changed data default is the kind of thing that belongs in a review rather than discovered on an invoice. Structured vendor contract monitoring is what turns these buried clauses into actionable internal signals.

How Venpo detected it

Venpo continuously monitors vendor legal documents and detects changes as soon as they are published. When Mixpanel updated its Supplemental Terms, Venpo immediately:

• detected the consolidation and the shifted revision date

• separated the substantive changes from the structural reorganization

• flagged the new overage fees and the AI usage default as the items that actually matter

• translated the legal language into clear business impact

Instead of skimming past a document that looks like a formatting cleanup, teams immediately understood which clauses carried real cost and data consequences. Building continuous vendor risk monitoring into the compliance stack is what makes that possible.

Business outcome

Companies that caught this change early were able to:

• check their Experimentation and Feature Flagging usage against purchased limits before hitting an overage

• brief finance on the new premium billing structure ahead of renewal conversations

• update internal AI governance to reflect the de-identified improvement default

• review whether their data-processing records still matched Mixpanel's AI provider framework

Instead of reacting to an unexpected invoice or a compliance gap found during an audit, they had time to adapt on their own schedule.

Key takeaway

The most consequential vendor changes often arrive disguised as cleanup. A consolidated Terms document can introduce premium overage fees and quietly flip an AI data default in the same revision. Venpo helps companies catch these updates the moment they happen, turning quiet legal shifts into clear insights before they become budget surprises or compliance gaps. A deeper look at why scheduled reviews miss this kind of change is explored in our analysis of manual vs automated vendor monitoring.