How Ignis handles data.
Last updated 14 November 2026. Operated by Ignis Intl LLC ("Ignis", "we"), 502 W 7th St STE 100, Erie, PA 16502, United States.
Ignis provides commerce and content software to merchants ("brands") who run their stores on Shopify. This page explains what data the Ignis platform processes, why, how we protect it, and how it is deleted. We do not sell personal data.
The short version
These rules apply to every feature and to every account a brand connects. The rest of this page gives the facts behind each one.
- We do not sell personal data.
- We act for the brand on its shoppers' data. It flows through Ignis only to power the features the brand switches on, and the brand decides how it is used. We do the same for the messages people send to a brand.
- A brand chooses what Ignis can reach. Someone at the brand connects each account. We act only on the Pages, profiles, channels, properties and accounts that person chose.
- We say what we read and what Disconnect deletes. For each connected account, the table under Accounts you connect to Ignis shows what we read, what Ignis does there and what Disconnect deletes.
- Disconnect works straight away. It deletes the account's token at once, with the other data that account's entry names. Copies in our backups age out about a month later.
- Messages from Facebook, Instagram and Threads are read, not stored. The Inbox reads the messages people send to a brand's Facebook Page and Instagram account, and the replies to and mentions of its Threads profile, live, so the brand's team can answer them.
- Google user data follows Google's Limited Use rules. We use it only for the features this page describes. We never use it for advertising or to train AI.
- AI training is the owner's choice. We train on a workspace's assistant records only if its owner switches training on. Data from Google, Slack messages and the messages the Inbox reads from Facebook, Instagram and Threads are never used to train AI.
- You can ask us for a copy or a deletion at [email protected], whether you are a brand or a shopper.
- This version takes effect on the date at the top. When a change affects how we handle personal data, we can require each person at a brand to accept it before they keep using Ignis.
Whose data this covers
Three groups, and our role for each:
- Brands. The merchant accounts that install and use Ignis (account, contact, billing, and store-configuration data). For this data we decide how it is used.
- Shoppers and visitors. The customers of those brands, and people who browse a brand's store without buying. Their data flows through Ignis only to power the features a brand has switched on, for example inventory sync, reviews, returns, loyalty and analytics. For this data we act for the brand.
- People who write to a brand. People who message a brand's own Facebook Page or Instagram account, or reply to or mention its Threads profile, where the brand uses the Inbox in Ignis, and people who post in the Slack channel a brand connects. For their messages we act for the brand.
Our legal basis. For shopper data we act for the brand, and the brand decides the legal basis. For a brand's own account data we decide, and we rely on: the contract with the brand, to provide and bill the service; our legitimate interests, to keep the service secure, prevent abuse, send the one setup reminder email and check and improve the service; our legal obligations, for tax and accounting records; and consent, where we ask for it, such as for training on a workspace's assistant records.
What we collect and why
- Order data (product IDs, variant IDs, quantities, order identifiers). Used to keep pack inventory accurate, credit loyalty points, and process returns. This, with the order detail below, is the data behind the "protected customer data" access the app requests. We request it for app functionality only.
- Order detail, where a feature needs it. For the features that cannot work without it, such as return labels, loyalty, order-level analytics and server-side conversion tracking, Ignis reads the full order from Shopify under the brand's own connection. That order carries the shopper's name, email address, phone number and addresses.
- Product and store configuration. Catalog, pricing rules, and per-brand module settings needed to render features in the store and dashboard.
- Reviews and returns content that a shopper voluntarily submits (name, email address, rating, review text, optional photos; return reason and shipping address for a return label), used only to provide that feature to the brand.
- Support messages a shopper sends through a brand's Ignis support form: their name, email address, order number if given, what they chose to write and any files they attach. To report a damaged item, a shopper first gets a one-time code by email to show the order is theirs.
- Site activity, where a brand has switched on Ignis analytics. Page views, product views, cart events, checkout steps, the page address, the referring page, the campaign tags and advertising click identifiers on the link the shopper arrived on, and browser details such as user agent, language, screen size and timezone offset, and the search terms the shopper enters. Each event is tied to a random visitor identifier. Where a shopper signs in, or types an email address or phone number into a store's newsletter or Klaviyo form, the theme embed sends us a SHA-256 hash of it, not the value itself. At checkout, the Ignis Analytics pixel sends the email address, phone number, name, city, state, postcode and country the shopper entered, and we hash them before we store them. An Ignis popup sends us the email address typed into it. Where the shopper signs in or places an order from that browser, we link the visitor identifier to their customer ID, email address and phone number.
- Data from accounts a brand connects to Ignis, such as its Facebook Page, Instagram account, Google Analytics, Google Merchant Center or Slack. What we read from each, why, and how long we keep it are set out under Accounts you connect to Ignis below.
- Messages people send to a brand's own social accounts, where the brand uses the Inbox in Ignis: private messages to its Facebook Page and its Instagram account, and replies to and mentions of its Threads profile. We read them to show them to the brand's team, and we do not store them. This is set out under Messages in the Inbox below.
- Brand account and billing data for authentication, support, and subscription management.
The shopper's IP address. We receive it, as any web service does, when a browser sends us a request. We use it to rate limit abuse, to let a brand exclude its own staff from its own figures, and, where a brand has switched on server-side conversion tracking, to send the conversion to Meta on that brand's behalf. We do not store a raw IP address in our analytics tables. When a shopper asks for a one-time code to report a damaged item, we keep the IP address with that code to limit abuse.
Data minimization. Where a feature does not need a shopper's name, email, phone or address, Ignis does not ask for it. The Ignis Shopify app's own order webhook, for example, forwards only the order identifier, the customer identifier, order totals and line items, with no contact details and no addresses. One exception: where Ignis analytics is on, we keep a visitor's email address, phone number and Meta's browser identifiers with the visitor identifier, so conversion tracking can match them if the brand turns it on.
Cookies and similar technologies
On this website (ignishq.com) we use none. This site loads no analytics and no advertising tags, and it sets no cookies. Our network provider, Cloudflare, adds one small script of its own to pages that show an email address, to help keep the address away from robots that collect addresses for spam. The pages load fonts from Google Fonts (fonts.googleapis.com and fonts.gstatic.com), so your browser requests those font files from Google, which receives your IP address and browser details as part of that request.
In the Ignis dashboard we keep your sign-in session in your browser's local storage so you stay signed in. We do not use it for advertising and we do not share it.
On a brand's storefront, Ignis places first-party identifiers in the shopper's browser. There are three ways this happens, and they are not the same:
- The Ignis Analytics pixel is off until someone at the brand turns tracking on, on the Tracking page of the Ignis app in Shopify. Shopify's own pixel manager will not load it at all unless the shopper has allowed analytics.
- The Ignis theme embed is off unless the brand switches it on in their theme editor. It runs on the storefront page itself, so it can also see marketing links, cart changes and form submissions.
- The Ignis popup is off unless the brand adds it in their theme editor. It keeps which popup a browser saw and any prize, and, where Ignis analytics is off, its own random visitor identifier.
These are what we set and why:
_pxa, a cookie and a matching local-storage entry, 400 days. A random visitor identifier so repeat visits by the same browser can be counted once. The theme embed renews it on each visit. Where the Ignis analytics endpoint is on the same site as the store, our server re-issues the same identifier for 400 days, because browsers cut short the lifetime of cookies set by a script. If a browser has lost_pxabut still has Shopify's own visitor cookie, our server gives back the identifier it had before. It contains no name, email address or phone number._pxs, a cookie, 30 minutes. A random session identifier, so a single visit is measured as one visit. The theme embed extends it with each event._pxl_excluded, a cookie and a local-storage entry, 10 years. Set when a browser opens the store through the exclusion link a brand gives its staff, so their visits do not distort that brand's figures. A device carrying this flag sends nothing at all._fbpand_fbc, cookies, 90 days. Meta's browser identifier and click identifier. Unless the shopper has declined marketing, Ignis reads them wherever they exist, keeps them with the visitor identifier, and can restore either from that record if the browser has lost it. Where the brand has set up Meta in Ignis, Ignis also writes_fbcfrom an ad click on the link. Where the brand has also chosen the browser-side option, Ignis creates_fbpif the browser has none, and the storefront loads Meta's own script fromconnect.facebook.net, which is Meta's code operating under Meta's terms._px_ftand_px_lt, local storage, until cleared. The first and most recent marketing source for this browser, taken from the link the shopper arrived on. This includes campaign tags and advertising click identifiers where the link carried them._px_buf,_px_attr_syncedand_px_blocked, local storage and session storage, until cleared. Working state, so events are not lost when a page closes and are not counted twice.- _px_id, local storage, until cleared. The SHA-256 hashes of an email address or phone number the shopper entered, kept so later visits from this browser can be matched.
- _pxl_ic, a cookie and a cart attribute, 15 minutes. Marks a checkout started from this browser, so the order can be matched to it.
- _px_cart, _px_ic_mark, _px_search_typed and _px_wallet_start, local storage and session storage, until cleared. Working state for cart, checkout and search events.
- _pxl_test, a cookie and a local-storage entry, until the time the link sets. Set only from a signed link a brand's admin opens, to mark their own device as a test device.
- _pplab_anon and the other _pplab_ entries, local storage until cleared, and one session storage entry. Set by the Ignis popup: which popup a browser saw, whether it finished, snoozed or dismissed it, any prize, and a random visitor identifier where Ignis analytics is off.
Where the theme embed is on, Ignis also writes those identifiers onto the brand's own Shopify cart as cart attributes, so an order can be matched to the visit that produced it.
Consent, described exactly. The Ignis Analytics pixel is not loaded by Shopify unless the shopper has allowed analytics, so no analytics consent decision of ours is involved. The theme embed reads the shopper's choice from Shopify's Customer Privacy API and re-reads it whenever the shopper changes it. Where that shopper has declined analytics, the embed sends no events once it has read that choice. Where Shopify's Customer Privacy API is not on the page, or does not load within about a second, the embed treats analytics as allowed, so a brand using the theme embed should make sure their consent banner is in place. The brand is responsible for the consent banner on their own storefront and for their own cookie notice. This page is here so a brand can publish an accurate one.
A shopper can clear all of this at any time by clearing cookies and site data for that store in their browser.
How we protect it
- Encryption in transit. All traffic uses TLS/HTTPS; server-to-server calls are additionally authenticated, most with an HMAC-SHA256 signature and the rest with a secret both sides hold, and are rejected if that check fails or its secret is not configured.
- Encryption at rest. Data is stored in a managed Postgres database that encrypts data at rest, and files are stored in object storage that does the same. On top of that, the third-party credentials the Ignis API stores are encrypted by Ignis itself with AES-256-GCM before they are written. The Shopify app's own access tokens for each store are not yet encrypted this way, and rely on the database's encryption at rest. That extra layer covers credentials. Other data is protected by the database's own encryption at rest and by the controls below.
- Tenant isolation. Most of each brand's data is separated at the database level with row-level security. The rest, such as what our public storefront endpoints write, is separated by the brand's ID on every query.
- Least-privilege access. Access to production data is limited to the personnel who operate the platform, and, by default, application code runs under a restricted database role that row-level security applies to. Migrations, jobs that work across brands and some requests use the owner role instead.
- Purpose limitation. Data is used only for the feature the brand switched on, and is never resold. Where a brand enables server-side conversion tracking, the data that feature sends is described under Who we share it with below.
No software or hosted service can be guaranteed secure, and we do not claim otherwise. We hold no security certification, and this page names only measures that are actually in place.
Artificial intelligence
Several Ignis features use AI. Here is what happens to data in them.
- We record assistant conversations. Each turn of a brand's Ignis assistant is recorded: which tools it used, short extracts of what it asked them and what they returned, the model, what it cost, how it ended, and any rating a person gave it. When a tool returns data from Google, the record keeps no extract of it. Email addresses, card numbers, credentials, and phone numbers and street addresses in their usual written forms are removed before the record is stored. A name written in free text, or a phone number written as one run of digits, is not reliably removed. We use these records to run, check and improve the assistant, keep them for 180 days, and delete them when a brand's store is redacted or its workspace is deleted. A brand can ask us to stop recording its workspace at [email protected].
- We train on those records only with the owner's consent. Training is off unless the owner of a workspace switches it on in Settings. Only records made while it is on can be used, and switching it off stops any further use. For a tool that read a customer's record, the copy kept for training holds only the tool's name and the names of its inputs, not the customer's details. Unless the owner switches training on, no assistant record of that brand is used for training.
- Data from Google is not used to train AI. Ignis does not use the data it receives from Google to train AI models. When the assistant uses a tool that reads Google data, such as Google Analytics or Merchant Center figures, that record and every later record of the same conversation are left out of training, even when the owner has switched training on. Content a brand saves in Ignis, such as a note or a brief, is the brand's own content, even if it was written while looking at Google figures. If the brand uses that saved content in a later conversation, and the owner has switched training on, that later conversation can be used for training like any other.
- Slack messages, and messages the Inbox reads from Facebook, Instagram and Threads, are not used for AI. No AI feature in a brand's Ignis workspace reads the messages copied from its Slack channel or the messages the Inbox reads from those platforms, so they are never used to train AI. Conversations a brand's team records by hand in the Inbox are different, as set out under Messages in the Inbox.
- We use third-party AI providers, named in the list below. When a brand uses an AI feature, the prompt and any files that feature needs are sent to the relevant provider so it can produce the output the brand asked for.
- What a provider may do with a request is set by that provider's own terms, and those terms differ between them. We send every request through our own account with each provider. If this matters to your business, read the terms of the provider behind the feature you use.
- Providers keep a short-term copy. AI providers generally retain a request for a limited period to detect and prevent abuse, and may have staff review content their systems flag as a possible policy breach. The providers we use publish periods of up to 55 days for this. That is their retention, not ours.
- If a brand uploads photographs of a real person, that brand is responsible for having that person's permission. Our terms say the same thing.
- AI output can be wrong. We say so plainly in our terms, and they make the brand responsible for checking AI output before it reaches a customer or an ad platform. Ads and emails always wait for a person. A brand that switches on the Planner's autopilot lets organic posts the assistant drafted go out on their own.
We record how much AI a brand used, so we can bill it and show them their spend. That record holds the provider, the model name, the operation, the amount of work and the cost.
Who we share it with
We do not sell personal data. We share it only with the providers we need to run the features a brand has switched on, and only for the purpose named. This is the current list.
Always used, because they run the platform:
- Google Cloud, application hosting (Cloud Run, in the us-west1 region in Oregon, United States) and the platform's own keys. Processes everything the platform handles.
- GitHub, our code host. The daily database backup is made on GitHub's servers on its way to our file storage, so it passes through them.
- Neon, managed Postgres database. Stores everything the platform stores.
- Cloudflare, our network and file storage. Every request to this website and to our API passes through Cloudflare's network. Its R2 storage holds review photos, files attached to a support message, return labels, images and videos a brand uploads or generates, videos creators upload, images shared in team chat, copies of assistant records, and database backups.
- Upstash, a data store for working data, such as the counters that limit how often a request can be made, and the Inbox's new-message notes described under Messages in the Inbox.
- WorkOS, sign-in for brand accounts. The email address and name of the brand's staff. No shopper data.
- Vercel, hosting for this website and our documentation. Static pages only. It receives the IP address and browser details every web request carries.
- Stripe, payments and creator payouts. Brand billing details and creator payout details. No shopper data.
- Resend, email from Ignis to the people who use it, and to our own team. The email address and name of a brand's staff, to send account, billing and usage emails and the daily summary a brand asks for. When someone asks for access to the Ignis dashboard, their name, email address, company and note, in a notice to our own team. When a creator asks a brand for samples, the creator's name or email address and their shipping address, in a notice to our own team. We never send email to a brand's shoppers through our own Resend account.
- ntfy.sh, push alerts. Alerts to our own team about the platform, and, for a brand that sets up its own alerts, alerts about its own store, such as its daily sales summary.
Used only for the feature named, and only when a brand switches it on:
- EasyPost, return labels, bought on the brand's own EasyPost account. The shopper's name, shipping address and phone number.
- OpenAI, text generation, and the assistant where it runs on an OpenAI model. Support message content, so the brand can be offered a drafted reply, the brand's own prompts and product text, the images the brand uploads or attaches, voice recordings a brand's team makes to give an instruction, text for voiceovers, and the store data the assistant reads to answer.
- Anthropic, the assistant where it runs on an Anthropic model. The brand's prompts and the store data the assistant reads to answer.
- fal.ai, image and video generation. The brand's prompts and the images, videos and sound the brand uploads.
- Google, image generation on the routes where Ignis is configured to use it, and the assistant where it runs on a Google model. The brand's prompts, the images the brand uploads, and the store data the assistant reads to answer.
- Klaviyo, where the brand connects it. The email address, and for text sign-ups the phone number, a shopper gave that brand, with events such as a review they wrote (first name, rating, title and text) or a code they won.
- Meta, where the brand connects Facebook, Instagram or Threads. The replies and posts the brand's team sends, the ads the brand confirms, and, with server-side conversion tracking switched on, the data in the paragraph below.
- Google Ads and TikTok, where the brand connects those accounts. Ignis reads the brand's own Google Ads campaign and spend figures, and, where Ignis offers it, creates or switches on a Google Ads campaign only when the brand confirms. On TikTok it reads the account's profile and video figures, and can publish content the brand made to that account. We do not send shopper personal data to either.
- Pinterest, LinkedIn, X and Snapchat, where the brand connects those accounts. Ignis reads the account's name, and, where Ignis supports it, reads that account's own figures and publishes what the brand schedules. It changes the brand's own ads there only when the brand confirms. We do not send shopper personal data to them.
- Google Analytics and Google Merchant Center, where the brand connects them. Ignis reads the brand's own analytics figures and product listings, and changes listings only when the brand confirms. We do not send shopper personal data to either.
- Slack, where the brand connects its own Slack workspace. The messages, alerts and approval requests Ignis posts to the channel the brand picks, and the messages people send back in that channel.
- An AI app the brand connects, such as Claude. What the person who connected it allowed that app to read. Covered under AI apps you connect to Ignis below.
Shopper details and AI. Unless a brand switches on sharing in Settings › AI & privacy, a shopper's name, email address, phone number, order number and tracking details are removed before support content is sent to an AI provider. Free text a shopper wrote can still hold personal details, such as a name typed into a message, and we do not claim to remove those.
Store email to a brand's shoppers, such as return labels and verification codes, goes out only through a mail account the brand connects: its own Resend account, its own SMTP host or its own Gmail account. Ignis writes each email and hands it to that account.
When a brand connects Meta and turns on server-side conversion tracking, we send order and event data to Meta on the brand's behalf and on the brand's instruction. Identifiers such as email address, phone number, name, city and postcode are hashed before they are sent. The shopper's IP address, browser user agent and Meta's own browser cookie identifiers are sent as they are, because Meta requires that form. This happens only for brands that have connected Meta and turned this on. It is never sent for a shopper who declined tracking. For an order, it is not sent in a country that requires consent first unless the shopper gave it. Other storefront events follow the choice the store's consent banner passes to Ignis. Where no choice reaches us, they are sent, so the brand's consent banner has to be in place. For a shopper who opted out of sale or sharing, it is sent only with Meta's limited-data-use flag, or not at all, as set out under California below. Server-side conversion tracking to any other advertising platform is not something Ignis does today.
We also disclose personal data where the law requires it, and to a buyer of our business if Ignis is ever sold, in which case this page is updated first.
Changes to this list. We keep this list current on this page, and we update it before adding a new provider that handles personal data. The date at the top changes when we do. A brand who wants to be told directly can ask us to email them when this list changes, at [email protected].
Accounts you connect to Ignis
Section added 6 November 2026.
A brand can connect accounts it holds on other platforms. Most connections are made by someone at the brand signing in to that platform and approving the access Ignis asks for. A few can instead use a key the brand pastes in, as set out below. Ignis then acts on that account only for the features the brand uses, and only on the Pages, profiles, channels, properties and accounts that person chose. On none of these platforms do we read a personal profile, a friends or followers list, or an email inbox. We read private messages in one place only: the Inbox in Ignis, for the brand's own Facebook Page and Instagram account, set out under Messages in the Inbox below. We also read the messages people post in the one public Slack channel the brand picks, set out under Slack below.
Tokens, and how long we keep them. A connection gives us an access token, and on some platforms a refresh token. We encrypt them with AES-256-GCM, as described above, and keep them, together with the account details listed below, only while the account stays connected. When the brand presses Disconnect in Ignis under Settings › Connectors, or removes Ignis in the platform's own settings and the platform tells us, we delete them straight away. A few business records stay: the Facebook Pages and Instagram accounts the brand turned on in Ignis or chose as its default for posting, the ID and currency of an ad account whose spend the brand's reports used, and ad figures and launch records already in the brand's reports. Our Data Deletion page lists them. Copies in our backups age out about a month later. Step-by-step instructions for every platform, and what we keep afterwards, are on our Data Deletion page.
At a glance. What we read from each platform, what Ignis does there, and what pressing Disconnect deletes. The entries below the table give more detail where a platform's rules ask for it.
| Platform | What we read | What Ignis does there | What Disconnect deletes |
|---|---|---|---|
| Facebook and Instagram | The names and IDs of the Pages, Instagram accounts, ad accounts and datasets the person manages. Ad performance and Instagram account figures. Their messages, read in the Inbox and not stored. | Publishes the posts the brand schedules. Creates the ads the brand builds, paused, and turns them on, pauses or changes them only when the brand confirms. Sends the brand's conversion events to its own dataset. Sends the replies a person on the brand's team writes. | The token and the account details. The Pages and Instagram accounts the brand turned on or chose as its default stay, and so do the ad account's ID and currency and the ad figures already read. |
| Threads | The user ID and username. Post and profile figures. Replies and mentions in the Inbox, read and not stored. | Publishes the posts the brand schedules. Sends the replies a person on the brand's team writes, and hides a reply when they choose. | The token and the account details. |
| TikTok | The display name, username and posting options. The account's public videos and their view, like, comment and share counts. | Uploads the videos the brand schedules. | The token and the account details. Cancels our access at TikTok. |
| YouTube | The channel's ID, title and picture. When the brand asks, the channel's own figures, held for up to 30 minutes and not stored. | Uploads the videos the brand schedules. | The token and the channel details. Cancels our access at Google. |
| Gmail | The address emails are sent from. | Sends the brand's store emails. It cannot read the mailbox. | The token and the address. Cancels our access at Google. |
| Google Ads | The accounts the person can access, and the chosen account's campaign and spend figures. | Where Ignis offers it, creates the campaigns the brand builds, paused, and switches one on only when the brand confirms. | The token and the account address. Cancels our access at Google. The Ads account's ID and currency, and the record of campaigns launched from Ignis, stay. |
| Google Analytics | Daily traffic reports for the chosen property. They are totals, not records of individual visitors. | Nothing. The access is read-only. | The token, the account address and every report we stored. Cancels our access at Google. |
| Google Merchant Center | The chosen account's products, their status and issues, and its data sources. | Sends a fix only after the brand turns fixes on and approves that fix. | The token, the account address, the latest read, the issues and every suggested fix. Cancels our access at Google. |
| Slack | The lists of its public channels and people, only to set up, and not stored. The messages in the one public channel the brand picks. | Posts team chat, agent updates and approval cards to that channel. | The bot token, every copied message, the member links, the button records, and the Slack IDs and channel name. Cancels our access at Slack. |
| Klaviyo | Lists, segments, profiles, metrics, campaigns, flows and templates. | Adds the profiles and events the store collects. Creates the campaigns, templates and images the brand builds. Schedules a campaign to send only when a person confirms. | The token. Cancels our access at Klaviyo when the brand connected by signing in. |
| Pinterest, LinkedIn, X and Snapchat | The account's name, and where Ignis supports it, that account's own figures. | Publishes what the brand schedules, where Ignis supports posting there. Changes the brand's own ads there only when the brand confirms. | The token. For Snapchat, the ad account records and figures already read stay. |
Facebook and Instagram
We read the names and IDs of the Facebook Pages, the Instagram professional accounts linked to them, the ad accounts and the datasets that person manages, so the brand can choose which ones Ignis uses. With that access we publish the posts the brand schedules to its Page and Instagram account, create the ads the brand builds in Ignis (created paused, so nothing spends until the brand turns it on), turn them on, pause or change them only when the brand confirms, read the performance of the brand's ads, read the figures Instagram gives for the brand's own account, such as reach, views and follower count, and send the brand's conversion events to its own dataset. Where the brand uses the Inbox, we also read and answer the messages people send to its Page and Instagram account, as set out under Messages in the Inbox below. You can remove Ignis in Facebook under Settings & privacy › Settings › Business integrations. Facebook then tells us and we delete the connection.
When the brand presses Disconnect or removes Ignis in Facebook, or Facebook passes us a request to delete its data, we delete the token and the account details. The Facebook Pages and Instagram accounts the brand turned on in Ignis or chose as its default for posting stay, and so do the ad account's ID and currency and the ad figures already in the brand's reports. The Pages' and accounts' names and IDs stay until the brand asks us to delete them, at [email protected].
Meta Conversions API. Server-side conversion tracking, described under Who we share it with above, runs either on the Meta connection or on a dataset ID and access token the brand pastes in from Meta Events Manager. A pasted token is stored and protected like every other credential, and deleted when the brand removes it in Settings › Connectors.
Threads
We read the Threads user ID and username, publish the posts the brand schedules to that Threads profile, and read the views, likes, replies, reposts and quotes of the brand's own Threads posts and profile. Where the brand uses the Inbox, we also read the replies to its Threads posts and the posts that mention its Threads profile, as set out under Messages in the Inbox below. You can remove Ignis in Threads under Settings › Account › Website permissions. Threads then tells us and we delete the connection.
TikTok
We read the TikTok display name and username, the posting options TikTok allows the account, and the account's public videos and their view, like, comment and share counts, and upload the videos the brand schedules to that TikTok account with the privacy and interaction settings the brand chooses for each one. You can remove Ignis in TikTok under Settings and privacy › Security and permissions › Apps and services permissions. TikTok then tells us and we delete the connection.
- YouTube. Ignis uses YouTube API Services. We read the YouTube channel's ID, title and picture, and upload the videos the brand schedules to that channel. When the brand asks, we can also read the channel's own figures, such as subscribers, views, likes and watch time. We hold them for up to 30 minutes and do not store them. By using Ignis's YouTube features you agree to be bound by the YouTube Terms of Service, and Google's handling of your data is governed by the Google Privacy Policy. You can revoke Ignis's access at any time on Google's security settings page, security.google.com/
settings/ . Once a day, Ignis checks with Google whether each YouTube connection still works. While it does, that check refreshes the channel's title and picture. If you have removed Ignis's access, the check records that. It does not delete anything. To have the channel ID, title, picture and token we stored deleted, press Disconnect in Ignis, or ask us at [email protected].security/ permissions - Gmail. A brand can send its store's emails, such as return labels, verification codes and support replies to its shoppers, from its own Gmail address. For that we ask Google only for permission to send email and for the address it is sent from. We cannot read, search, change or delete anything in the mailbox. What we receive with this permission is used only to send the brand's email and to show which Google account is connected. It is never used to develop, improve or train AI or machine-learning models. You can revoke this at security.google.com/
settings/ or by pressing Disconnect in Ignis.security/ permissions - Google Ads. Where a brand connects it, we read the Google Ads accounts that person can access and the campaign and spend figures of the account the brand chooses, and, where Ignis offers it, create the campaigns the brand builds in Ignis, paused until the brand turns them on. The Ads account's ID and currency, and the record of campaigns launched from Ignis, stay after Disconnect.
- Google Analytics. We ask Google for read-only access. We read the Google Analytics accounts and properties that person can see, so the brand can choose one, and that property's name, time zone and currency. Then we read the property's traffic reports. They give sessions, users, engaged sessions, key events, purchases, purchase revenue and page views for each day. Each report is split by channel, source and medium, landing page, device or page. The first read covers the last 30 days. After that we read the latest days once a day. We keep the top 25 rows of each split for each day. When the brand or its assistant in Ignis asks a question, we can also read one extra report. It is split by country, campaign, source, medium, or new and returning visitors. We hold it for 10 minutes. These are totals, not records of individual visitors. We use them to show the brand where its visitors come from, which pages lose them and which channels lead to sales, next to its orders and ad spend. We change nothing in Google Analytics. When the brand presses Disconnect, we delete the token and every report we stored, in the same step.
- Google Merchant Center. We ask Google for Merchant Center access. It is the only permission Merchant Center offers, and it allows changes, so Ignis makes none unless the brand approves them as described here. We read the Merchant Center accounts that person can open, so the brand can choose one. Once a day, and when the brand asks for a refresh, we read that account's products and their approval status. We also read the issues Google found with each product, the account's own issues, and its data sources. We keep the latest read: the counts, the issues, and the ID and title of each product with an issue. We use it to show the brand which products Google is not showing and why, and to suggest fixes taken from the store's own Shopify data. Ignis changes the account only after the brand turns on "Let approved fixes update Merchant Center" and then approves each fix. A fix adds one data source named "Ignis product data" and links it to the main data source that holds the product. It sends only the price, availability, colour, size or material that the fix changes. We never create a main product feed, add a product, or set a product's gender or age group. When the brand presses Disconnect, we delete the token, the latest read, the issues and every suggested fix, in the same step. The "Ignis product data" source and the fixes already sent stay in the brand's Merchant Center.
Google user data. Ignis asks Google for these permissions, each one only when a brand connects that product:
- Google Analytics:
analytics.readonly. - Google Merchant Center:
content. - Google Ads:
adwords. - Gmail:
gmail.send. - YouTube:
youtube.uploadandyoutube.readonly.
With each of these except YouTube, we also ask for openid and email. They tell us the address of the Google account that made the connection, so the brand's team can see which account is connected. We keep that address with the connection and delete it on Disconnect.
Ignis's use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements. In particular:
- We use the data we receive through these permissions only to provide the features the brand uses and that this section describes.
- We do not sell it. We do not use it for advertising, and we do not use it to decide credit or lending.
- We do not use it to train AI models, as set out under Artificial intelligence above.
- We pass it on only to provide a feature the brand asked for, for security, or to comply with the law. Two features pass it on. When the brand asks its assistant in Ignis a question, the data needed for the answer goes to the AI provider that runs the assistant, named under Who we share it with, so it can produce the answer. When the brand connects an AI app to Ignis, that app can read the same figures, as set out under AI apps you connect to Ignis.
- What the assistant read in a conversation stays in that conversation until the person who started it deletes it.
- If Ignis is ever sold, we ask the brand first before any Google user data is transferred.
- Nobody at Ignis reads it unless the brand asks us to in a support request, it is needed for security or to investigate abuse, or the law requires it.
Slack
A brand can run its Ignis team chat in Slack. The brand adds Ignis to one Slack workspace as a bot and picks one public channel for it. To set this up, we read the names of the workspace's public channels and the display names of its people. The brand uses them to pick its channel, and each team member to find themselves. We do not store those lists.
The bot posts to the picked channel: what the team writes in the matching Ignis channel, updates from the brand's AI agents in Ignis, and approval cards. An approval card shows counts and IDs, never a shopper's name, address or message. We copy the messages people post in the picked channel into the matching Ignis channel, so the team sees one conversation. We clean each message and cut it at 4,000 characters before we store it. If someone adds the bot to another channel, Slack sends us those messages too, and we drop them without storing them. We do the same with edits, join notices and posts by bots. We do not record message text in our logs. A team member can link their Slack account, so they can answer their own approval cards from Slack. We then keep their Slack user ID until they unlink it in Ignis or Slack is disconnected. Approval and link buttons work once and expire within a day. We cannot read private channels or direct messages. No AI feature in a brand's Ignis workspace reads the Slack messages, and neither does an AI app a brand connects to Ignis, so they are never used to train AI.
We keep the bot token, the workspace ID and name, the bot's IDs and the picked channel. When the brand presses Disconnect, or removes the Ignis app from its Slack workspace and Slack tells us, we delete straight away: the bot token, every message we copied from Slack, the links between Slack people and team members, the records of approval and link buttons, and the Slack IDs and channel name we stored. Pressing Disconnect also cancels our access at Slack. The matching Ignis channel stays, with the messages the team wrote in Ignis and the updates from its agents. While Slack stays connected, anyone in that Slack workspace can ask the brand, or write to us at [email protected], to have their copied messages deleted.
Klaviyo
We read the brand's lists, segments, profiles, metrics, campaigns, flows and templates, add the profiles and events the brand's store collects, and create the campaigns, templates and images the brand builds in Ignis. Ignis schedules a campaign to send, or changes its send time, only when a person on the brand's team confirms it. When the brand connected by signing in, pressing Disconnect in Ignis also cancels our access at Klaviyo. A private key the brand pasted in stays valid until the brand deletes it in Klaviyo.
Messages in the Inbox
Where a brand connects Facebook and Instagram, or Threads, its team can read and answer messages in the Inbox in Ignis.
What we read.
- Private messages people send to the brand's own Facebook Page, through Messenger, and to its own Instagram professional account. For each message we read the sender's name or username, their ID on that platform, the text, any attachments and the time. We also read the brand's own replies in the same conversation.
- Public replies to the brand's own Threads posts, and public Threads posts that mention the brand's Threads profile. For each one we read the author's username, the text, the time and its link.
Why. So the brand's team can see and answer its customers' messages in one place. Ignis sends a reply only when a person on the brand's team writes it and sends it. On Messenger and Instagram, a reply can only be sent within 24 hours of the customer's last message. On Threads, a person on the brand's team can also hide a reply to one of the brand's posts.
Who sees them. Only people signed in to that brand's Ignis workspace. An API key cannot read them. No AI feature in Ignis reads them, and neither does an AI app a brand connects to Ignis, so they are never used to train AI. Nobody at Ignis reads them unless the brand asks us to in a support request, it is needed for security or to investigate abuse, or the law requires it.
How long we keep them. We do not store the messages. Each time the brand opens the Inbox, Ignis reads them from Facebook, Instagram or Threads and shows them without keeping a copy. When a new Facebook or Instagram message arrives, Meta sends us a notice that includes it. From that notice we keep only a short note that the customer wrote: the platform, the sender's ID and the time. We never keep the text or the sender's name. A note is deleted when the brand opens that conversation. We keep at most the 200 newest notes for a brand, and all of them are deleted 7 days after the brand's last new message. We also record who on the brand's team opened or answered a conversation, and when. That record holds IDs and counts, never the text.
How it is deleted. Because we keep no copy of the messages, there is nothing of them to delete in Ignis. The messages stay on Facebook, Instagram or Threads, under those platforms' own rules. When the brand presses Disconnect, or removes Ignis in Facebook or Threads, Ignis can no longer read them, and the new-message notes are gone within 7 days. If Facebook still sends us a new-message notice for that Page or Instagram account, we drop it without keeping anything.
Conversations the brand's team records by hand in the Inbox are different. We store them with the brand's other support conversations, and the brand can ask AI to analyze one or suggest a reply.
AI apps you connect to Ignis
A brand can connect an AI app it already uses, such as Claude, to its Ignis workspace. The app can then use Ignis's tools inside a conversation, for example to read the brand's figures, products and content, or to make content in Ignis. There are two ways to connect:
- Our hosted connector at
mcp.ignishq.com. A person at the brand signs in to Ignis and approves the app on Ignis's own screen. - The Ignis MCP server on the brand's own computer. The brand creates an API key in Ignis and gives it to the server.
What a connected app can see and do. When a person approves an app, they choose what it may do:
- Always: see the workspace's analytics, content, products, inventory and job status. This does not include shoppers' personal details. It does include the figures Ignis reads from the brand's connected accounts, such as Google Analytics, Google Ads and Merchant Center.
- Only if the person allows it: create and edit content, assets, briefs, video edits and Planner drafts.
- Only if the person allows it: read customer details, such as orders with names, returns, sign-ups, support conversations and reviews.
- Only if the person allows it: post, schedule and send to the brand's channels and customers. No Ignis tool uses this permission yet.
An app can never do more than the person who approved it can do in Ignis. It can never change billing, payouts, store credit, team members, roles, API keys, workspace settings, privacy settings or stock. On our hosted connector, tools that cost credits, and some that cannot be undone, such as generating an image or deleting a project, wait for that same person to say yes. They answer in the app or on a confirmation page in Ignis.
The app is yours. A connected app runs on the brand's own account and plan with the company that makes it, such as Anthropic for Claude. Ignis does not run or pay for the app's AI model. What the app reads from Ignis goes to that company under the brand's account and that company's terms, not ours. That includes shoppers' details, if the person allowed customer access. Connecting an app is the brand's instruction to us to give it what the person allowed. Ignis receives only the requests the app sends to Ignis's tools, including any details the app puts in them. It does not receive the rest of the conversation.
What we keep. For each connected app we keep who approved it, for which workspace, the app's name, the permissions granted, and when it was approved, last used and disconnected. An access token lasts 1 hour. A refresh token lasts 30 days and is replaced each time it is used. We store only a one-way hash of each token, never the token itself. When an action needs a yes, we keep the tool's name, the details the app sent for it and the answer. Our hosted connector keeps no copy of a conversation or of a request's details. Its log records the address called, the result and the time taken. What the app makes in Ignis stays in the brand's workspace, like anything else the brand makes there. A brand can ask us to delete these records at [email protected].
Files on the brand's computer. The Ignis MCP server on the brand's own computer reads a file in one of two cases: the person confirms that file, or the file is in a folder the brand named for it. It reads a file only to check it and send it to the brand's Ignis workspace.
Disconnecting. A brand can see its connected apps in Ignis under Settings › Connected AI apps and disconnect any of them. That ends the app's access and every token it holds at once. An app can also end its own access. For the server on the brand's own computer, deleting its API key in Ignis ends its access. What an app already read stays with the company that runs it, under that company's terms.
Where your data is processed
Ignis is a United States company. If you are in the United Kingdom, the European Economic Area or Switzerland, your personal data leaves that region when it reaches us.
We do not claim a transfer mechanism we have not put in place. A brand can enter into the European Commission Standard Contractual Clauses with us for that transfer, with the UK International Data Transfer Addendum where the UK is involved, as part of our data processing terms, and can ask us which transfer terms apply with each provider named above. Write to [email protected] and we will send you the current position.
Where a brand connects a provider of their own, such as their own advertising account or their own marketing platform, that provider's own location and terms apply to what the brand instructs us to send there.
How long we keep it and how it's deleted
How long we keep data depends on the brand's clean-up settings, the requests below, and what the law requires. Ignis honors Shopify's mandatory data-protection requests:
- Customer data request. We assemble a record of the data we hold for a specific shopper, such as their reviews, returns, support conversations, store credit, loyalty, saved items, signup and consent records and orders, and provide it to the brand, who is the one that received the request. It does not include the text of their support messages or the pages they visited.
- Customer redaction. We delete a specific shopper's data when Shopify sends a redaction request, or remove their identity from a record that has to survive for the reasons given below. What the shopper wrote in a support message, and the photos on a review that stays published, stay with the record. A copy we made for an earlier data request is kept until clean-up removes it.
- Shop redaction. When a brand uninstalls the app and Shopify sends a shop-redaction request, we delete that store's shopper data: reviews, returns, support conversations and messages, one-time codes, store credit and loyalty records, visitor tracking and identity records, signup-form captures, and our copy of the brand's order list.
Shop redaction removes the shopper data listed above. Our log of earlier customer data requests, including the exports in it, is kept. It deliberately leaves the brand's own Ignis workspace in place, meaning their account, their team, their settings and the content they made, because a brand's account can outlive one Shopify install and deleting it would destroy work that is not customer data. A brand who wants that removed as well can ask us at [email protected].
Retention periods
Each brand chooses its own retention periods in their Ignis settings and switches automatic clean-up on. Until a brand turns clean-up on, these periods do not apply, and shopper data stays until the brand deletes it or a redaction request above removes it. Clean-up also has to be switched on by us, per environment, before any of it runs. A brand can see in their settings which of the two applies to them, and nothing is removed until it says clean-up is running. When both are on, a daily job applies that brand's periods. These are the periods a brand can set, the default we suggest for each, and exactly what happens at the end of one:
- Visitor tracking. Default 13 months. The record linking a visitor to their email address, phone number or customer account is deleted. Those identifiers and the browser user agent are also removed from stored events. Traffic, order and revenue figures are not affected.
- Shopper records. Default 2 years. The name, email address, customer id and any stored shipping-label link are removed from reviews, returns, support conversations and our copy of the brand's order list. Those records themselves stay, so the brand keeps its reviews, its returns history and its accounts. Saved items are different: a saved item is only a shopper and a product, so the whole record is deleted rather than stripped, counted from when the shopper last saved that item. An open return or an open support conversation is never touched, no matter how old it is.
- Reviews the brand rejected. Same 2 year period. A review the brand chose not to publish is deleted outright, and any photos on it are deleted from our file storage at the same time.
- Signup form captures. Default 2 years. The email address captured by a popup or signup form is removed from the prize record. The prize and the discount code stay. A gift that was promised and not yet sent is never removed.
- One-time codes. Default 30 days. Expired sign-in and verification codes, which hold an email address and an IP address, are deleted.
A brand can set any of these to a longer period, up to ten years, or to "keep forever". Each one also has a shortest period we allow, so a brand cannot set clean-up so aggressive that it removes data the service still needs. Those minimums are 30 days for visitor tracking, 90 days for shopper records and signup captures, and 1 day for one-time codes. Separately from the periods, an open return, an open support conversation or a gift not yet sent is never touched at any age. That protection comes from the record's own state, not from the minimum period.
Clean-up never removes store credit or loyalty balances. They are a financial record of what a brand owes a shopper, and deleting one would not cancel the obligation, it would only destroy the brand's record that it exists. When a shopper asks to be erased, we remove the identity from the store credit record and keep the amount, so the shopper can no longer be linked to it and the brand's accounts stay correct. A loyalty points balance is removed on that request, because a points balance is not a debt in money. The history of how points were earned and spent stays, with the shopper's identity taken off it, so the brand's own records still add up.
Backups. We take a backup of the database every day, store it in encrypted object storage, and delete each one after about a month. When a record is deleted, whether by a retention period, a redaction request or a brand uninstalling, it is removed from the live system straight away and then ages out of the backups about a month later. We do not restore a backup in order to bring deleted personal data back. If we ever have to restore one to recover from a failure, we run again the store deletions and clean-up made since that backup.
Other records we keep. Assistant records are kept for 180 days, as described under Artificial intelligence. The record of which version of our terms each person accepted, and when, is kept for at least 6 years after the workspace is deleted, because it is the record of a contract. It holds no IP address.
Your rights
If you are a brand. You control which modules are active. You can change your retention periods at any time in your Ignis settings. You can disconnect Ignis by uninstalling the app, which starts the deletion described above. You can ask us for a copy of your account data, for a correction, or for deletion, by writing to [email protected].
If you are a shopper. Depending on where you live, you may have the right to ask for a copy of the personal data held about you, to have it corrected, to have it deleted, to receive it in a portable form, to object to or restrict certain processing, and to withdraw a consent you previously gave. You will not be treated differently for exercising any of these rights.
The brand you shopped with is the one who decides how your data is used, so the fastest route is to ask them. Their request reaches us through Shopify's own compliance channel and we act on it.
You can also come to us directly. Write to [email protected] and tell us which store you shopped with and the email address you used there. Because we hold your data on that brand's behalf, we will verify the request and then either action it or pass it to that brand to confirm, and we will tell you which we did. We answer within one month of receiving a request we can verify, and we will tell you if we need longer than that.
We may need to keep some records even after a deletion request, where the law allows or requires it. Store credit is the clearest example, and it is described above.
If you are in the United Kingdom or the European Economic Area and you are not satisfied with how we handled your request, you can complain to your national data protection authority.
California and other US state privacy rights
If you live in California, or in another US state with a comprehensive privacy law, you have rights to know what personal information is collected about you, to have it deleted, to have it corrected, to receive a copy, and to opt out of its sale or sharing. You will not be discriminated against for using them.
We do not sell personal information, and we do not share it for cross-context behavioural advertising for our own purposes.
Where a brand has connected Meta and switched on server-side conversion tracking, we send that brand's order and event data to Meta on the brand's instruction. Under California law that can count as "sharing" by the brand. The brand decides whether it happens, and the brand is responsible for offering their shoppers the opt-out. Ignis applies the choice a shopper makes in the store's Shopify privacy settings: where a shopper has opted out of the sale or sharing of their data, we either stop sending person-linked data for that shopper, or send it with Meta's own limited-data-use flag set, so the conversion still counts and Meta limits what it does with the person-linked fields.
To exercise any of these rights with us directly, write to [email protected]. An authorised agent may make a request for you if they give us proof that you authorised them.
Children
Ignis is business software sold to brands. It is not directed to children, and we do not knowingly collect personal data from anyone under 16. Where a brand's storefront reaches shoppers, that brand is responsible for the age of its own audience. If you believe a child's personal data has reached us through a brand's store, write to [email protected] and we will delete it.
If something goes wrong
If we become aware of a breach of security that affects personal data we hold for a brand, we notify that brand without undue delay after we become aware of it, tell them what we know, and help them with the notifications they have to make. Because we hold the data on the brand's behalf, the brand decides whether and how to notify their own customers and any regulator. We do not notify a brand's customers on the brand's behalf.
Changes to this policy
We update this page when the platform changes. This version takes effect on the date at the top, the day we publish it. We keep every earlier version.
When a version changes how we handle personal data, we can require each person at a brand to accept it before they keep using Ignis. The app then shows them what changed and asks them to accept. Until they do, they cannot use their workspace in the Ignis dashboard, and the API keys and AI apps they set up stop working. In the Shopify app, one owner, admin or verified staff member accepts for the store. They can still sign in and read what changed, and an owner can still export or delete the workspace. We record which version each person accepted, and when, as described under Other records we keep.
Before we use Google user data in a new way, we ask the brand to accept an updated policy.
Contact
Questions about privacy or a data request: [email protected].