Skip to content

Data Privacy and Security ​

When collecting data—especially from vulnerable populations or through surveys focusing on accessibility—protecting the privacy and security of respondents is paramount.

This document explains the conceptual framework and architectural decisions Accessible Surveys uses to ensure data is handled ethically, securely, and in compliance with modern privacy standards.

The Philosophy of Minimal Data Collection ​

Our core philosophy regarding privacy is simple: We only collect what is strictly necessary.

By default, Accessible Surveys does not collect PII (Personally Identifiable Information) unless you explicitly design your form to ask for it. We do not track respondent GPS coordinates, fine-grained location, browser fingerprints, or persistent advertising trackers. At the network edge, the platform infers coarse country-level location for regional aggregate reporting (via standard CDN headers), but raw IP addresses and granular location data are never written into response records.

Operational email addresses are the exception worth naming. A survey can collect a respondent's address solely to send emails — such as a submission confirmation, unsubmitted-form reminders, or follow-ups for survey results or later studies — without adding that address to the dataset. The address serves the survey's emails only: it is stored apart from the answers, excluded from every export unless the survey's Dataset email purpose makes it an answer, and never used to identify a respondent or link them to their answers (see Email Purposes).

When the editor opts in to invitations, a team audience's addresses serve the invitation, its reminders and — while the survey is invitation-enabled — the results and follow-up emails as well. Those addresses stay in the invitation records, where the editor sees a masked form only; they never enter a response, a dataset or an export.

A consent is a record with a lifetime: the address, the session, the notice version agreed to, the purposes it covers and the sends, held for 180 days by default (1–365, editor-adjustable) and deleted when it ends. The Email page's Sent history displays aggregate counts for each send event (total, delivered, opened, failed); individual masked recipient addresses are not exposed there, but can be monitored during active invitation runs in Analytics → Progress. A follow-up is sent by hand from the Email page, to the respondents who opted in for that purpose and, when invitations are on, to the invited audience that has not unsubscribed. The survey's owner can purge the collected data at any time, which removes the consent records with the responses; the survey itself survives.

Account Types and Anonymity ​

The platform supports different levels of respondent authentication, allowing you to choose the right balance between data validation and respondent privacy.

  1. Anonymous Surveys: The default mode. Anyone with the link can respond. The system generates a randomized session ID to handle the technical submission process, but this ID cannot be tied back to a specific individual. This is ideal for sensitive feedback where anonymity guarantees honest answers.
  2. Verified Accounts: For situations where you need to track participation, prevent duplicate submissions, or conduct longitudinal tracking, you can require respondents to log in. In this scenario, responses are linked to an authenticated account.

Data Isolation ​

When a respondent logs in (Verified Account mode), the architecture maintains a strict boundary between their identity data (email, name) and their survey responses. Researchers must explicitly choose to export identifying information; otherwise, the dataset remains fundamentally anonymized, referencing only abstract User IDs.

Anonymized Exports ​

Marking a question — or a whole section — as anonymous in the builder decides what the anonymized export contains. The decisive detail is when that decision is applied: anonymization is baked into the stored response at submission time, using the build the respondent actually answered. Downloading the anonymized dataset does not re-read your form; it serializes the responses that were anonymized when they were submitted.

Plan around three consequences:

  • Mark fields anonymous before collecting responses. Marking a field anonymous later only affects responses submitted under a build that includes the change; responses collected earlier keep their values in the anonymized dataset.
  • Changing the setting does not rewrite history. Removing values already collected requires a backfill of the stored anonymized dataset.
  • Anonymous fields still appear in the export. The column exists for every question, but its cells are empty for responses that were anonymized.

To mark a question or section anonymous, open it with Advanced Mode enabled and switch on Anonymous under Anonymous Settings. See Exporting Survey Data for the download steps.

Invitation Records ​

Sending a survey to a list changes what the organizer can see, so it is worth stating exactly what an invitation records.

A team's audience is a reusable address book: one recipient per person per team, with the address, an optional name, language and external id. The team is the controller of that list and confirms a lawful basis and permission for it before importing (the import warranty). The platform processes the list only to send the survey and its reminders; it never uses audience addresses for its own marketing or profiling.

An invitation records status, not answers. For each recipient it stores a masked address, a language, status timestamps (sent, delivered, opened, started, submitted, bounced, unsubscribed, complained, failed) and reminder counts. The full address and the unique link live in a separate server-only record. No invitation id, token, address, audience id or actor id is written into a response document, an export or a dataset: the organizer sees that someone submitted, never which answers are theirs. The Progress table and its status CSV show masked addresses only.

An unsubscribe, a complaint or a hard bounce adds the address's keyed hash to a customer-wide suppression list, so a later survey cannot re-contact someone who opted out. The list holds no plaintext address, survives the survey's or audience's deletion, and only a superadmin can remove an entry (with an audit record). Invitations are survey data: purging a survey deletes its runs, invitations and links, while the audience and the suppression list remain until the team or customer removes them.

Security Architecture ​

Accessible Surveys is built on enterprise-grade infrastructure designed to protect data at rest and in transit.

  • Data at Rest: All survey configurations, logic rules, and respondent data are encrypted while stored in our databases.
  • Data in Transit: All communications between the respondent's browser and our servers are encrypted using TLS/SSL, ensuring that data cannot be intercepted during submission.
  • Role-Based Access Control (RBAC): Access to survey data is strictly governed by user roles. Only team members with Owner or Admin roles can purge data or manage collaborators. Editors can build questions, configure settings, and deploy updates, while Guests are limited to read-only review, and Translators hold permissions restricted to localized text. All members of the active team share read access to team surveys.

The "Why" Behind the Design ​

In the context of the disability movement and inclusive design, trust is critical. Respondents need to know that their data is safe, especially when answering questions about their habits, experiences, or medical needs.

By building privacy controls directly into the foundational architecture—such as offering truly anonymous sessions and decoupling identity from response data—we empower researchers to collect high-quality data ethically, while respecting the boundaries and safety of their respondents.