# Wants

URL: /code/builders/startup/wants · Id: `page:code/builders/startup/wants`

> Unmet demand sensed across every Startup becomes graded Evidence on a Problem, and a refundable deposit proves it before anything is built.

When building costs almost nothing, the scarce thing is demand. [StartupBuilder](/code/builders/startup) records every unmet want as graded [Evidence](/business/nouns/evidence) on a [Problem](/business/nouns/problem), and asks for a refundable deposit before anything is built.

## Where wants come from
A want is never typed in. It is written when someone acts on a need that no [Offer](/business/nouns/offer) meets, and each source writes Evidence at the grade its act deserves.

| Where | What happened | Grade |
| --- | --- | --- |
| [Agent](/business/nouns/agent) request | a buying agent asks for an Offer and nothing matches | behavior |
| Empty search | a [Visitor](/business/nouns/visitor) searches a [Site](/business/nouns/site) and finds nothing | behavior |
| Uncovered request | a [Customer](/business/nouns/customer) asks support for something no Offer covers | behavior |
| Refused [Order](/business/nouns/order) | an Order fails because no Offer can fill it | behavior |
| [Waitlist](/business/nouns/waitlist) join | a Visitor joins the line for an Offer that is not open | behavior |
| [Interview](/business/nouns/interview) answer | an [ICP](/business/nouns/icp) member describes the last time the Problem cost them | behavior |
| Deposit | a buyer pays a refundable deposit on a reserve Offer | commitment |
| Stated wish | someone says they would use it | opinion |

Opinion stays on the record and weighs zero. Frequency, urgency, and willingness to pay are computed from the Evidence, never entered as fields.

## The want ledger
The want ledger is one read across every [Startup](/business/nouns/startup) in the [Portfolio](/business/nouns/portfolio). It ranks Problems by the outside Evidence behind them and shows the distinct ICP members who asked. It is a [Dashboard](/business/nouns/dashboard) over the kept record, not a separate store.

A Problem that 40 people hit this month, with three deposits, outranks one with 400 opinions. The [Studio](/business/nouns/studio) drafts new Startups from the top of the ledger, so their first buyers exist before their first [Page](/business/nouns/page).

When such a Startup goes Live, everyone who asked and agreed to hear back is told first. A Dormant Startup stays subscribed to its Problem and wakes when commitment-grade Evidence returns.

One Startup's Customers never become another Startup's prospects by default. Wants cross between Startups as counts per Problem, and a person's identity crosses only with their [Consent](/business/nouns/consent).

## Reserve-first
Reserve-first is a new Startup's first test: a reserve page, published on day one, that takes a refundable deposit on the real Offer. The page states the Problem in the ICP's own words, the approach, the [Price](/business/nouns/price), and a deliver-by date.

A deposit is an authorization against that Price. If the Startup misses its date, every deposit refunds automatically, so the buyer risks nothing but attention.

Deposits are commitment-grade Evidence, the strongest grade a Startup can collect before it has a [Product](/business/nouns/product). Saying yes is free; paying is not.

A meal-planning app for people with celiac disease posts its reserve page at $9 a month, with a $9 deposit and a 30-day date. A payroll tax-table [Dataset API](/products/templates/dataset-api) for developers posts at $400 a month with a $40 deposit. Within two weeks each knows whether anyone pays.

## What a deposit proves
A deposit counts only from an outside buyer: one who shares no owner, Studio account, [Agreement](/business/nouns/agreement), or sibling chain with the Startup. Purchases between sibling Startups are real revenue, reported apart, and never pass a gate.

Commitment-grade Evidence from outside ICP members moves a Startup from [Idea](/business/nouns/idea) to [Building](/business/lifecycle/states/building). The Building state's one [Metric](/business/nouns/metric) that matters is [Committed ICP members](/agents/founding#committed-icp-members).

Simulated buyers can prune variants before real traffic, and their results are graded opinion. A simulated conversion is a claim about buyers, not a buyer, so it never decides a [Lifecycle](/business/nouns/lifecycle) transition.

## Wants across the Studio
Sibling Startups race as variants of one Idea. Each takes a different ICP or approach, opens its own reserve page, and is judged on deposits. The losers go Dormant and keep their Evidence.

An audience is a source of wants too. A creator's comments, questions, and complaints read as Evidence on Problems. The Studio drafts two or three Startups for that audience, each with a reserve page, and offers the creator equity for distribution.

Every Startup writes to the same record, so each new Startup starts with the demand that every earlier one sensed.

## Next
- [Test demand](/workflows/processes/test-demand): the cheapest test that settles each [Hypothesis](/business/nouns/hypothesis).
- [Problem interview](/workflows/processes/problem-interview): Evidence from the people who have the Problem.
- [Experimentation](/code/builders/startup/experimentation): how every change becomes a test.