Skip to content

Data processing agreement

Last updated: November 17, 2026

This agreement is concluded under Article 28 of Regulation (EU) 2016/679 (the GDPR) between you, the controller, and NCT HOLDING, a SASU with a share capital of €1,000, 57 avenue du Maréchal Juin, 64200 Biarritz, France, registered under number 131 634 651 (RCS Bayonne), the processor.

It applies whenever you use the service to process personal data about people other than yourself: your customers in Stripe, the people your product surfaces in PostHog, the contacts in your CRM, the people who fill in your capture forms, and the recipients of the campaigns, sequences and transactional emails that Canopus sends for you. For your own account data, on the other hand, we are the controller, and the privacy policy applies.

It forms an integral part of the terms of service and takes effect when you accept them; the version, date and IP address of that acceptance are kept. A signed version can be obtained on request from contact@canopushq.io. Where the two texts conflict on the processing of personal data, this agreement prevails.

1. Subject matter and duration

Subject matter. Our processing of personal data on your behalf, in order to provide the service described in the terms of service: reading the tools you connect, maintaining your CRM, producing analysis, and sending from your domains the campaigns and sequences you approve and the transactional emails your product entrusts to us.

Duration. From your acceptance of the terms of service until the later of the following two dates: the closure of your account, or the deletion or return of the data under section 10.

2. Nature and purpose of the processing

We process personal data only to provide the service, and in particular to:

  • read and store data from the tools you connect, so we can show you your own numbers;
  • maintain your contact records, including the fields Stripe supplies, imported contacts and those received from your capture forms;
  • compute aggregates over your customer base and submit those aggregates to a model, so an agent can produce a brief, an ICP, an audit or a campaign draft;
  • send from your domains, through Amazon SES, the campaigns and sequences you have approved and the transactional emails your product sends through the API, and record the send, the delivery, the bounce, the complaint and the click;
  • monitor the reputation of your sending domains from bounce and complaint rates;
  • honour unsubscribes, hard bounces and complaints, and keep the suppression list that makes them effective;
  • erase a person when your product asks us to (the user.deleted event);
  • host, back up and secure all of the above.

We do not process the data for our own purposes. We do not sell it, we do not rent it, and we do not use it to train models. We do not combine it with data from other controllers; the one internal view that spans customers is restricted to aggregates and contains no data at the level of an end customer.

3. Categories of personal data

| Category | Typical content | Where it comes from | | --- | --- | --- | | Contact identifiers | Email address, name, company | Your CRM, your imports, your capture forms, your Stripe customer records, the transactional API | | Commercial data | Plan, monthly amount, subscription status, sign-up date, Stripe customer id | Your connected Stripe account | | Fields you define | Any field you choose to keep on a contact | You | | Sending data | The sending of a message to a recipient, its delivery, its bounce with type and reason, the complaint, the click on a link, the unsubscribe and its date | Generated by the service and by Amazon SES | | Transactional messages | Recipient, sender, subject, status. Never the body of the message | Your product, through the API | | Suppression list | Address, reason (unsubscribe, hard bounce, complaint) and date | Generated by the service | | Proof of erasure | A fingerprint of the address and of the external identifier, the date of the request | Generated during an erasure | | Product aggregates | Number of people per day, retention, event volumes | Your connected PostHog project, as query results, never a row per person |

What is not in this list. No card number or bank detail: those stay with Stripe and never reach our servers. No opens, no tracking pixel. No transactional message body. No raw HTML from an analysed page.

4. Categories of data subjects

  • Your customers and former customers.
  • Your trial users, prospects and leads, including the people who fill in your capture forms.
  • Users of your product, as counted by your analytics tool.
  • The recipients of the campaigns, sequences and transactional emails that Canopus sends for you.

5. The controller's instructions

We process personal data only on your documented instructions. Your use of the service is that instruction: what you connect, what you import, the forms you publish, the segments you build, the campaigns and sequences you approve and the messages your product sends through the API.

We will tell you if an instruction appears to us to infringe the GDPR or another data protection provision, and we may suspend the processing concerned until it is resolved.

If Union or Member State law requires us to process data other than on your instructions, we will inform you beforehand, unless that law forbids it on important grounds of public interest.

You warrant that you have a lawful basis for the data you bring into the service, that you have informed the people concerned as the GDPR requires, and that you hold the consent or other lawful basis needed for every campaign you approve. The obligations of the section Sending email of the terms of service are part of this warranty.

6. Confidentiality of personnel

Every person we authorise to process personal data on your behalf is bound by an obligation of confidentiality, whether contractual or statutory, which survives the end of their engagement. Access is limited to those who need it to provide or maintain the service.

Administrative access to an account is itself constrained: it requires a recent authentication, it writes an entry to an audit trail naming the person who acted, and impersonating a user displays to that user a banner that cannot be dismissed.

7. Security measures

We implement the appropriate technical and organisational measures under Article 32. The measures in place are the following.

Access to the data

  • Row level security is enabled and forced on every table in the database, with no permissive policy: any role other than the application's own sees no rows at all. The Supabase anon and authenticated roles hold no privilege on the schema, its tables, its sequences or its functions.
  • Every query touching your data is scoped by an organization identifier resolved from a verified membership, never from a form field or from the URL alone. A request from outside the organization receives a "not found" answer.
  • Authorization is decided in a single module, called by every protected page and every action. It is tested end to end by a suite that actually attempts cross-organization access.

Credentials you entrust to us

  • The API keys you connect are encrypted at rest with AES-256-GCM. The encryption key lives only in the environment, never in the database.
  • Each ciphertext is bound to the organization and connector type it belongs to through additional authenticated data: an envelope copied onto another organization's row does not decrypt.
  • The plaintext exists in two places only: the action that verifies a key against the API before storing it, and the worker that spends it. The queries that display a connector omit the column entirely.
  • Disconnecting a tool deletes the credential in the same request.

Authentication

  • Passwords are at least twelve characters, scored by zxcvbn and checked against Have I Been Pwned through its k-anonymity range API. They are hashed with scrypt.
  • Two-factor authentication by TOTP with backup codes, and passkeys, are available. Magic links and email one-time codes are single-use, expire in ten minutes and are stored hashed.
  • Sessions last seven days with rolling refresh and thirty days at most. Cookies are HttpOnly, SameSite=Lax and secure in production. No cookie cache is used: a revocation takes effect immediately.

Transport and browser

  • HTTP Strict Transport Security with a maximum age of two years, subdomains included, with preloading.
  • A Content-Security-Policy carrying a per-request nonce with strict-dynamic on every route rendered on demand, frame-ancestors 'none', X-Frame-Options: DENY, X-Content-Type-Options: nosniff, a strict referrer policy, a permissions policy denying camera, microphone and geolocation, and browsing context isolation.

Abuse and availability

  • Rate limiting at several levels: a global per-address limit and stricter per-endpoint rules on sign-in, sign-up, password reset, the beta form, connector actions and analyses. It relies on Redis in production, the module refusing to start without it.
  • An anti-bot check on the sign-up, password-reset and beta sign-up forms, when it is configured.
  • Uploaded files are checked by their signature bytes rather than their extension, size-bounded, re-encoded, stripped of their metadata, given a random name and kept in a private bucket reachable only through expiring signed URLs.
  • Requests to a URL you supply are guarded against server-side request forgery: a scheme allow-list, every resolved address checked against the private ranges, and redirects followed by hand and re-checked at each hop.

Traceability

  • An append-only audit trail records security-relevant events with the action, the actor, the target, the organization, the IP address and the user agent. It is never updated or deleted, except by the retention job.
  • Webhooks are verified against the raw body before any other operation, and replay is prevented by an idempotency table keyed on the sender's event id.

Error reporting and logs

  • Personal data is scrubbed from error events before they leave the process, and user context is reduced to an identifier.
  • The logger applies a redaction list, tested against the real list. Secrets, tokens and request bodies are never logged, and neither prompts nor model answers are written to the logs or to the run ledger.
  • Background job payloads are identifiers by construction: personal data therefore cannot accumulate in a queue row or reach an error report that way.

Data minimisation by design

  • Only aggregates reach a model: counts, medians, percentiles and distributions. That boundary lives in the query layer, and so applies to every agent.
  • Analysed pages are reduced to facts; the document is discarded. No screenshot is taken or stored.
  • Sending records the send, the delivery, the bounce, the complaint and the click. No pixel is placed and no open is recorded. The body of a transactional email is never kept.

8. Sub-processors

You give us general authorisation to engage sub-processors. The current list, with each one's role, its place of processing and what it receives, is at /legal/subprocessors, which forms the annex to this agreement.

Before adding or replacing a sub-processor, we inform you at least 30 days in advance, by email and by updating that page. If you object on reasonable data protection grounds within that period, we will work with you to find a solution; failing one, you may terminate the affected part of the service without penalty and obtain a refund of any unused prepaid period.

Each sub-processor is bound by written terms imposing data protection obligations at least as protective as those of this agreement, and we remain fully liable to you for its performance.

9. Assisting you

Data subject rights. Taking into account the nature of the processing, we assist you by appropriate technical and organisational measures in responding to requests under Chapter III. The interface lets you read, correct, export and delete a contact yourself, which answers most requests without our involvement. If a data subject approaches us directly, we will not answer on your behalf: we will direct them to you and inform you within five working days.

Other obligations. We assist you, given the information available to us, with your obligations under Articles 32 to 36: security of processing, breach notification, data protection impact assessments and prior consultation.

Personal data breach. We notify you of any breach affecting data processed on your behalf without undue delay, and in any event within 48 hours of becoming aware of it, so that you can meet your own 72-hour deadline towards the supervisory authority. The notification describes the nature of the breach, the categories and approximate number of data subjects and records concerned, the likely consequences, the measures taken or proposed, and a contact point. Where we cannot provide all of this at once, we provide it in phases without further undue delay. Notifying your supervisory authority and the data subjects is your responsibility as controller.

10. Deletion or return

At the end of the contract, at your choice, we delete or return to you all the personal data processed on your behalf.

  • Deleting an organization cascades to its contacts, connectors, sending domains, campaigns, sequences, transactional emails, suppression list, analyses, audits and reports.
  • Closing your account deactivates it immediately and purges it 30 days later. That window exists so that a compromised account can be recovered, and the privacy policy states it for the same reason.
  • The CRM contacts, with all their fields, can be exported as CSV from the interface at any time, including while a SaaS is paused, for at least 30 days after the plan ends, and during the notice period if the beta closes. That is the return provided for in the first paragraph.

We may retain data where Union or Member State law requires it, for as long as that law requires and for no other purpose. Accounting records such as invoices are the usual case.

Backups disappear on their own rotation and are restored only in the event of an incident.

11. Audit

We make available to you the information necessary to demonstrate compliance with Article 28, and we allow for and contribute to audits, including inspections, conducted by you or by an auditor you mandate.

In practice: the measures in section 7 and the sub-processor list are published and kept current, and we answer a reasonable security questionnaire. An on-site inspection may be requested once in any twelve-month period, with 30 days' notice, during business hours, under an obligation of confidentiality, and conducted so as not to disrupt the service or compromise the confidentiality of other customers' data. You bear the cost, unless the audit reveals a material breach of this agreement.

12. International transfers

Databases, file storage, email sending and application functions are in the European Union.

Where a sub-processor processes personal data outside the European Economic Area and no adequacy decision applies, the transfer relies on the EU-US Data Privacy Framework where the sub-processor is certified under it, and otherwise on the European Commission's standard contractual clauses (Implementing Decision (EU) 2021/914), Module Three (processor to processor), supplemented where necessary by additional measures following a transfer impact assessment. The sub-processor list states where each one processes.

Where those clauses apply, they prevail over this agreement in case of conflict, and their annexes are completed as follows: sections 1 to 4 of this agreement for the description of the processing, section 7 for the technical and organisational measures, and the sub-processor list for the annex of authorised sub-processors.

13. Contact

Questions about this agreement, and any request made under it, go to contact@canopushq.io.