“We're moving from seats to usage. How do we not break the book?”
A value metric that tracks what you deliver, and a migration that protects the base.
A founder or CFO at an AI-native company, or anyone whose cost to serve now moves with usage.
Seat pricing stopped describing the product. An agent does work rather than occupying a chair, and cost to serve moves under you per customer. The model has to change — but the existing book is priced on seats, and a migration done badly costs more than the old model did.
The evidence, before the recommendation.
Nothing here is a market average or a rule of thumb. Every number is computed from your own data or carries the source it came from.
The right value metric
Which unit actually tracks the value the customer receives, tested against your own usage and billing data rather than chosen by analogy.
What customers will accept
Willingness-to-pay for the new metric by segment, and the price points at which the switch reads as fair rather than as an increase.
What it does to every account
The new model projected onto your live subscriptions, customer by customer: who pays more, who pays less, who leaves.
The hybrid, if there is one
Most migrations land on a platform fee plus usage, not pure usage. The shape falls out of the book, not out of a preference.
What you get, and what keeps going.
One engagement produces something you decide on. What happens next — and on what rhythm — is the section below.
The work product.
- The value metric, and the evidence it was chosen on
- The new architecture — platform fee, included usage, overage, commits
- A per-customer current-vs-new table and the net ARR consequence
- A cohort rollout: immediate, at renewal, grandfathered, or new customers only
The metric, the price points, and which cohorts move when.
Nothing reaches your billing system until you approve it.
A migration is not done when the pricing page changes.
The risk in a model change is spread across the year that follows it, in cohorts that move at different speeds and a metric that may quietly stop tracking value.
These read your own data, so they are current the moment a sync lands — no waiting for a reporting cycle to find out what you charged last week.
- Value-metric fit — does the candidate metric track value?lead
Whether the unit you want to charge for moves with what the customer actually gets
- Usage audit — the distribution behind the metric
Who the new model makes expensive, before they discover it themselves
- Cohort retention and tier migration
How each cohort behaves through the move, instead of one blended number that hides the damage
- Revenue simulation on the live book
The net consequence, customer by customer
A tracked set of six to eight competitors changes something every two or three weeks. Monthly catches each move within weeks; daily would report nothing almost every day.
- Pricing-model comparison — who charges per seat, per unit, hybridlead
Whether the category has already moved, and what your buyer now treats as normal
- Total cost at a reference buyer
What a comparable customer pays across the set once model differences are removed
Demand moves with the category, not the week. Simulated respondents are what make a quarterly re-measure affordable at all — a conventional study is priced to be run once.
- Choice-based conjoint — metric and level traded off togetherlead
Which metric buyers accept and at what level, tested together rather than one at a time
- Gabor-Granger on the new metric
The price per unit the new model actually clears at
A value model is wrong the day you ship something that changes the offer, not on a date. Quarterly is the floor; any release that moves what the buyer gets pulls it forward.
- Economic value expressed in the new metric
That the new unit still carries a price you can defend
- Next-best alternative = the deal they already have
What you are really competing with in a migration, which is your own old contract
Your own book is live because it is your data — the analyses recompute as soon as a sync lands. The rest are cadences a person keeps, not alerts a dashboard fires: your pricing expert does the refresh and brings you the read. Each one either confirms the price you are charging or turns into a proposal, and nothing changes until you approve it.
Where this question usually comes from.
Judge the method before you buy it.
We run this analysis on public pricing pages and publish what we find — outside-in, every claim bounded. It is the closest thing to watching us work before you hire us.
All teardowns →An $8 AI add-on worth about a dollar a seat — withdrawn, and replaced with a $20 tier boundary worth an estimated $120–165M.
Read the Notion teardownBefore you book a call.
What do you need from us to start?
Read-only access to your billing system, or an export of customers, subscriptions and invoices. Competitive and demand evidence we bring ourselves.
Who does the work?
A senior pricing operator, with Albi's agents doing the continuous groundwork on your data. You review a proposal — you are not being handed software to learn.
What if we disagree with the recommendation?
Then it does not ship. Every number arrives with the evidence behind it, so you can argue with the reasoning rather than the conclusion.
Does this end, or continue?
Either. A scoped engagement ends at a proposal you approve or reject. A standing mandate keeps the willingness-to-pay, value and competitive evidence current, and brings you a move when one is warranted.
The Albi workspace
Your engagement runs in a workspace you log into. Your expert works there on your data; you review the evidence, read the proposal, and approve the move. Approved pricing provisions into Stripe from the same place.
- One workspace per pricing initiative, with the evidence behind every number
- The proposal, the working underneath it, and the approval in one surface
- Nothing reaches your billing system until you approve it
Next: the answer where the question gets asked
Not built yetNobody asks a pricing question in a pricing tool. They ask it in the thread where a deal is being argued, on the record where the opportunity lives, or through an agent already halfway through a task. We are building toward Albi answering in those places, from the same evidence — and none of it changes who approves a price.
- In the deal threadChat, wherever your team argues deals — Slack, Teams, or whatever you actually use.
- In your CRMThe guidance on the opportunity record, where the rep already is.
- To your own agentsAn MCP server, so agents you already run can read your pricing evidence directly.
- Through the APIYour own surfaces, your own workflow, the same evidence behind it.
Approval does not move. However the question reaches us, the answer comes back as a proposal a person signs off.
Not the one you're asking?
"We're moving from seats to usage. How do we not break the book?"
Bring us the question. A scoping call gets you the shape of the engagement, what your data makes possible, and a number.