Parmenion data processing addendum

Last updated: October 20, 2026

The short version

This summary helps you read the DPA. It isn't part of it.

  • This DPA is part of our terms. It covers the personal data in what your brand connects or gives us, such as the comments on your ads or details in files you upload.
  • For that data, you decide and we act for you. We only read your Meta ad data. We never change your ads.
  • Your data serves only your brand. We don't sell it, share it, use it for other brands or train AI models on it.
  • If a security incident affects it, we tell your team's owner without undue delay, and within 48 hours of becoming aware of it.
  • When you leave, you can ask for a copy first. We delete your data within 30 days, or within 7 days if you ask us to.
  • The terms' limits on liability apply to this DPA.

1. Who this is between, and how it applies

  • This data processing addendum ("DPA") is between:
    • CHIBBS MANAGEMENT LLC, an Illinois limited liability company, whose mailing address is 1640 Marcie Court, Deerfield, Illinois 60015, United States, and which runs Parmenion ("we", "us"); and
    • the business that accepts our terms of service ("you").
  • It's part of our terms of service ("the terms") automatically, whenever we process personal data for you. You don't need to sign it or ask for it. If you'd like a signed copy, email sam@chibbsmanagement.io.
  • It covers any free audits we offer as well as paid plans, and the private beta. Some features it describes are still being built, such as reading the comments on your ads and your Facebook Pages and Instagram accounts, keeping copies of your ad images and of still frames from your video ads, your brand's board in the Sandbox, uploading files to it, Parmenion AI on the board, the analyzer, the briefer and the ad library. Where this DPA describes them, it applies when Parmenion offers them.
  • If documents conflict: a signed order form comes first, then this DPA on data protection, then the terms (terms §1). No order form or DPA can loosen sections 6, 9, 15 or 17 of the terms, or allow any use of ad data that Meta's terms don't allow. Where the transfer clauses in Schedule 4 apply, they win over the rest of this DPA.
  • Words defined in the terms, such as "your data", "ad data", "outputs" and "team members", mean the same here.

2. Words we use

  • Personal data: information about an identified or identifiable person. It includes "personal information" under US state privacy laws.
  • Covered data: the personal data in your data that we process for you under the terms. Schedule 1 lists what it includes.
  • Data protection laws: the privacy and data protection laws that apply to covered data. That includes the California Consumer Privacy Act, as amended (the "CCPA"), other US state privacy laws and, where they apply, the EU GDPR and the UK GDPR.
  • Controller, processor, business, service provider, sell and share mean what data protection laws say they mean.
  • Meta's terms: Meta's Platform Terms and Developer Policies, and the other Meta terms named in section 5 of the terms.
  • Subprocessor: a company we use that processes covered data for us.
  • Security incident: a breach of security that leads to covered data being destroyed, lost or changed by accident or unlawfully, or disclosed to or accessed by someone without permission. Attempts that don't get through, such as blocked sign-ins or scans, aren't security incidents.

3. Our roles

  • For covered data, you're the controller (a "business" under the CCPA), and we're your processor (your "service provider" under the CCPA).
  • If you connect ad accounts for an advertiser, as an agency does, you're that advertiser's processor, and we're your subprocessor. You confirm the advertiser has authorised you to use Parmenion for it (terms §5).
  • Under Meta's terms, we're a "Tech Provider" and you're our "Client". We process your ad data for you, at your direction (terms §6).
  • For your account, billing and usage data, we're the controller. This DPA doesn't cover that data; our privacy policy does. Usage data holds none of your ad data or files, and nothing your team types to Parmenion AI or it writes back (terms §8).

4. What we process, and your part

  • What and why. Schedule 1 sets out the covered data, whose it is, what we do with it and why.
  • Your part. You're responsible for:
    • having the right, and a lawful basis, to give us covered data, including people's details in files your team uploads;
    • giving the people in it any notices your law requires, including for sensitive details people write in comments, or in reviews your team uploads as reviews, such as health details;
    • making sure your instructions follow data protection laws and Meta's terms;
    • not typing sensitive personal information into Parmenion yourself, or uploading files that hold it or children's details (terms §5);
    • claims that come from what your team uploads, which you cover us for (terms §20).
  • Files that break this. We may remove a file your team uploads that breaks section 5 of the terms.

5. Your instructions

  • We process covered data only on your documented instructions. They are:
    • the terms and this DPA;
    • the choices your team makes in Parmenion, or asks us for by email, such as which ad accounts to connect and, when Parmenion offers them, which Pages and Instagram accounts, your settings, exports, disconnecting and deleting, inviting our staff to your team for support and removing them, and, when Parmenion offers them, running the analyzer and the briefer;
    • if your brand is on the Parmenion Super Strategy, your contract with us, which lets our staff switch into your account as section 6 says, and your owner's switch for it in Settings;
    • any other written instructions we both agree to.
  • Commenters and reviewers. You also instruct us to delete a commenter's data, or a review your team uploaded as a review, when that person asks us directly. We'll tell you when we do.
  • If an instruction is a problem. We'll tell you if we think an instruction breaks data protection laws, and we don't have to follow it. We can't follow an instruction that Meta's terms don't allow.
  • If the law requires something else. If a law requires us to process covered data in another way, we'll tell you first, unless that law forbids it.
  • Requests from authorities. If a government body or court asks us for covered data, we check that the request is lawful, push back on any that isn't or that asks for too much, give only what's required, keep a record, and tell you unless the law forbids it.

6. Your data serves only your brand

This section covers all of your data, not only personal data.

  • We use your data only to run Parmenion for your brand, which means measuring and improving your own Meta campaigns.
  • Your data may improve your own plan and strategies. We never use it for another customer: not for benchmarks, pooled or shared models, cross-brand reports, or to improve Parmenion for other customers.
  • We keep each customer's data separate from every other customer's. Each team is for one advertiser.
  • We don't train AI models on your data. We may fit your brand's own settings and suggestions to your brand's own history, for your brand only. What we fit for one brand is never used for another, and it's deleted with your data.
  • We work out your brand's P1 score (how often its suggestions beat a plain guess on its own past) and its own fitted settings from its own history, for your brand only.
  • What your team types to Parmenion AI, in Ask and, when Parmenion offers AI on your board, on the board, and what it writes back, are used only to answer you and to fix problems you report. Anyone on your team can share their own conversation with us after a thumbs-down, when Parmenion offers it, with what the AI read for it. We read that one conversation only to fix the problem, and don't keep it in our tests or use it to train any AI. We delete what your team types and what the AI writes back after 90 days, or with your ad data if that's sooner. What your team adds to its board stays with its board (strategies keep their own clock).
  • We never use your ad data to decide whether to offer you a larger plan, in our marketing emails, or for any other business of CHIBBS MANAGEMENT LLC.
  • Our staff look at your data only: when your owner invites one of us to your team for support, as a member of your team, until the owner removes us; to investigate a security problem; to read a conversation your team shares after a thumbs-down, when Parmenion offers it (above); to check a report that a file your team uploaded breaks the terms; or, if your brand is on the Parmenion Super Strategy, in a switch-in (below). The invite is the record of a support visit, and we keep a record of every other look.
  • The Parmenion Super Strategy. If your brand is on it, our staff can switch into your brand's account to build and tune your Super Strategy, as your contract says. They can do what a member of your team can, including exporting your data and adding members to your team, and they can save your goal, winner rules and plan setup as an admin would. They can't see billing, connect or disconnect an ad account, delete your data or close your account. Your owner can turn this off in Settings, and sees each switch-in, change and export there. What our staff export serves your brand only: it never goes into another brand's account or work, or into an outside AI tool.
  • The one exception is your strategies: our staff can see each strategy's rules, settings, names, own notes and versions, with the definitions of the custom metrics a saved version uses, which hold none of your ad data, whenever needed, to run, fix and support Parmenion, except a Super Strategy's own strategies, which staff see only in a switch-in. What a strategy shows when it runs on your numbers is covered by the line above. Staff never see a brand's tags, tag rules, customer map, custom metrics or 90-day customer value, except a custom metric's definition inside a saved strategy version, a conversation your team shares after a thumbs-down (above), and as a member of your team your owner invited, or in a Super Strategy switch-in. The notes and docs on a brand's board, and the files on it, are covered by the line above.
  • We don't keep summarised or "anonymised" versions of your data after it's deleted, and we don't make de-identified data from it for any other use.

7. No selling or sharing

For covered data, we act as your service provider for the specific business purpose in Schedule 1. So:

  • We don't sell or share it, as the CCPA uses those words, and we don't rent or license it.
  • We don't keep, use or disclose it for any purpose other than that business purpose and the other uses the CCPA's rules allow a service provider, such as keeping Parmenion secure, fixing errors and meeting legal duties. That rules out any other commercial purpose.
  • We don't keep, use or disclose it outside our direct business relationship with you.
  • We don't combine it with personal information from other customers. We don't combine it with personal information from other sources, except as the CCPA's rules allow.
  • We don't use it for advertising or retargeting, to build profiles of people, or to make decisions about people, such as for jobs, credit or housing.
  • We give it only to:
    • our subprocessors (section 10);
    • services you tell us to send it to;
    • others where the law requires it, and then only what the law requires;
    • a company that takes over Parmenion, under section 23 of the terms, and only if Meta's terms allow it.

    We never give it to partners who refer customers to us.

  • We comply with the CCPA, and with the other US state privacy laws that apply to us as your processor, and give covered data the level of privacy protection they require of you.
  • We'll tell you promptly if we decide we can no longer meet these duties.
  • You may take reasonable and appropriate steps to check that we use covered data as this DPA says (section 14). After telling us, you may also take reasonable steps to stop and fix any use that isn't allowed.

8. Confidentiality

  • Only people who need covered data to run Parmenion for you, keep it secure or meet a legal duty can reach it.
  • Everyone who can reach it, our own people and our subprocessors, is bound by confidentiality duties at least as strict as this DPA.
  • Section 10 of the terms also applies. For your data, confidentiality lasts as long as we hold it.

9. Security

  • We protect covered data with the technical and organisational measures in Schedule 2. They're chosen to fit the risk to people, given the kind of data, how we process it and the state of the art.
  • We may change these measures over time, but we won't make them less protective overall.
  • No system is perfectly secure. Section 11 says what happens if something goes wrong.

10. Subprocessors

  • Your authorisation. You authorise us to use the subprocessors on our subprocessors page, which is this DPA's list (Schedule 3).
  • How we bind them. Before a subprocessor handles covered data, we bind it with written data protection terms that limit it to our instructions and are at least as protective as this DPA. We're responsible for what our subprocessors do with covered data, as we are for our own acts.
  • New ones. We'll tell your team's owner by email, and update the page, at least 30 days before a new subprocessor starts handling covered data, or an existing one starts handling a new kind of it.
  • Emergencies. If we must replace a subprocessor to keep Parmenion running or secure, or Meta tells us to stop using one, we'll tell you as soon as we can. You can still object.
  • If you object. Email sam@chibbsmanagement.io with reasonable data protection grounds. We'll look for a fix with you. If we can't find one, you can cancel that brand's plan and get a refund of what you prepaid for the unused time on a plan paid by the month or year. A lifetime founding seat isn't refunded (section 13 of our terms).
  • When we stop using one, we stop sending it data, ask it in writing to delete what it holds, and keep its reply.
  • Their own subprocessors aren't covered by our 30-day notice. Each one announces its changes on its own list, linked from our page.
  • Meta isn't our subprocessor. It's where your ad data comes from.

11. Security incidents

  • Telling you. If a security incident affects covered data, we'll tell your team's owner by email without undue delay after we become aware of it, including when a subprocessor tells us, and within 48 hours of that. If a law sets a shorter time, we follow the law.
  • What we tell you, as far as we know it at the time:
    • what happened and when;
    • which data was involved, and which wasn't;
    • about how many people and records were involved;
    • what it could mean for them;
    • what we've done, and what you need to do, if anything;
    • what we don't know yet, and when you'll hear from us next;
    • who to contact.

    If we can't give it all at once, we'll send it in stages as we learn more.

  • Your notices. Covered data is yours, so you decide whether to tell the people in it, or a regulator. We'll give you reasonable help and the information we have.
  • What we do. We act promptly to stop the incident, limit the harm, find the cause and keep the evidence.
  • Meta. If an incident involves data from Meta, we also report it to Meta, as Meta's terms require.
  • Telling you about an incident isn't an admission of fault.
  • To report a security problem to us, email sam@chibbsmanagement.io.

12. Helping you

  • People's requests. Owners and admins can use Export everything in Settings, under Your data, or ask us for an export of your data, and the owner can use Delete my data there or ask us to delete it, by email at sam@chibbsmanagement.io. If you need more help to answer a person's request about covered data, for example to see, correct, delete or move it, or to object, email us. We'll reply within 7 days, and help in full within 30 days.
  • Requests that reach us. If someone asks us directly about covered data, we'll pass the request to you promptly, and won't answer it ourselves unless you tell us to or the law requires it. The exception is a commenter who asks us to delete their comment, or a reviewer who asks us to delete a review your team uploaded as a review: under your instruction in section 5, we delete it within 30 days and tell you.
  • From Meta. If Meta sends us anything about a person's request about their data, we pass it on to you.
  • Assessments and regulators. We'll give you reasonable help, with the information we have, for your data protection assessments, your consultations with regulators, and your own security and breach-notice duties.

13. Deleting and returning your data

This section covers all of your data, not only personal data. Our How to delete your data page gives every step and copy, with the same numbers.

  • Getting a copy first. Until your data is deleted, owners and admins can use Export everything in Settings, which holds the ad data we read from Meta, your brand's settings and answers, your strategies, tags, tag rules, custom metrics, customer map and angles, and, when Parmenion offers the board, your boards and a list of the files on them, or ask us for a copy of any of it, including during the 30 days below, even without a plan. That's how you get your data back. We build an export from Settings when it's asked for and keep no copy of it.
  • When you leave. We delete our access keys (tokens) within 1 hour, unless other ad accounts on the plan still use that access, and your data, including the outputs made from it, within 30 days after any of these (your board and its files, and your settings and strategies, follow your plan's clock, below):
    • you disconnect an ad account, Page or Instagram account (then only that one's data goes);
    • our access stops working, including when you remove Parmenion in Meta's settings and Meta refuses our access at our next daily read;
    • your paid period ends after you cancel, or your plan ends for any other reason, including 7 days after a payment fails.

    If you reconnect, or restart your plan, before then, we stop the clock and nothing is deleted. We read nothing new while the clock runs.

  • When you ask. If your owner asks us to delete your data, by email or with Delete my data in Settings, or is your last owner and closes their account, we delete the tokens within 1 hour of confirming the request is the owner's, and your data within 7 days, with no waiting period.
  • Your board. Your board, when Parmenion offers it, and the files on it belong to your brand, not one ad account, so "When you leave" works differently for them. We delete them within 30 days after your plan ends, or as "When you ask" says. If you restart your plan before then, nothing is deleted. Disconnecting an ad account, or our access to one stopping, doesn't delete them, but the outputs made from that account's data leave your board, and its saved versions, with that data, including anything Parmenion AI made from it. If your team deletes a file from your board, we delete it, with a CSV's rows, within 7 days.
  • A board you delete. When Parmenion offers the board, deleting a board deletes its notes, docs, versions and files within 7 days, and undo can't bring them back. Its strategies stay.
  • Your settings and strategies. Your goal, winner rules, strategies, tag rules, custom metrics, metric cards, customer map (its avatars and pain list) and 90-day customer value belong to your brand, not one ad account, so "When you leave" works differently for them. So do, when Parmenion offers them, your ad library, the angles your team keeps, concepts your team writes, your subscriber counts, your logo and colours, and a Super Strategy onboarding's answers. We delete them within 30 days after your plan ends, or as "When you ask" says. If you restart your plan before then, nothing is deleted. Disconnecting an ad account, or our access to one stopping, doesn't delete them, but facts in them drawn from that account's data go with it. The labels your team puts on ads, and the avatars linked to an ad, go with those ads' data. A file your team deletes from its ad library is deleted within 7 days.
  • What's worked out from an ad account's data. When Parmenion offers them, the analyzer's reads, transcripts and sound numbers, its checks and accuracy history, concepts Parmenion AI writes from that account's data, and the dated record of each live suggestion and its inputs follow that account's clock under "When you leave". A video file the analyzer fetches is deleted within 1 hour.
  • A strategy you delete. When Parmenion offers the board, a strategy deleted from your board is kept for 24 hours, with its versions, so undo can bring it back, then deleted.
  • Free audits. If we give you a free audit, we delete the token within 1 hour of building it, and the audit's data 30 days after it's ready, unless you start a plan.
  • When Meta asks. We delete what Meta names, as Meta directs, within 7 days at the latest.
  • Backups and vendor copies. Deleted data leaves our database backups within a further 7 days. Copies at our subprocessors expire on their own schedules, which the deletion page lists.
  • Outside these clocks: anything an AI provider flags for misuse and keeps under its own rules, anything the law makes us keep, and anything under a legal hold. We keep these only as long as we must, and this DPA still covers them while we do.
  • What we keep after deletion. A record of each deletion (its dates and what kind of data went) for 3 years, and the other records in section 14 of the terms. None of them holds the ad data we read from Meta.
  • Proof. On request, we'll confirm in writing when a deletion is done.
  • Copies you hold. Ad data you've exported, or copied into other tools, is yours to delete, and section 6 of the terms still applies to it.

14. Showing we keep to this DPA

  • Information. On request, we'll give you the information reasonably needed to show we meet this DPA, such as our subprocessors list, a summary of our security measures, and written answers to a reasonable security questionnaire once a year.
  • Certifications. We don't hold a security certification, such as SOC 2 or ISO 27001.
  • Audits. If that information isn't enough, or a regulator requires more, you or an independent auditor you choose may audit our compliance with this DPA. You give us 30 days' written notice, audit no more than once a year unless a regulator requires more, and pay your own costs. The auditor must keep what it sees confidential, and no audit may expose other customers' data or our trade secrets beyond what the audit needs.
  • Fixing problems. If an audit shows we don't meet this DPA, we'll fix it promptly, at our cost.
  • Meta. Meta may also review how we handle data from its platform, under Meta's terms.

15. Where covered data is processed

  • US only, for now. Parmenion is only for businesses based in the United States, apart from founding members (section 13 of our terms).
  • Where it's kept. We and our subprocessors store covered data in the United States: our database with Neon, on AWS in N. Virginia, and our app, file copies, the files you upload (when Parmenion offers the board) and background jobs with Vercel, in Washington, D.C. Vercel's network may cache pages and files briefly near wherever you open Parmenion. Anthropic may process AI requests in other countries. Our subprocessors page says where each one works.
  • People outside the US. Comments on your ads can come from people anywhere.
  • Transfer clauses, later. Before we accept a business based in the EU, the EEA, the UK or Switzerland, or a founding member's brand there connects, we'll add the transfer clauses in Schedule 4.

16. AI

  • Who. Our only AI provider for covered data is Anthropic (Claude), a subprocessor. We call it directly through its API, under our own agreement with Anthropic.
  • What for. It may label your ads, sort comments and reviews by the questions customers raise, write the short summaries around the numbers our code works out, answer questions in the Ask box, and help draft and explain strategies. It may also explain a chart when your team clicks Analyze (when Parmenion offers it), and, when Parmenion offers AI on your board and your team uses it there, read the board and its files and draft items for your team to add or discard. When Parmenion offers them and your team asks, it also reads your ads for the analyzer (their pictures, on-screen text, words and transcripts), finds angles and writes ad concepts, and drafts avatars for your customer map from comments.
  • What it gets. Only what the task needs: numbers our code worked out, ad text, ad images and still frames from videos where a task needs them, and, when Parmenion offers the analyzer, on-screen text and transcripts, comment and review text, your team's Ask questions, what your team types for the briefer, and, when Parmenion offers AI on your board, the board items and uploaded files a task needs, with any names in the files. Never your tokens, your billing data, or commenters' names or IDs from Meta. It has no direct access to our database.
  • No identification. Nothing in Parmenion identifies a person from a face or a voice. The analyzer makes transcripts on our own servers and measures how a voice sounds as numbers; it keeps no audio, makes no face or voice templates, labels no speaker, and labels no one's emotions or traits.
  • How. We send covered data to Anthropic only under a written agreement that bars it from training on that data, and every request that carries your data uses zero data retention. Under those settings, Anthropic doesn't keep the request or the answer after it replies, except where the law requires it or a request is flagged for misuse of its service. When Parmenion offers AI on your board and it reads your files, they go inside the request like everything else; we don't use Anthropic's file storage.
  • You decide. The AI can't change your ads, and nothing it writes is carried out on its own. When Parmenion offers AI on your board, what it makes there stays a draft until your team adds it. Everything Parmenion AI writes carries a "Parmenion AI" mark until someone on your team edits it (it then shows "Edited") or removes the mark, and we record who removed which mark and when.

17. Meta's terms

  • For data from Meta, Meta's terms apply alongside this DPA, and nothing in this DPA overrides them.
  • If Meta asks us to, we'll suspend or end your access to ad data through Parmenion, and we'll give Meta our list of clients, with your business name and contact details (terms §6).

18. Liability

  • This DPA is part of the terms, so section 19 of the terms (limits on liability) applies to all claims under this DPA and the terms together, including its exceptions, such as anything the law doesn't let us limit.

19. How long this lasts, and changes

  • This DPA lasts as long as we process covered data for you. It keeps applying after the terms end, until your data is deleted under section 13.
  • We may update this DPA as section 22 of the terms says. We'll tell your team's owner at least 30 days before a change that matters takes effect. We won't change how we use your data, or protect it less, without your express consent.

20. Contact us

CHIBBS MANAGEMENT LLC, 1640 Marcie Court, Deerfield, Illinois 60015, United States


Schedule 1. The processing

  • Subject matter: providing Parmenion to you under the terms, including any free audits.
  • How long: while you use Parmenion, and until your data is deleted under section 13.
  • What we do with it: read it from Meta with read-only access; store it, with the files your team uploads when Parmenion offers the board; work out your numbers (winners, testing, scaling, angles, your plan and your suggestions); show your files on your board and use them in your strategies and metrics; use AI to label your ads, sort comments and reviews, write summaries and Ask answers, and, when Parmenion offers AI on your board and your team asks, read your board and its files and draft items for it; when Parmenion offers them and your team asks, read your ads with the analyzer (with a transcript we make and sound numbers we measure), keep your ad library, and find angles and write ad concepts; show it to your team; export it when you ask, and copy a concept out when your team uses Copy or Hand off; and delete it. We never change your ads.
  • The business purpose: analysing your own Meta ad data, ad comments, uploaded reviews and, when Parmenion offers the board, the files your team uploads, to show your testing, winners, scaling and angles, to keep your files and show and use them on your board and in your strategies, and to write your plan text, Ask answers and, when Parmenion offers AI on your board, the drafts Parmenion AI makes there when your team asks. When Parmenion offers them, also reading your ads with the analyzer, keeping your ad library, and finding angles and writing concepts when your team asks; and, for a brand on the Parmenion Super Strategy, building and tuning its Super Strategy, and running the briefer for it, under its contract. Also keeping Parmenion secure, fixing problems, giving you the support you ask for, and meeting our legal duties.
  • How often: about once a day while an ad account is connected on a plan, after a first read of up to a year of history. Once for a free audit.
  • Whose data:
    • people who comment on your ads on Facebook or Instagram;
    • people whose reviews you upload;
    • people who appear in, are named in, or are heard in your ads, including their voices in video ads (never identified);
    • the person on your team whose Meta login connects Parmenion;
    • anyone your team names in an Ask question;
    • people named or shown in files your team uploads, or named in what your team writes on its board.
  • What data:
    • comment text, when it was posted and which ad it's on. Commenters' Meta IDs are kept only as a code made with a secret key for your brand. We drop names, usernames, profile pictures and @-mentions;
    • review text, rating, date and product. We drop reviewers' names and emails when you upload them;
    • your ads' text, which images and videos they use, and the IDs of the Facebook or Instagram posts they run as, and, when Parmenion offers them, copies of your ad images and of still frames from your video ads (we keep no whole video files), which can show people;
    • when Parmenion offers the analyzer: a video ad's file, held only briefly and deleted within 1 hour; its transcript, and numbers measured from its sound (pace, pauses, loudness, how much is speech, music under the voice and pitch variety), never the sound itself; and the analyzer's reads;
    • when Parmenion offers them: the ads your team uploads to its library (images and video, with location data removed); avatars, pains and angles; ad concepts; subscriber counts (never names); your logo and colours; and a Super Strategy onboarding's answers;
    • your ad account's activity history: the budget and status changes made to its campaigns and ad sets, and when. We don't keep who made them;
    • the Meta access tokens your connection uses, stored encrypted, and the ID Meta gives your business;
    • what your team types to Parmenion AI, in Ask and, when Parmenion offers AI on your board, on the board, and what it writes back, including a conversation a team member shares after a thumbs-down, when Parmenion offers it, with what the AI read for it (drafts your team adds become part of its board, below);
    • anything personal in your goal, winner rules, customer profile or other settings;
    • files your team uploads to its board when Parmenion offers it (CSV, PDF, JPG and PNG), kept as uploaded in Vercel's private file store, with a CSV's rows also in our database;
    • files your team exports from Meta itself, such as Ads Manager reports, which we keep and delete with your board, and count as data from Meta when we report a security incident (section 11);
    • the notes, text, docs and flowcharts your team puts on its board.
  • Sensitive data: we don't ask for any. Comments, and reviews your team uploads as reviews, can hold anything people chose to write, including health details. The terms ask you not to type sensitive personal information into Parmenion, and not to upload files that hold it or children's details (terms §5).
  • How long we keep it: section 13, and our deletion page.
  • Subprocessors: Schedule 3.

Schedule 2. Security measures

We'll have these measures in place before we process covered data for you, and keep them in place while we do.

  • Walls between brands. Each customer team holds one advertiser's data. Every table with brand data carries the team's ID, our code filters every read by the signed-in team, and the database enforces the same wall on its own (row-level security).
  • Read-only access to Meta. We ask Meta only for the permissions we need to read. Our Meta client sends only read requests, apart from the call that revokes a token.
  • Tokens. Meta tokens are encrypted with a key held in a separate key service, outside our database. They never reach a browser, a log or an AI provider.
  • Encryption. HTTPS only, with TLS 1.2 or higher in transit. Encryption at rest for the database, the file store and background jobs.
  • Uploads. When Parmenion offers the board, we accept only CSV, PDF, JPG and PNG files, checked by their contents, not only their names, and up to a size limit. They go to a private store that needs a key for every read, in a folder for your brand, which our deletion process empties and checks. We remove the location saved in images when they're uploaded. When Parmenion offers the ad library, it also takes MP4, MOV, WebP, GIF and ZIP files, with its own limits, skips duplicates and removes location data.
  • Access. Everyone on our side who can reach covered data signs in with multi-factor authentication, and gets only the access they need. Access is removed when someone no longer needs it. Staff see a brand's data only as a member of its team its owner invited for support, to investigate a security problem, to read a conversation the brand's team shares after a thumbs-down, when Parmenion offers it (section 6), to check a report that an uploaded file breaks the terms, or in a Super Strategy switch-in under the brand's contract: on only while the owner's switch is, recorded switch-in by switch-in with every change and export, one brand at a time, and never reaching billing, deletion or the brand's connections. We keep a record of each look outside an invite. The one exception is strategies (their rules, settings and versions, with the custom-metric definitions a saved version uses, which hold no ad data), which staff can see whenever needed, except a Super Strategy's own.
  • The AI. The AI calls a few read-only functions locked to one brand, never the database. Comments, reviews and uploaded files are treated as material to read, never as instructions. When Parmenion offers AI on your board, it reads a file's contents only when your team asks it to.
  • Logs. A log of sign-ins, data reads, exports, deletions and admin actions, kept 12 months, with IP addresses removed after 90 days. It never holds tokens, secrets, comment or review text, file names or contents, or anything typed to or written by Parmenion AI.
  • Backups. The database can be restored to any point in the last 7 days. After a restore, finished deletions are re-applied before the app goes back up.
  • Secure development. Automated tests and security checks, including secret scanning and dependency checks, run on changes before they go live.
  • Devices. No covered data on staff laptops or phones, apart from Meta's list of IDs to delete, held for less than a day on an encrypted laptop.
  • Incidents. A written incident response plan.
  • Vendors. A written agreement with each subprocessor before it gets covered data.
  • Deletion. One deletion process for every trigger, a check after each deletion, and a record of each one.

Schedule 3. Subprocessors

The list is our subprocessors page, as updated under section 10. It says what each subprocessor does, what data it gets, where it works, and where to read its terms.

Schedule 4. Transfer clauses

Until we accept a business based in the EU, the EEA, the UK or Switzerland, this schedule is empty. Before we do, we'll add the EU Standard Contractual Clauses (Module 2, controller to processor, and Module 3, processor to processor, for agencies that connect for an advertiser), the UK Addendum and the Swiss changes, with their annexes filled from Schedules 1 to 3.

Signed copies

This DPA applies without a signature (section 1). If you'd like a copy signed by both of us, email sam@chibbsmanagement.io.