question.technology · one platform, any question type · accounts opened by hand

Every question you need answered, one platform, one link, one feed.

A survey, a quiz, a poll, an intake form, a vote, an RSVP, an application, an NPS check, a feedback form, a waiver -- question.technology is built around one registry of question types instead of a separate tool for each. Sign up, build a question or a whole set, share a link, and read what comes back. The response engine underneath already keeps every answer free of any identity column and coarsens the stored response time to the hour -- a real, working property today, not a slogan.

Ask us to open an account See how it works →

No self-serve signup route and no billing route are built: apps/api/src/routes/question-technology.ts and question-technology-authoring.ts register no signup, checkout, billing or payment route at all, so no card is charged. Migration 1164_qt_question_type_wall_widening enforces five question types at authoring (open text, a 1-to-5 scale, single-choice, ranking, and conjoint); three of them -- text, scale, and choice -- accept a submitted answer, because validateAnswer (packages/shared/src/question-technology.ts) refuses a ranking or conjoint answer with "ranking/conjoint answers are validated by the submit lane". The rest of the registry below is what the self-serve builder is being built toward.

3 shapes shippedopen text, a 1-to-5 scale, and single-choice -- enforced by a database wall

No identity columna response row has no name, email, IP, account, or session field

Hour-coarsenedthe stored response time is rounded to the hour, never stored precise

Moderated + rate-limitedevery response is checked server-side before it reaches your feed

What the product does, and what is still being built

Five pillars. Two are shipped and already running in production in our own response engine. Three -- the open self-serve signup, the full type registry, and live billing -- are in development. Each is marked honestly.

One authoring surface, aimed at any question type

The pitch is breadth: surveys, quizzes, polls, forms, assessments, intake, voting, RSVPs, applications, NPS, feedback, and waivers, all built on one registry of question types instead of a different tool for each. Today the database wall admits five question types at authoring -- open text, a 1-to-5 scale, single-choice, ranking, and conjoint -- and that is a real constraint, not just a client-side check. Three of those shapes accept a submitted answer today (text, scale, and choice); ranking and conjoint refuse answers until their answer wall lands, so the submit lane does less than the authoring wall allows and we state both numbers rather than average them. The self-serve builder composes those shapes into the formats above; a scored quiz, a wider NPS scale, and a signature capture each need an answer shape the registry does not enforce, and we say so plainly rather than badge them available early.

Five question types enforced at authoring (migration 1164); three accept an answer today

Sign up, create, distribute, analyze

The whole loop lives in one place: ask us to open an account, build a question or a whole set, share a link, and read what comes back in a response feed. Authoring is already open: any signed-in staff or studio account can author questions and question sets today, on the engine running in production. What does not exist is a PUBLIC SELF-SERVE SIGNUP -- there is no form on this page that opens an account, so an account is opened by us, by hand. That signup is what this page is pitching; the authoring surface behind it is not waiting on it.

Authoring open to staff and studio accounts -- no public self-serve signup

A response engine with no identity column, by construction

There is no identity column in the response table at all -- no name, email, IP address, account, browser fingerprint, or session column. Not hidden, not masked, simply never created, so a future bug cannot fill in a column that does not exist. The stored submission time is coarsened to the hour before it is written, so even the moment of a response cannot be lined up against the moment a link went out to work out who answered. This is a real, provable property of the data model, backed by our own response engine, already proven in production, and the self-serve builder is being extended to expose the same option on every creator's own sets.

Shipped -- proven in production today

Moderation and abuse limits run before a response ever reaches your feed

Every incoming response passes a server-side moderation pass for likely foul language or spam before it lands anywhere -- a flagged response is labelled, never silently thrown away, so nothing disappears without a record. A salted, rotating rate limit bounds how many responses one source can send in a time window; the limiter key cannot be reversed back to a sender and is never written to the response table. Both checks are server-authoritative: neither can be skipped by a client, and both are already proven in production in our own response engine today.

Shipped -- server-authoritative, already running

Self-serve signup and billing are not built: no signup form, no checkout, no card charged

There is no public signup form or live checkout on this page. A contact request does not itself create an account or charge a card. No pricing is offered on this page.

Not built -- no signup route and no billing route exist; no card is charged

Sign up, create, distribute, analyze

Four stages, described exactly as they exist today.

Ask us to open an account. That means an email address and a request; no open self-serve signup route is built. Authoring itself is not the thing being waited on -- any signed-in staff or studio account can author on the engine today; what a request is queued for is an ACCOUNT, which we open by hand. No card is asked for and none is charged.
Create a question or a whole set. Build from the shapes that are shipped today -- open text, a 1-to-5 scale, and single-choice -- which already cover a plain survey, a poll, an intake form, an RSVP, or a feedback form. Formats that need a shape the registry does not enforce (a scored quiz, a wider NPS scale, a waiver acknowledgement) are still in development and are labelled as such, not offered early.
Distribute a link. A share link resolves an opaque token to your question set -- the token is used only to look up the set and count opens; it is never written onto a response row, so the link and a specific answer cannot be joined back together from that token alone.
Analyze the responses. Read a response feed of what has already passed the server-side moderation and rate-limit checks -- counts by question, and by set, with no unreviewed noise. The feed shows what actually came in; it never fabricates a placeholder response to fill the screen.

The north star

One registry of question types -- surveys, quizzes, polls, forms, and everything between

This is the breadth the self-serve builder is aimed at: twelve formats built from one shared registry of answer shapes instead of twelve separate products. Today the database wall (migration 1164) enforces five question types at authoring -- open text, a 1-to-5 scale, single-choice, ranking, and conjoint -- and three of them, text, scale and choice, accept a submitted answer. A plain survey, a poll, an intake form, a vote, an RSVP, and a feedback form already compose cleanly from those three. A scored quiz or assessment, a wider NPS-style scale, a signature-style waiver, and an attachment-carrying application each need an answer shape the registry does not enforce -- said plainly below, format by format, rather than badged available before its shape is built.

Surveys

A sequence of scale and choice questions -- works with today's shapes.

Quizzes

Choice questions with a scored right answer -- needs a scoring shape the registry does not enforce.

Polls

One choice question, shared widely -- works with today's shapes.

Forms

Open text and choice fields composed into a general-purpose form -- works with today's shapes.

Assessments

Scored, multi-question evaluation -- needs the same scoring shape as quizzes.

Intake

A structured first-contact form for a new client or case -- works with today's shapes.

Voting

A single choice question with one response per link -- works with today's shapes.

RSVPs

A short choice-plus-text response to an invitation -- works with today's shapes.

Applications

A longer structured form, often with attachments -- the attachment shape is absent from the registry.

NPS

A single 0-10 style scale question -- needs a wider scale than today's 1-5 range.

Feedback

Open text plus an optional scale -- works with today's shapes.

Waivers

A statement plus an acknowledgement capture -- needs a signature/acknowledgement shape the registry does not have.

The differentiator

A response that carries no identity, by construction -- not by policy

The table a response is written to has no column for a name, an email address, an IP address, an account, a browser fingerprint, or a session. It was never built into the table, so there is no field for a future bug to accidentally fill in. This is a property of the schema, not a promise layered on top of one that could hold identity if something went wrong.

The stored response time is rounded to the hour before it is written. A precise timestamp on a low-volume set can be lined up against the moment a link went out and work out who most likely answered; the coarse, hour-level value cannot do that same correlation. Every response also passes a server-side moderation pass for likely foul language or spam and a rotating rate limit before it reaches your feed -- a flagged response is labelled, never silently thrown away, so there is always a record of what came in.

These are properties of the response-collection path specifically -- this is not a claim that every corner of the product carries no identity. Your own account, your billing details (once billing is live), and your dashboard obviously do identify you as the creator. It is the person answering your question who can stay free of any identity requirement, and that option is already proven in production in our own response engine today; the self-serve builder is being extended to expose the same option on every creator's own sets.

Ask us to open an account Read the FAQ →

Self-serve signup and billing are not built

There is no public signup form or live checkout on this page. A contact request does not itself create an account or charge a card. No pricing is offered here. This is a for-profit product; nothing here is a nonprofit or tax-deductible arrangement.

Opening an account

Ask us to open an account -- no card, no commitment

We show the current state honestly: which of the twelve question-type formats already compose from the answer shapes that accept answers today (text, scale, and choice), how the identity-free response option works today, and what the self-serve authoring surface looks like -- authoring is open to any signed-in staff or studio account; it is the public SIGNUP that is not built. There is no pricing commitment and no signup form to fill out on this page.

To ask: email [email protected].

FAQ

Common questions

Is this the same thing as the question.technology I already answered a few questions on?

It is built on the same response engine, but it is a different surface for a different audience. If you clicked a link and answered a few questions with no account and no personal information, that is the respondent side -- it stays exactly as it is. This page is the self-serve product for the person who wants to BUILD a question or a set, share it, and read the responses.

What question types actually work today?

The database wall enforces five question types at authoring: open text, a 1-to-5 scale, single-choice, ranking, and conjoint. Three of them accept a submitted answer today -- text, scale, and choice -- while ranking and conjoint refuse answers until their answer wall lands. A survey, a poll, an intake form, an RSVP, and a simple feedback form all compose cleanly from those three answering shapes today. A scored quiz or assessment, a wider NPS-style scale, a signature-style waiver acknowledgement, and an attachment-carrying application each need an answer shape the engine does not validate -- those are the ones still in development, and we label them that way rather than claim them early.

Is a response really free of any personal information?

The response table has no column for a name, an email, an IP address, an account, a browser fingerprint, or a session -- it was never built, so nothing can leak into a column that does not exist. The stored response time is rounded to the hour before it is written, so the exact moment of a response cannot be matched against the moment a link was sent. This is a property of the data model, proven in production in our own response engine today, and it is the option the self-serve builder is being extended to expose on your own sets.

What happens to a response that gets flagged?

It is labelled, not deleted. A server-side pass checks every incoming response for likely foul language or spam before it reaches any feed; a flagged response is marked as flagged so there is a record of what came in and why it was set aside, rather than being silently discarded.

Can I sign up today? Will I be charged?

There is no public signup form or live checkout on this page. A contact request does not itself create an account or charge a card. Authoring access for an existing signed-in staff or studio account is a separate capability; the contact request grants no such access.

Why build one platform instead of separate tools for surveys, quizzes, forms, and the rest?

Because the underlying question is the same shape problem every time: ask something, collect an answer in a known shape, and read back what came in. A single registry of answer shapes -- text, scale and choice, which accept answers, plus ranking and conjoint, which the wall stores but the submit lane refuses -- lets one authoring surface, one share-link engine, and one response feed serve supported survey answer shapes, instead of maintaining a separate product per format.

What is built and what is honest-off

Built and running today: five question types enforced by a database wall at authoring (open text, a 1-to-5 scale, single-choice, ranking, and conjoint), three of which -- text, scale, and choice -- accept a submitted answer today, with ranking and conjoint refusing answers until their answer wall lands; a response table with no identity column at all; hour-coarsened stored response times; server-side moderation and a rotating rate limit on every incoming response, with a flagged response labelled rather than discarded; and an opaque share-link token that resolves a set without being written onto any response row. Honest-off: a public self-serve signup (an account is opened by us, by hand) and live billing. Authoring is NOT honest-off -- any signed-in staff or studio account can author on the engine today. No card is charged today. No adoption numbers, named customers, or customer quotes appear on this page — it is a new product and there is nothing to cite yet.

Ask us to open an account