chatleadr Docs

Privacy Policy

Effective date: 1 September 2026

Last changed: 15 November 2026

Who we are: Jadin Roste Andrews, trading as Chatleadr, a sole proprietor trading in the Republic of South Africa, of 71 Old Main Rd, Hillcrest, 3650. Chatleadr is not a registered company, and one person trading in their own name has no registration number to publish. In this policy "we", "us" and "Chatleadr" mean that person.

Information Officer: Jadin Roste Andrews, support@chatleadr.com. POPIA makes the head of a private body its Information Officer, so this is the position whether or not registration with the Information Regulator has been completed.

Contact: support@chatleadr.com

1. Two Roles in This Policy#

Chatleadr handles two different kinds of personal information, and we are not in the same legal position for each. Reading the rest of this policy without this distinction will give you the wrong answer.

Account information is ours to answer for. When you sign up, buy a plan or email support, we decide why and how that information is processed. Under the Protection of Personal Information Act 4 of 2013 ("POPIA") we are the responsible party; under the General Data Protection Regulation ("GDPR") we are the controller. Sections 3, 5 and 8 of this policy describe that information.

Conversation information belongs to the customer whose bot collected it. When a visitor talks to a bot built by one of our customers, the customer decides what that bot asks, what it stores and what it is for. The customer is the responsible party or controller. We process it only on their instructions, which makes us the operator under POPIA and the processor under GDPR. The terms of that processing are in the Data Processing Terms. They include using conversations to find faults and to check and improve the quality of the service, and never to train models.

If you are a visitor who spoke to a bot and you want your information removed, the business that runs the bot is the right place to ask. Section 10 explains what to do if you cannot identify them.

2. Who the Parties Are#

TermMeaning
CustomerThe business or person with a Chatleadr account.
VisitorAnyone who talks to a bot a Customer has published.
AgencyA Customer who builds and runs bots for its own clients.
ClientA business an Agency runs a bot for.

Where an Agency runs a bot on behalf of a Client, the Agency is responsible to us for that bot and the Client's own relationship is with the Agency. Clause 4 of the Data Processing Terms sets out what the Agency has to have in place before it does that.

3. Account Information We Collect#

Collected from you directly, when you create an account or use the Dashboard:

  • Identity. Your email address and, if you provide it, your full name. Sign-in is handled by Google Firebase Authentication; we store the account identifier it issues, not your password. We never see your password.
  • Workspace and configuration. Your bots, their settings, your knowledge documents and your workflows.
  • Usage records. Conversation counts per day, workflow executions, and the credit ledger that supports your invoice.
  • Billing records. Which plan you are on, when it renews, and the reference our payment provider gives each transaction. We do not receive or store your card number. Payments run through our payment provider, which is the merchant of record and holds the card details. The provider that took a payment is named at checkout and on the receipt for it.
  • Support correspondence. Anything you send us by email.

Visitors to Our Website#

When you visit chatleadr.com or our documentation at docs.chatleadr.com, we use Google Analytics to count visits and to learn which pages, searches and adverts bring people to the two sites. Google Analytics sets cookies in your browser and records the pages you view, how you reached the site, your approximate location, and your device and browser. It does not receive your name or email address. Google processes this information for us and keeps it for at most 14 months.

Google Analytics does not run if your browser is set to a time zone in Europe, which covers the European Union, the United Kingdom and Switzerland, or if your browser sends a Global Privacy Control signal. To stop it on any visit, block cookies for chatleadr.com in your browser.

Google Analytics runs only on chatleadr.com and docs.chatleadr.com. It does not run in the Dashboard, and it does not run on the websites where Customers publish their bots.

4. Conversation Information a Bot Collects#

This is the information described in section 1 as belonging to the Customer. We process it on their behalf. What a bot collects depends entirely on how that Customer configured it, so the list below is what the software is capable of storing rather than what any particular bot does.

  • The conversation. Every message in the exchange, the bot's replies, and the record of which workflows ran.
  • Lead details, if the bot asks for them. Name, email address and phone number are stored as named fields. Anything else a form collects is stored with it.
  • Channel identity. On WhatsApp or Instagram this is the account the visitor messaged from. On the web widget there is no such identifier.
  • Uploaded files, where a Customer's form includes a file upload, or a Customer's workflow receives a file the visitor sends in the chat. A curriculum vitae or an identity document uploaded this way carries far more personal information than the rest of a conversation. Uploaded files have a retention window of their own, which a Customer can set shorter than the one for conversations.
  • Photos sent in a conversation on WhatsApp or Instagram, where a Customer has asked us to switch on photo understanding for a bot. Our model provider checks each photo for harmful content and describes it in a few words, which the bot reads before it replies, and we keep the photo with the conversation, where it is screened like any other stored image. A photo of a person is described only as a photo of a person. A photo of a bank card or an identity document is described by its kind alone and is never kept. A photo the check stops is neither described nor kept. A photo from somebody the Customer has erased is described but not kept. Without photo understanding, a photo sent outside a workflow is not kept.
  • Traffic source, meaning the page the widget was opened on, any campaign parameters in its address, the address of the site the visitor followed a link from (never the page on it), and which ad network marked the visit as an ad click (never the click code). On WhatsApp or Instagram it is what Meta sends about how the conversation started: the ad, story, shared post or catalogue product, and the account's profile name, username and number kept as separate fields.
  • Screening results. The result of screening each image from a visitor that the service stores, and a mark on any visitor message flagged as harmful content, with the categories it was flagged for. Section 8 explains the screening.

We do not record a visitor's IP address or browser user agent. No field for either exists on any record the chat runtime writes.

5. Why We Process Account Information#

PurposeBasis under GDPRBasis under POPIA
Providing the service you signed up forContractNecessary to perform a contract
Taking payment and keeping billing recordsContract, and legal obligation for tax recordsContract, and obligation imposed by law
Security, abuse prevention and debuggingLegitimate interestsLegitimate interests
Service email about your account, outages and payment failuresContractNecessary to perform a contract
Marketing email about features and offersConsentConsent
Counting visits to chatleadr.com and which adverts bring themLegitimate interestsLegitimate interests

Marketing email is separate from service email and you can withdraw consent at any time without losing access to anything you have paid for. Service email about a failed payment or an outage is not something we can exclude you from, because it concerns the contract itself.

6. How Long We Keep It#

Retention differs by class of data and by plan. These windows are enforced in software rather than by policy alone: transcripts and technical logs expire automatically, and stored documents are swept by a job that deletes the encrypted file as well as the record pointing at it.

DataFreeStarterGrowth, Scale and Agency
Uploaded documents30 days1 year3 years
Conversation transcripts30 days1 year3 years
Technical execution logs7 days7 days30 days
Saved ad and story picturesWhile a conversation or lead shows themWhile a conversation or lead shows themWhile a conversation or lead shows them

A photo kept from a conversation follows the uploaded documents row, and counts toward the storage a plan allows for uploaded documents.

Daily analytics totals are kept for 400 days on every plan, so that a year-on-year comparison still works in January.

Saved ad and story pictures are kept for as long as something shows them. When a Visitor messages a Customer's bot from a Meta ad, or replies to the Customer's own story on Instagram, we copy the ad's picture, a still from a video ad, or the story's picture as that message arrives, because the links Meta provides to them stop working. A lead never expires, and the copy is what shows the lead's ad after those links have gone. The pictures are the Customer's own ads and stories rather than anything a Visitor supplied, so erasing a contact does not remove them. A job that runs every night deletes each one once no conversation or lead shows it, and every one is destroyed when its bot or its workspace is erased. They are encrypted and stored in the same way as uploaded documents, and they do not count toward the storage a plan allows for uploaded documents. A picture from a Visitor's own story is never kept.

Leads and contacts are not deleted by age on any plan. A contact list that empties itself is not a contact list. They are deleted when the Customer deletes them, or under section 7.

Opt-out and suppression records outlive every window above. If someone asks not to be contacted, we have to keep enough information to honour that. Deleting the record of an opt-out would cause the very contact it was meant to prevent.

A Customer may choose a shorter window than their plan allows. They cannot choose a longer one.

When a Customer moves to a plan with a shorter window, nothing is deleted for at least 30 days, so that there is time to export first.

7. Deletion, and What It Removes#

Three different things can be deleted, and they behave differently. All three are available in the Dashboard.

Erasing One Person#

A Customer can erase a single contact from their workspace on request. This is the mechanism a Visitor's erasure request is answered with, and it is irreversible.

What is destroyed: the lead record and its tags, every file that person uploaded, and every photo of theirs kept from a conversation. Each file's encrypted contents are destroyed before the record pointing at it, so nothing is left behind that we hold but can no longer identify.

What is redacted rather than deleted: the conversation transcript. Every message is replaced with a marker, including the description of any photo, and the sender identity, the stored summary, the channel account the person messaged from and the record of what brought them are removed. What remains is how many turns there were and when. That is billing evidence for a month already invoiced, and it identifies nobody once the contents are gone.

What is kept, deliberately: the record that the person asked. We keep a one-way fingerprint of their email address, phone number or channel account so that the same person is not captured again by any bot in that workspace. The record applies only in the workspace where the erasure was made. It is copied to another workspace only when a bot is moved there, and it does not apply anywhere else on the service. Deleting that record would cause the contact it exists to prevent. Billing and usage totals are kept for the same reason as always: they are financial records and carry no contact details.

The Customer receives a receipt setting out exactly what was removed. It names nobody, because reprinting the address would undo the erasure.

Deleting a Bot#

Archiving a bot stops it answering on every channel at once, removes it from your lists, and frees the plan slot immediately. Nothing is destroyed and it can be restored whole.

Erasing an archived bot destroys its workflows, knowledge documents, conversations, leads, uploaded files and analytics. It is irreversible, and a bot has to be archived before it can be erased.

Deleting a Workspace or Closing an Account#

Deleting a workspace is reversible for 30 days, and the workspace is still billed for that period, because we are still storing it. Restoring it during that window brings everything back.

Erasing a workspace destroys everything in it: every bot, workflow, knowledge document, conversation, lead, uploaded file, integration, invitation and usage record. The per-workspace key that the uploaded files are encrypted under is destroyed last, which makes any remaining file permanently unreadable. Live WhatsApp and Instagram bindings are released with the provider, so a number is not left pointing at a bot that no longer exists.

A workspace that pays for client workspaces cannot be deleted until those are transferred or deleted first.

Where the Customer does not ask, we erase a closed account's data within 90 days of it closing. Backups follow their own cycle and are overwritten within 35 days.

8. Who Else Processes It#

We use other companies to run parts of the service. Each one receives only what it needs for its function. The current list, with what each one does and where it does it, is on the sub-processors page, which we update when the list changes.

Three of them are worth naming here because of what they receive:

  • The model provider receives the conversation text and the bot's configuration in order to generate a reply, and to run the quality checks the Data Processing Terms describe. It also receives the words of each message a visitor sends, and a copy of each image from a visitor that the service stores in a format the model provider can read, in order to screen them for harmful content, as described below. Those images include images stored before screening began and photos of documents uploaded through a form. Where a Customer has asked for photo understanding, it also describes each photo a visitor sends on WhatsApp or Instagram, as section 4 explains. It is instructed not to use any of that content to train models. It keeps what it receives for up to 30 days to check its service for abuse, and for longer only where the law requires it or where it needs to protect its service or another person from harm.
  • Our payment providers are the merchants of record for purchases, and each is named at checkout and on the receipt for the payment it takes. Each collects and holds payment details directly and is a controller in its own right for that information, not our processor.
  • Postmark delivers email. That includes account messages such as workspace invitations, and the lead notifications a Customer switches on, which carry a lead's contact details and a summary of their conversation.

Two more things happen to a photo the model provider describes. With each one it receives a code made from the visitor's account with a key only we hold, so that it can tell one visitor's activity from another's without learning who either of them is. It also scans each photo it describes for child sexual abuse material, and keeps any photo that scan flags for its own manual review.

Screening for harmful content. The model provider's moderation service reports whether a visitor's message or image contains harmful content, such as harassment, hate, violence, illegal activity, sexual content or self-harm. Screening is automated and can flag an ordinary message or image, so a flag is not a finding about the visitor. The Dashboard calls the screening of an image a safety check. We keep no separate copy of a screened message or image, only the result:

  • we mark a flagged message on the stored transcript with the categories it was flagged for, and leave every other message unmarked;
  • we record the result of screening, with any categories it was flagged for, on every stored image.

We use the result to keep a flagged image out of the Dashboard and out of anything a workflow forwards afterwards. The Dashboard does not show a mark on a message, and nobody is notified of a mark or of a flagged image.

A mark on a message, or the result recorded on an image, can record that the message or image suggested self-harm. POPIA and the GDPR treat that as special category information about the visitor's health, so a Customer's transcripts and files can hold special category information even when its bot never asks for any.

Each result is kept only as long as the transcript or the image it belongs to, apart from a line in our servers' operating logs naming the conversation or file and the categories, which our servers overwrite as the logs rotate. Erasing a contact under section 7 removes the results kept with that person's transcripts and files.

A flagged message is answered as normal. We keep a flagged image but withhold it: the Dashboard does not show it, its link no longer opens it, and a workflow that forwards its contents after it was flagged sends a notice instead. An image stays withheld until we find that it was flagged in error and release it. Screening runs after an image is stored, so a workflow that forwards an image as soon as it arrives can forward it before the result is known. Retention and the deletions in section 7 remove a flagged image in the same way as any other file.

Integrations you switch on with your own credentials are different. If a Customer connects ClickUp, that service receives lead information because the Customer told us to send it, under the Customer's own account. That is the Customer's own arrangement and the Customer is responsible for having a lawful basis for it.

Email notification is switched on the same way and is not the same thing. It runs on our Postmark account and sends from our address, so Postmark is our sub-processor rather than the Customer's.

9. Cross-Border Transfers#

Our infrastructure and several sub-processors are outside South Africa, in the United Kingdom and the United States. Sending personal information out of the country engages section 72 of POPIA, and out of the European Economic Area engages Chapter V of GDPR.

We transfer personal information to a recipient outside South Africa only where that recipient is bound by terms giving it protection substantially similar to POPIA, including a restriction on passing it on again. In practice that is the data processing agreement each provider on the sub-processors page publishes and we have accepted, which where the recipient sits outside the European Economic Area incorporates the European Commission's Standard Contractual Clauses.

We will not add a sub-processor that will not agree to those terms.

10. Your Rights#

Under POPIA you may ask us to confirm what we hold about you, give you a copy, correct it, or delete it where we have no lawful ground to keep it. You may object to processing based on legitimate interests. You may complain to the Information Regulator of South Africa.

Under GDPR, if it applies to you, you additionally have the right to restrict processing, to receive your data in a portable format, and to complain to your own supervisory authority.

How to exercise them. Email support@chatleadr.com. We will respond within 30 days. We may ask you to verify your identity first, because acting on an unverified erasure request is itself a breach.

If you are a visitor rather than a Customer, the business whose bot you spoke to has to answer your request, not us. Their name is usually on the website the bot was on. If you cannot work out who they are, write to us and we will pass the request to them and tell you we have done so.

Some things survive an erasure request, and the law allows this: billing records we are required to keep for tax purposes, and the suppression records described in sections 6 and 7.

11. Security#

  • Traffic to and from the service is encrypted in transit.
  • Stored files are encrypted at rest under a key held per workspace: uploaded documents, saved ad and story pictures, and photos kept from a conversation. One Customer's files cannot be read with another Customer's key.
  • Sign-in is delegated to Google Firebase Authentication, so we hold no passwords.
  • Each workspace's data is separated at the query level.
  • Access to production systems is limited to named administrators. There is currently one.

No system is immune. If a breach affects your personal information we will notify you and the Information Regulator as POPIA requires.

12. Children#

Chatleadr is sold to businesses and is not directed at children. We do not knowingly collect information from a child. A Customer who configures a bot to collect information from children is responsible for the consent that requires, and POPIA treats a child's information as a special category. A photo a visitor sends can show a child. Where photo understanding is on, a photo of a person is described to the bot only as a photo of a person, whatever their age.

13. Changes#

We will post any change on this page and update the Last changed date at the top. If a change materially affects a Customer's rights we will email them before it takes effect.