개인정보처리방침

1. Scope and controller

This policy applies to easy-gogo (www.easy-gogo.com), operated by Easy Trust Information Co., Ltd. (“Easy Trust”), including the public discovery and provider pages, the customer marketing workspace, and the data interfaces made available to AI agents. Easy Trust is the controller of the personal data described below.

Customer and lead records that you create inside the workspace are data you collect yourself. For those records you are the controller, and Easy Trust only provides the tools that store and process them on your instruction.

2. Data we collect

We collect only what is necessary for the feature you use.

• Threads engagement: after you sync the connected brand account, we store available public posts, replies, platform identifiers, usernames, timestamps, source links, lifetime insights, reply drafts and handling/deduplication records for up to 90 days. You can pause background sync or clear local records; disconnecting removes these records. Content on Threads is not deleted. Sending or uncertain replies must be verified before local records can be cleared. AI reply drafting sends the selected public conversation, basic brand information and enabled brand knowledge to the AI provider configured by the platform. Replies are published only after you confirm the recipient and text and send them; no automatic replies or Threads private messages are included.
• Account: email, name, company or brand name, and a securely hashed password. We never store plain-text passwords. Accounts created with Google or LINE have no password at all.
• Multiple brands: one account can add several brand workspaces (only a brand name is required; a website is optional) and switch between them in the workspace. Each brand keeps its own brand knowledge base, social channel authorizations, content, leads, AI Agent settings, and public provider page as described in this section, as well as the briefs and posts submitted or imported to that brand through ChatGPT or Claude. Sign-in methods (email, password, linked Google or LINE accounts), browser push subscriptions, and ChatGPT or Claude connections belong to the account, not to an individual brand, and notifications for any brand are sent to the devices subscribed under the account. One ChatGPT or Claude connection can use every active brand in the account (including brands added later), and each read or write must name the brand. We record which brand you are currently working in so that data is saved to the right brand.
• Social sign-in: the provider’s account identifier, email, and display name (see sections 4 and 5).
• Brand knowledge base: the brand website, brand description, target audience, products and services, FAQs, and brand tone and writing rules (terms to avoid, commitments that must not be made, fixed terms, and call-to-action preferences) you enter; the company direction and updates you add (including events, customer cases, and source notes); and the images and videos you upload. For the writing rules and the company direction and updates, we also record who last modified them and when.
• Customer and lead records: entered by you; the content is your choice.
• Content calendar: content drafts you create or generate with AI (including any first comment), their review status, planned times and archive status. When you use the AI content assistant, we also keep topic proposals, per-platform drafts, and a snapshot of the referenced information (entry IDs, versions, and generation time) so the basis of each draft can be traced; images generated by the AI content assistant are stored in your media library and are kept and deleted the same way as files you upload. Content you archive is kept until you restore or delete it; when you delete content, the content itself is removed from our database and we keep only a record of which onboarding tasks you completed (no content text). Archiving or deleting content does not affect posts already published or their publishing records.
• Social lead radar: when you enable it, authorized channels provide available comments on recent posts. We retain platform/account identifiers, author identifiers and names, comment text, time, and source links to identify needs and create leads, explainable scores, and follow-up tasks. We do not retrieve liker lists, private messages, or individual website browsing records, infer contact details, or merge identities across platforms. Channel tokens are also used for this server-side reading.
• Comment auto-replies: after you create and enable a rule (keywords, match type, which posts it covers, the private message, and optional public replies), when the lead radar reads a new comment posted after the rule was enabled that matches its keywords, we send that commenter one private message from your brand account through Meta’s private reply feature (any {name} in the message is replaced with the commenter’s name on the platform), and may post a public reply under the comment after the private message is sent. We store the rule and when it was enabled, and a match record for each matching comment: the rule, platform and account identifiers, comment and post identifiers, a hash of the commenter’s identifier (used so the same person is not messaged twice; the raw identifier is not stored), the delivery status and reason for the private message and public reply, and the time. Match records are kept for ninety days. The text sent and the records that prevent duplicate sends are kept with the radar comments for up to thirty days. You can pause or delete a rule at any time; private messages not yet sent are then cancelled.
• Social channel authorization: Meta OAuth retrieves your authorized Pages, linked professional Instagram accounts, and app-scoped user identifier. Account choices and tokens are encrypted temporarily for up to ten minutes; only your selected channel is retained after confirmation. When you connect Threads, Threads authorization gives us the identifier and username of the Threads account you are signed in to, and the confirmation screen shows that username and profile picture (the picture is not stored). Threads authorization expires after about 60 days; we keep the expiry time and renew it automatically before then, and if renewal fails the channel is marked as needing reconnection. LINE and advanced manual connections use tokens you supply. Credentials are encrypted with AES-256-GCM, used only on the server to verify connections, renew authorization and publish, and never returned to the browser.
• Social engagement and messages: when you like a comment or delete a published post from the workspace, we record the time of that action. When you open “FB/IG messages”, conversations and message content are read live from Facebook (Messenger) or Instagram for display and are not stored. When you send a reply, we keep only the request ID, conversation identifier, delivery status, and message identifier needed to prevent duplicate sends, and delete them after seven days. When you set a first comment for a Facebook, Instagram or Threads post, we keep the comment text, its delivery status, the comment identifier returned by the platform, and the time. When a post or first comment fails, we keep the failure category, the number of automatic retries and the next retry time, and a result code for each send or check attempt (never the platform’s raw error text; attempt records are kept for ninety days). When it is unclear whether a send went through, we read the recent posts of your Page, Instagram account, Threads account, or Google Business Profile location (or the comments or replies under that post) to check whether it was published, and keep only the identifier and URL of the matching item — no other posts are stored.
• Social post performance: once the platform enables social post performance for you, when you open “Analytics”, we use your connected channel’s authorization to read, live from Facebook or Instagram, the reaction (like), comment, and share counts of posts published through the service. Where the platform enables it and you grant the insights permissions when connecting the channel, we also read each post’s views, reach, saves, and shares, and your Page or Instagram account’s views for the selected period. These are aggregated figures provided by the platform and contain no data about individual users; they are cached in server memory for up to 30 minutes to avoid repeated requests, are not written to the database, and are never read in the background. Channel tokens are also used for this server-side reading.
• Site audit: the URL you enter, together with the public page content we fetch from it and the resulting findings.
• Reports you import: the contents of the Search Console CSV files you upload; or, after you connect your own Google Search Console, the search performance reports brought in when you press “立即同步” (sync now), together with the encrypted authorization (see section 7).
• Google Business Profile: after you connect your own Google Business Profile, the encrypted authorization, the business account and location you selected, the business posts you publish, and records of review replies you send; reviews, performance figures, and business information are read live from Google each time and their contents are not stored (see section 8).
• Market research: the industry classifications, service areas, and target customer industries you choose (your research scope), and when and by whom (you or a platform administrator) it was last changed. No unified business number is needed.
• Technical and security logs: IP address, browser and device information, access time, errors, and anti-abuse data. IP addresses used for rate limiting are stored as hashes, not in their original form.
• Public provider page statistics: only “which page, which day, how many times” (views, match hits, AI lookups). We do not record the visitor’s IP address or browser identifiers.
• Analytics: public pages load Google Analytics when the platform has configured it. The signed-in private workspace loads no analytics at all.
• Platform admin Google data: only after a platform super administrator authorizes it, we store an encrypted Google refresh token and the selected property settings so that the admin console can read this platform’s website traffic and search performance reports (see section 6).
• AI Agent automatic work: after you turn on an AI Agent in the workspace, we store its settings (on/off switches for each Agent and task, frequency, pause date, and who changed them and when), a record of each run (time, status, IDs of items produced, and reasons for skips or failures), the figures in your weekly performance summary, and the read status of in-app notifications and the delivery status of browser push notifications. Automatic work only uses data already in your workspace (plus the public pages of your website fetched again by the scheduled site check) and collects no new personal data. If you turn on Content Agent weekly drafting, we also keep a record of each task handed to the AI Agent service (time, status, progress text, IDs of the drafts produced, and failure codes) and a log of each data read or write by the AI Agent service (function name, result, and time, without content); if you turn on “auto-publish when overdue”, we keep the deadline you chose, when you turned it on, and each draft’s auto-publish time. If you turn on the Brand, Research, Publish, or Growth Agent work handled by the AI Agent service, we likewise keep task records and keep the suggestions the AI Agent makes: weekly topics and improvement suggestions (title, description, angle, source name and URL, priority, and whether you used or skipped them); brand data check suggestions are saved as drafts in the brand knowledge base (marked as created by the AI Agent, with the source URLs it read), and scheduling suggestions are written to the content’s planned time. If you have connected a Facebook Page, the brand data check task record also temporarily keeps posts that Page itself published in the last 90 days (up to 20, storing only each post’s URL, publishing time, and first 200 characters) and the website pages and post URLs read in that run; these are kept and deleted with the task record. The brand data check also collects public image candidates from your website (logo, brand images, product or service images) and stores the re-encoded image file, source page and image URL, source evidence, size, format, and file hash; only you can see candidates, and candidates that have not reappeared for 30 days and were never edited are deleted with their files. A copy is added to your media library for drafts and short videos only after you select Enable, and you decide whether to disable or delete a brand asset.
• AI video production: when you submit a request under “New short video → AI video”, or after you turn on Content Agent automatic short videos (daily or weekly, as you choose), we store each production job’s request (video topic, key points, call to action, length, the template and virtual actor you chose, voice option, whether brand background is included, the reference assets chosen (media library IDs, use and notes) and any reference images drawn for it, and the request actually sent to the video studio), the draft caption and first comment, who submitted it (you or Content Agent), the studio’s task ID, status, progress, a summary of any reason it is waiting for the studio, failure codes, and times. When the finished video is fetched, it is saved to your media library and a short video waiting for your review is created. If you separately enable automatic publishing after video review, we also keep the grant ID, time, selected social accounts you own, and the delivery review revision, video hash and publication binding. Only new Content Agent requests created after that grant qualify.
• ChatGPT and Claude connections (easy-gogo Plugin): when you use the easy-gogo connection in ChatGPT or Claude, sign in to the service, and approve the connection, we store a connection record (the client ID and callback URL registered by that service, the permissions you approved, the expiry date, and whether it was revoked); connection credentials are stored only as hashes, never in plain text. The connection belongs to your account and can use every active brand in the account (including brands added later); the AI assistant must name the brand on each read or write, and briefs, imported posts and changes made through the connection are saved to the brand it names. Company briefs submitted through the connection (title, facts confirmed by the company, marketing direction, target audience, tone, call to action, source, effective dates, whether it may be used publicly, and requested platforms), the times they were submitted and confirmed, and the ID of any drafting task started from them are stored in your account. We store the summary ChatGPT or Claude prepares from your conversation and any post text, first comment and image files you explicitly select for import, not the full conversation. Temporary image download URLs are used only to fetch the file and are not stored in post metadata or activity records. After you personally confirm a brief on this website, it is added to “company direction and updates” with “ChatGPT brief” or “Claude brief” as its source, and an activity record is written. If you separately grant brand maintenance permission and confirm a preview in the conversation, the AI assistant can change your brand profile (except the website and account name), content rules and company information, or add, turn on or turn off company information; these changes are saved directly to your brand knowledge base, company information added this way uses “ChatGPT conversation” or “Claude conversation” as its source, and activity records are written. If you separately grant comment auto-reply permission and confirm a preview in the conversation, the AI assistant can create, change, enable, pause or delete comment auto-reply rules; rules are saved to the brand it names and activity records are written. The AI assistant itself cannot send private messages. If you separately grant AI Agent settings permission and confirm a preview in the conversation, the AI assistant can turn the AI Agent automations of the brand it names on or off and change their task list, schedule and drafting platforms (it cannot turn on or change automatic publishing of drafts left unreviewed); the settings are the same as those made in the workspace, and activity records are written. We do not store the preview itself, and its confirmation token is valid for 10 minutes.
• Meta ad boosts: when you connect your own Meta ad account, we keep the ad account name, ID, currency, time zone, and status, plus the encrypted ads authorization. For each boost we keep the post you chose, the goal, total budget, number of days, locations, age range and gender, when you confirmed and a hash of what you confirmed, the Meta ad object IDs, delivery and review status (only a reason code if rejected), and daily spend, impressions, reach, clicks, and engagements (see section 10B).

3. Purposes

We use the data to:

• create and maintain your account, session, and workspace content;
• generate marketing drafts, run site audits, bring in the Search Console search performance you authorize, and publish the social posts you request and read their performance figures for you to view in the workspace — including, when you set a first comment for a Facebook, Instagram or Threads post, posting that comment as your Page, Instagram or Threads account once the post has been published (on Threads, as a reply to your own post). When publishing or commenting fails temporarily, it is retried automatically within 24 hours of the scheduled time; when it is unclear whether a send went through, we first check what was recently published on the platform and resend only if it is not there, to avoid duplicate posts;
• analyze comment needs, rank follow-up priority, and provide reply drafts according to the radar settings you turn on — the score is not a probability of closing a sale, and we never confirm a deal on your behalf. Apart from comment auto-reply rules you separately create and enable (which message only people whose comments, posted after the rule was enabled, match its keywords — once per comment and once per person per rule), we never message prospects automatically;
• at the frequency you choose for the AI Agent automatic work you turn on, regularly review the state of your brand data, content waiting for review, scheduling and publishing status, high-intent leads found by the comment radar, and a weekly performance summary, and check your website again. The automatic work above only produces reminders, summaries, and site check records; it never approves, schedules, or publishes content for you, and never replies to or messages anyone (weekly drafting is described next). Notifications go only to you (in-app notifications, and browser push if you turned it on; push messages are fixed text without customer names or comment text, are not sent between 22:00 and 08:00, and work finished overnight is combined into one message in the morning). You can pause or turn it off at any time;
• for the Content Agent auto-drafting you turn on (daily or weekly, as you choose), have the AI Agent service write drafts for the platforms you chose from your brand data and place them in “待你確認” (waiting for your review); drafts are scheduled only after you approve them. If you separately turn on “auto-publish when overdue”, a draft you have not handled within the deadline you chose is first checked (guarantee wording, terms to avoid, unknown links, platform rules, channel connection, and the limit on automatic posts per 7 days, which is 14 by default, may be adjusted per account, and can be set lower by you within that limit) and is approved and scheduled on your behalf only if every check passes, with at most 1 post per platform per day; the first 3 drafts after you turn it on still need your manual approval, you are notified within the final 6 hours before the deadline (for a 4-hour deadline, after the countdown starts), and editing, sending back, or cancelling a draft stops it from being published automatically;
• for the Brand, Research, Publish, and Growth Agent work you turn on, have the AI Agent service make suggestions: each month compare your brand website with your brand knowledge base and save missing, outdated, or conflicting information as knowledge base drafts (effective only after you turn them on); each week suggest content topics based on recent industry information, placed in “Brand topic ideas” on the Content page; each week suggest publishing times for approved or in-review content that is not yet scheduled, filled in as planned times on the calendar (only empty ones, and nothing is scheduled); and each week make improvement suggestions in Analytics based on your results, site check, and search data. You decide whether to use them;
• for the requests you submit under “New short video → AI video”, or the Content Agent automatic short videos you turn on (daily or weekly, as you choose), have the group’s internal video studio make 9:16 short videos; finished videos are fetched and saved to your media library, awaiting your review by default. You may separately enable automatic publishing after video review and select your connected Facebook or Instagram accounts. Only new Content Agent requests created after that grant may be approved and queued on your behalf after checks of the same final AI-reviewed cut, file, caption, accounts and allowance. You can revoke permission at any time. Manual requests, earlier requests and existing drafts are excluded. Changed content or invalid permission stops further submission; unknown results are reconciled first. Queued does not mean published. Requests you submit and those Content Agent submits share the same AI video quota;
• when you click or confirm on screen, like or unlike comments as your brand account, delete posts that easy-gogo published from Facebook or Instagram, and reply to Messenger and Instagram messages within the 24-hour window the platforms allow (message content is never sent to AI providers);
• when you confirm on screen, publish business posts, reply to reviews or delete replies, and update business information such as opening hours through the Google Business Profile you connected, and read reviews and performance figures for you to view in the workspace (see section 8);
• after you review the maximum spend and tick the confirmation for each boost in Ad boosts, create and start an ad for that post with your Meta ad account; pause, resume, or stop it when you ask, and regularly read its delivery and review status and results (see section 10B).
• show market statistics aggregated from government registration data, and topic ideas, based on your research scope (counts and shares only, never individual businesses; see section 14);
• following your instructions in ChatGPT or Claude and your confirmation on this website, let that AI assistant read your brand data and save briefs for your confirmation; a brief is added to your brand knowledge base only after you personally confirm it here, and Content Agent creates drafts for your review, or Brand Agent drafts company information from your saved website and connected Facebook Page posts, only after you ask it to. With separately granted permissions, your own AI assistant can import post text and images and ask the service to publish publicly to your connected social account after showing the complete preview and exact account and obtaining your explicit confirmation in the conversation. With separately granted brand maintenance permission, the service changes your brand knowledge base only after the AI assistant shows each change before and after in the conversation and obtains your explicit confirmation. With separately granted comment auto-reply permission, the service saves or enables a rule only after the AI assistant shows its keywords, covered posts, full private message, public replies and whether it starts now in the conversation and obtains your explicit confirmation; private messages are sent by the service according to enabled rules, and the AI assistant cannot send messages directly. Importing or previewing does not publish automatically. Scheduling is not provided;
• serve your public provider page once you submit it and it is approved;
• provide statistics and support, and send essential service and security notices;
• detect abuse and protect the security and quality of the service; and
• comply with applicable law.

4. Google account sign-in and Google user data

The service offers “Sign in with Google”. If you choose it, Google provides us with the data below after obtaining your consent. If you do not use it, we receive none of your Google user data (if you connect your own Google Search Console, see section 7; for a platform administrator’s authorization to read this platform’s website data in the admin console, see section 6).

• Scopes accessed: only the basic openid, email, and profile scopes. Of these we actually use your Google account identifier, email address, and name. We neither request nor can read your Gmail messages, Drive files, contacts, calendar, or any other Google service content.
• How we use it: solely to create or sign you in to your account on this service, to identify you, and to show your name and email in the workspace.
• What we store: your Google account identifier, the email address at the time of linking (so a later sign-in maps to the same account), and your name. We do not store your Google password, nor any Google-issued access or refresh tokens.
• Sharing: we do not sell or trade Google user data, and do not provide it to third parties for advertising or marketing. We disclose it only where required by law or a competent authority.
• Limited use: our handling of Google user data complies with the Google API Services User Data Policy, including its Limited Use requirements.
• Unlinking: you can unlink Google under “Workspace → Settings → Sign-in methods” (set a password or keep another sign-in method first so you do not lose access), or remove this service from the third-party access page of your Google account. Once unlinked, we delete the link record.

5. LINE account sign-in and LINE user data

The service offers “Sign in with LINE”. If you choose it, LINE provides us with the data below after obtaining your consent. If you do not use it, we receive no LINE user data at all.

• Scopes accessed: only openid and profile. An email address is obtained only where we hold LINE’s email permission and you separately consent. We do not read your chat messages, friend list, or group information.
• How we use it: solely to create or sign you in to your account and to identify you.
• What we store: your LINE user identifier and display name, plus the email address where you consented to provide it. We do not store LINE-issued access tokens.
• Where no email is provided: we assign a placeholder address for identification only (ending in @no-email.easy-gogo.com) that cannot receive mail, and ask you in the workspace to supply your own email so we can reach you when needed.
• Unlinking: you can unlink LINE under “Workspace → Settings → Sign-in methods” (set a password or keep another sign-in method first). Once unlinked, we delete the link record.
• The separate feature “connect your own LINE Official Account to publish posts” is unrelated to sign-in; it uses the Channel access token you supply yourself. See section 10.

6. Platform admin access to Google Analytics, Search Console, and Bing Webmaster Tools data

This section applies only to the platform’s own admin console and is unrelated to the customer “Sign in with Google” in section 4 and to a customer connecting Search Console in section 7: a super administrator of the platform’s admin console may use their Google account to authorize this service to read, on a read-only basis, the Google Analytics (GA4) and Google Search Console data of this platform’s website. Customers and visitors are never asked to grant this authorization.

• Scopes accessed: only two read-only scopes, analytics.readonly (Google Analytics, read-only) and webmasters.readonly (Search Console, read-only). We do not request any scope that can change Google Analytics or Search Console settings, nor email, profile, or any other Google service scope.
• Data we read: the names of the GA4 accounts and properties and the list of Search Console sites that the authorizing account can access, so that the administrator can select this platform’s website; the measurement ID of the selected GA4 property’s web data stream; and site-wide aggregate reports for the selected properties — for GA4, users, new users, sessions, views, engagement rate, user engagement time, key events, top pages, and traffic sources; for Search Console, clicks, impressions, click-through rate, average position, top queries, and top pages. This authorization does not read data about individual visitors, nor any customer’s or visitor’s Google account data.
• “AI performance” aggregates: using the same analytics.readonly authorization, we read GA4 aggregates of sessions arriving from AI assistants — the default channel group “AI Assistant” and known AI service sources (for example ChatGPT and Perplexity): sessions, users, engagement rate, source platform, landing page path, key event count, and the share of sessions with a key event — plus the list of key events configured on the property (only to tell whether any are configured). No tracking code, event, or cookie is added, no GA4 setting is written, and nothing is sent to AI providers.
• How we use it: solely to show this platform’s website traffic and search performance on the admin “流量分析” (traffic analytics) page to platform staff who have permission for that page, and to let administrators choose which properties to show in the admin settings. When a GA4 property is selected, its measurement ID is applied to the Google Analytics setting of the public pages (see “Analytics” in section 2). Report figures never appear on public pages or in the customer workspace and are never sent to AI providers.
• What we store: the refresh token issued by Google, encrypted with AES-256-GCM in this service’s database; the scopes Google actually granted; the GA4 property ID and Search Console site URL the administrator selected; and the measurement ID. Short-lived access tokens and report data are held only in server memory (reports for at most 12 minutes) and are never written to the database or to files.
• Sharing: we do not sell or transfer this data and do not provide it to third parties; it is not used for advertising or to train AI models. We disclose it only where required by law or a competent authority.
• Revoking access: a super administrator can click “中斷連結” (disconnect) under “設定 → Google 數據” (Settings → Google data) in the admin console. The service then revokes the authorization with Google, deletes the stored refresh token, granted-scope record, and property selections, and clears its in-memory caches; these stored records are deleted even if revocation with Google fails. The authorizing user can also remove this service’s access on the “Third-party apps & services” page of their Google Account, after which the service can no longer read this data.
• Limited Use: easy-gogo’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.
• Bing Webmaster Tools: a super administrator may also separately use their Microsoft account to authorize this service with the read-only webmaster.read scope (an account-level authorization). The service’s reports query only site-wide clicks, impressions, top queries, and top pages of this platform’s main site in Bing web search; “Check main site” reads that account’s site list solely to confirm whether the main site is listed and verified, and other sites are neither displayed nor stored. The service does not add or remove sites or submit URLs. The refresh token issued by Bing and the client secret entered in the admin settings are encrypted with AES-256-GCM in this service’s database; reports are held only in server memory (for at most 12 minutes). A super administrator can click “中斷連結” (disconnect) under “設定 → Bing 數據” (Settings → Bing data); the service then deletes the stored refresh token and clears its caches. To also revoke the authorization on Microsoft’s side, the authorizing user can remove this service under API access in Bing Webmaster Tools.

7. Customers connecting their own Google Search Console

When the platform makes this feature available, you may choose, on the “SEO / AI Discovery” page of the workspace, to authorize this service with your Google account to read, on a read-only basis, the Google Search Console search performance of your brand website, instead of exporting and uploading CSV files yourself. The feature is optional; you can keep importing CSV files without connecting. This authorization is separate from “Sign in with Google” in section 4: signing in with Google does not let us read Search Console, and connecting Search Console does not change how you sign in.

• Scopes accessed: only one scope, webmasters.readonly (Search Console, read-only). We do not request any scope that can change Search Console settings or site properties, nor email, profile, or any other Google service scope. If you do not grant the Search Console read permission on Google’s consent page, the service immediately revokes that authorization with Google and stores nothing.
• Data we read: the list of Search Console sites your Google account can access (site URL and permission level), used only to find the verified property that covers the website entered in your brand knowledge base; and that property’s aggregate search performance for the last 90 days (up to about 3 days ago) — clicks, impressions, and average position by date, by search query, and by page, up to 1,000 rows each, keeping only pages on your brand website itself (where the matching property is a domain property, the date and query figures cover the whole domain). These are aggregate figures produced by Google and contain no data about individual searchers; we read nothing else in your Google Account.
• How we use it: solely to show search trends, query keywords, and page performance on the “SEO / AI Discovery” page of your workspace, handled the same way as CSV reports you import. These reports never appear on public pages or your public provider page and are never sent to AI providers; only if you yourself choose to create a content draft from a search query in the keyword table is that query text sent, as a topic, with your instructions (see section 10).
• What we store: the refresh token issued by Google, encrypted with AES-256-GCM in this service’s database, together with the time you connected and the time of the most recent authorization; and the three reports brought in by a sync, stored in your workspace and labeled “Google 自動帶入” (brought in from Google), each sync replacing the earlier report of the same type (including an imported CSV). Short-lived access tokens are used only when you check the property or sync and are never written to the database or to files; the site list and matching result are not stored. The service never syncs automatically in the background — reports are read only when you press “立即同步” (sync now).
• Sharing: we do not sell or transfer this data and do not provide it to third parties; it is not used for advertising or to train AI models. We disclose it only where required by law or a competent authority.
• Revoking access: while the feature is available, you can click “中斷連結” (disconnect) in the Search Console area of the same page. The service then revokes the authorization with Google and deletes the stored refresh token and connection record, together with the reports brought in from Google; CSV reports you imported are kept. These stored items are deleted even if revocation with Google fails. You can also remove this service’s access at any time on the “Third-party apps & services” page of your Google Account, after which the service can no longer read this data; reports already brought in remain in your workspace until you delete them with “清除報表” (clear reports), and the encrypted authorization record can be deleted by clicking “中斷連結” (disconnect) or by contacting us as described in section 15.
• Limited Use: easy-gogo’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

8. Customers connecting their own Google Business Profile

When the platform makes this feature available, you may choose, on the “Google 商家” (Google Business) page of the workspace, to authorize this service with your Google account to manage Google Business Profiles that you own or are authorized to manage: publishing business posts, replying to reviews, viewing performance figures, and updating some business information. The feature is optional and independent of “Sign in with Google” (section 4) and of connecting Search Console (section 7).

• Scopes accessed: only one scope, business.manage (manage your Google Business Profile). We do not request email, profile, or any other Google service scope. If you do not grant this permission on Google’s consent page, the service immediately revokes that authorization with Google and stores nothing.
• Choosing a location: we read the names of the business accounts your Google account can manage and the name, address, and posting eligibility of the locations under them, used only to let you choose the location to manage; before saving your choice we confirm with Google that the location belongs to that account.
• Business posts: only after you press “立即發布” (publish now) or set a schedule in the workspace does the service publish a business post (update, event, or offer) to the location you selected, consisting of the text you approved, one photo you chose from your media library (Google fetches it from the service’s public URL), and the call-to-action button, period, and offer details you set. Business posts are shown publicly on Google Search and Maps. If posting fails temporarily, the service retries automatically within 24 hours of the scheduled time; when it is unclear whether a post went through, it first reads that location’s recent business posts to check and resends only if the post is not there, keeping only the identifier and URL of the matching post. You can delete posts published by the service from your publishing history.
• Reviews and replies: when you open the Reviews tab, we read that location’s reviews from Google in real time (reviewer display name, star rating, review text, time, and your reply) solely to display them in your workspace; review content is not stored. Replies, edits, or deletions you confirm are published on Google under your business name. To prevent duplicate submissions we keep only a request ID, the review’s identifier, the processing status, and a hash of the reply text, deleted after seven days. What is sent to AI providers when you press “AI 擬草稿” (AI draft) is described in section 10.
• Performance figures: when you open the Performance tab or the analytics page, we use your authorization to read that location’s daily performance from Google — impressions on Google Search and Maps (desktop and mobile shown separately), website clicks, calls, direction requests, bookings, and menu clicks — and the search keywords customers used to find your business with their monthly impressions (for low-volume keywords Google only provides “fewer than N”). These figures are held only in server memory for up to 30 minutes to avoid repeated requests and are not written to our database; the service does not read them in the background.
• Business information: you can change opening hours, special hours, the business description, the primary phone number, and the website. After you confirm the before/after changes, we send only the changed fields to Google and read back Google’s response (for example, whether the change is pending review). Business information is read live from Google each time and its contents are not stored; your account activity log records only which fields were updated and when, not their values. Business name, address, and categories are only read and displayed, never changed. Submitted information is shown publicly on Google Maps and Search and is subject to Google’s terms and review policies.
• What we store: the refresh token issued by Google, encrypted with AES-256-GCM in this service’s database; the time you connected; the business account and location you selected (names and address); and the posts you published, their settings, and the post identifier and URL returned by Google (like other social publishing records). Short-lived access tokens are never written to the database or files.
• Sharing: apart from what is sent to AI providers when you press “AI 擬草稿” (AI draft), as described in section 10, we do not sell or transfer this data and do not provide it to third parties; it is not used for advertising or to train AI models. We disclose it only where required by law or a competent authority.
• Revoking access: you can click “中斷連結” (disconnect) on the same page. The service then revokes the authorization with Google, deletes the stored refresh token and location selection, and cancels Google Business posts not yet sent; posts, replies, and business information already on Google are not deleted or reverted. The stored token is deleted even if revocation with Google fails. You can also remove the service’s access at any time on the “Third-party apps & services” page of your Google Account.
• Limited Use: easy-gogo’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

9. Public provider pages: what becomes public

You may submit your brand information for review in the workspace. Once approved by the platform, the provider page is public content: anyone can view it, search engines may index it, and AI agents may read it through the interfaces this site offers (llms.txt, the Match API, MCP) and cite it in their answers.

Please put only information you are willing to publish on a public provider page. Do not include national identification numbers, financial account numbers, other people’s personal data, or anything that should not be public. You can edit and resubmit the page at any time, or contact us to take it down.

Content that has not been submitted or approved is not shown publicly, and your customer records, drafts, and media files are not made public along with the page.

10. Third-party services and credentials you supply

To complete the actions you trigger, data is sent to the third parties below, limited to what that action requires.

• AI providers: after AI content generation is enabled for your account, pressing “generate” sends brand materials and instructions to the AI provider and returns a draft for review. Signed-in users can also choose website analysis during onboarding without individual AI content access. When you press “analyze website”, or when a brand has a website but its description or products & services are still empty (when you create a brand with a website, save your brand basics, or open the workspace; at most one automatic run per website, and no automatic retry for 6 hours after a failure), we read the public homepage and up to two related pages; if the platform has configured AI, extracted text is sent to OpenAI to suggest brand information, audiences and services. We retain suggestions and source URLs, not the original web pages. Suggestions only prefill fields you have not filled in and become brand information only after you confirm and save them; they are not automatically published.
• AI content assistant (content calendar): when you request topic proposals, platform drafts, or revisions, we send your basic brand information (company name, website, brand description, products and services, target audience, FAQs), your writing rules (brand tone, terms to avoid, commitments that must not be made, fixed terms, call-to-action preferences), the goal, channels, focus, and link you chose, and only those company direction and updates entries that are active, approved for external use, already effective, and not expired, to the AI provider. Drafts, disabled entries, entries marked internal-only, and entries outside their effective period are not sent. When you ask for a revision, the current text in your editor is also sent. Your campaign service, primary action, keyword, confirmed reception next step, destination purpose and title requirements are saved and sent to AI. A content and routing check also sends the current internal title, body, first comment and draft image copy; the quoted input, findings and suggested revision are saved for you to compare and apply. AI output only becomes a draft or suggestion; you must confirm and approve it before it can be scheduled, and it is never published automatically. When AI images are enabled by the platform, after a platform draft is first completed (or when you press “generate image”), we send the topic title, that draft’s image suggestion, the key points and call to action to be shown on the image, and your company name to our group’s internal image-generation service; if you choose an AI editor’s image style, that style’s design description (the platform’s own content, not your data) is sent as well. When the image is ready, we send it to the AI service configured by the platform (OpenAI or our group’s internal AI service) to check people in it for obvious anatomical errors, and redraw it once automatically if needed; the image sent for checking is not stored separately. The image is saved to your media library and selected in that draft, and likewise must be confirmed and approved by you before it can be scheduled; it is never published automatically. When you press “AI image” or “Try another” while editing content or a social post, we send that post’s title (for a social post without a title, its first line of text), the image description you enter, and the design description of the AI editor image style you last chose to the same image-generation service, with the same anatomy check; the post text itself is not sent. The image is saved to your media library and added to the post you are editing; it is saved to the content only when you press Save, and it counts toward the same AI image quota.
• Google Business Profile review AI drafts: when you press “AI 擬草稿” (AI draft) under “Google 商家 → 評論” (Google Business → Reviews), the review’s star rating and text, the reviewer’s first name (only when written in Latin letters, without the surname; Chinese names are not sent), your brand basics (company name, brand description, products and services, FAQ), and your brand tone are sent to the AI service configured by the platform (OpenAI or the group’s internal AI service) to generate a reply draft. The draft is only placed in the reply box and is sent only after you confirm it; we do not store drafts.
• AI Agent service (auto-drafting): after you turn on Content Agent auto-drafting, the system hands a task to the group’s internal AI Agent service (Agno AgentOS) at the scheduled time for the frequency you choose (daily or weekly), sending only your account number, the task type, and the platforms you chose, together with a short-lived credential that can access only your account and only during that task. With that credential the AI Agent service can read from the service only the data the task needs: your brand profile (company name, website, brand description, products and services, main audience, FAQ), company direction and updates in the brand knowledge base that are active, approved for external use, already effective, and not expired, your content guidelines, the titles and status of recent content, the figures in last week’s workspace summary, and the file names, formats, and URLs of images and videos in your media library (for choosing images). This content is sent through the AI Agent service to Google Gemini to write the drafts; at launch the service uses Google’s paid service, and Google does not use this content to train models. While writing, the Agent browses public web pages for background information, for reference only, and never enters your data into other websites. The Agent can choose images from your media library or ask the service to generate an image for a draft: as with AI editor images, the draft’s internal title, image description and key points, call to action, and company name are sent to the group’s internal image generation service; the image is saved to your media library and attached to that draft (drafts you have already given images, approved, or sent back are not changed), and it counts toward the same AI image quota. By default the AI Agent service does not keep content: it is deleted once the result is delivered to the service, within 24 hours at most, and only usage summaries without content are kept; it also does not remember your data for later tasks. It can only create drafts waiting for review and cannot publish, schedule, send messages, or change brand data (auto-publish when overdue is performed by the service according to your settings).
• AI Agent service (brand data check, weekly topic research, weekly scheduling suggestions, weekly improvement suggestions): after you turn on this work for the Brand, Research, Publish, or Growth Agent, the way tasks are sent, the short-lived credential, processing by Google Gemini, and the AI Agent service not keeping or remembering content are the same as for weekly drafting. In addition to the brand and performance data weekly drafting can read, depending on the task it can also read: all items in your brand knowledge base (including drafts and disabled or expired items, to find gaps and outdated information); the titles, channels, status, and planned times of approved or in-review content that is not yet scheduled; the publishing times and results on each channel in the last 30 days (without post content) and channel connection status; the latest site check results; and Search Console search queries and page figures (clicks, impressions, average position). For the brand data check, the service reads the website in your brand profile (the home page and up to 4 other public pages on the same site; the AI Agent service opens your website in a browser only when the service cannot read its content) and the text of posts your connected Facebook Page itself published in the last 90 days (up to 20; never comments, messages, visitor posts, or other Pages, and the Page token is used only inside the service and never given to the AI Agent service), and passes them to the AI Agent service for comparison; each draft it proposes must cite a source URL it actually read, and is used for AI writing only after you turn it on. Website images are downloaded by the service in the background (public addresses only, with per-image and per-run size limits; SVG is not supported). The AI Agent service receives only descriptive data (URL, source evidence, size) and an ID, can only place an image into “Needs review”, and cannot enable or replace your official logo. Enabled brand assets are available to the Content Agent; their image files are sent to the AI video platform only when you or Content Agent choose them as reference assets for a short video (see “Group video studio” below). If you allow “brand source sync” in your ChatGPT connection, ChatGPT can also start this check on your instruction; weekly topic research and improvement suggestions browse public news, industry information, or public pages of your website, for reference only. This work can only write suggestions: brand knowledge base drafts, topic ideas on the Content page, planned times for content, and improvement suggestions in Analytics; it cannot publish, schedule, send messages, or change, turn on, or turn off your brand data.
• ChatGPT (OpenAI): after you connect ChatGPT, ChatGPT reads data from or sends data to the service during your conversations within the permissions you approved, and that data is processed by OpenAI. ChatGPT can read: the names and websites of the brands in your account (so you can choose which brand to work on), and, for the brand it names, your brand profile (company name, website, brand description, products and services, main audience, tone, FAQ), company direction and updates currently available to AI (active, approved for external use, already effective, and not expired), your content guidelines, and the briefs submitted through the connection (including content you marked internal only), their confirmation status, drafting progress, and the titles, platforms, status, and workspace links of the drafts, as well as the status of brand source syncs started through the connection, how many sources were read, and the titles, status, and workspace links of the company information drafts created (not the website or Page post text itself). Unless you grant brand maintenance permission, it cannot read existing internal information in your brand knowledge base; with that permission it can read all company information in your brand knowledge base (including drafts, disabled and internal-only items). With comment auto-reply permission it can also read your comment auto-reply rules, their delivery statistics for the last 7 days, and the identifiers, times, first 100 characters and links of the 25 most recent posts on your connected accounts; it cannot read commenters’ names or comment text. With AI Agent settings permission it can also read the automation settings and status of each AI Agent of the brand it names (whether it is on, any pause date, each task’s on/off state, schedule, next run time, latest result and missing requirements, and the drafting platforms and automatic publishing setting), not the content the agents produce. In either case it cannot read your account name or email, your password or any credentials, or any other account. ChatGPT sends the brief prepared from your conversation, brand changes you confirmed in the conversation, and, when you request import, your selected post text, first comment and image files. Temporary image download URLs are only used to fetch the file and are not stored in post metadata or activity records. Your ChatGPT conversations and the data the service returns to ChatGPT are processed by OpenAI under its own terms and privacy policy, not this policy, and may be processed outside Taiwan. When you ask for drafts, the task is handled by the AI Agent service described above, in the same way and with the same data as “AI Agent service (weekly drafting)”, plus the confirmed brief approved for external use, and it uses the same drafting and image quota; the service does not send your data to the OpenAI API for this.
• Claude (Anthropic): after you connect Claude, Claude can read and send the same data as described for ChatGPT above, within the permissions you approved, and that data is processed by Anthropic. Your Claude conversations and the data the service returns to Claude are processed by Anthropic under its own terms and privacy policy, not this policy, and may be processed outside Taiwan. The service does not send your data to the Anthropic API for this.
• Group video studio (AI video): when you submit an AI video request, or Content Agent automatic short videos submits one for you, the service uses the platform’s own service credential to send the request to the group’s internal video production platform: the video topic, key points, call to action, length and 9:16 format, and the template, virtual actor, and voice option you chose. If you choose to include brand background (requests submitted by Content Agent always include it), we also send your company name, brand description, products and services, main audience, brand tone, and up to 8 company direction and updates entries in your brand knowledge base that are active, approved for external use, already effective, and not expired. The studio may use GPU computing and the AI services it has configured to make the video; when footage or quality needs attention, studio staff may handle it before production continues. Reference assets: the reference assets you choose in the request, or that Content Agent chooses for the topic, are attached with their use notes. These can include your official logo from brand assets (attached by default; you can untick it), images or videos from your media library (each up to 50 MB, 200 MB in total, at most 12), and reference images AI draws first when there are not enough product or scene images (these go to the same internal group image-generation service as AI images, which receives only the video topic, key points and a shot description, count against the same AI image quota, and are saved to your media library). Media library files you did not choose are not sent. We do not automatically send your login details, separate contact fields, or social tokens to the studio, and the studio only makes videos — it does not publish. The finished video is fetched from the studio by the service with the same service credential, saved to your media library, and created as a short video waiting for your review; publishing is handled by the service’s existing flow under your individual approval or previously saved automatic-video grant and selected accounts; social tokens are not sent to the studio. When you cancel a production, the service tells the studio to stop the remaining steps, and a video that finishes after cancellation is not saved to your media library; under the studio’s rules, cancelled production projects and files are kept on the studio. Content Agent can submit requests only through the tools this service provides; the AI Agent service never receives the studio credential and cannot see other accounts’ productions.
• When social lead radar and AI access are enabled, comment text and limited brand materials are sent to OpenAI for intent analysis. Reply generation also uses these materials. Scores prioritize follow-up and are not conversion probabilities; only text you confirm for immediate or scheduled delivery is sent to the social platform. If AI is unavailable, the interface identifies the rule-based assessment or reply template.
• Social platforms: when you publish, the post text and media are sent to Facebook, Instagram, Threads, or LINE as you configured; if you set a first comment, its text is sent to Facebook, Instagram or Threads after the post is published and posted as your Page, Instagram or Threads account (not available for LINE). If publishing fails, we retry automatically depending on the cause; when it is unclear whether a send went through, we first read the recent posts of your Page, Instagram or Threads account (or the comments or replies under the post) and resend only if it is not there, and for LINE we use LINE’s official retry key so the message is not sent twice. Those platforms handle data under their own policies. You can remove channels in the workspace or revoke authorization or request deletion through Meta (including Threads website permissions). After verifying a Meta or Threads callback, we remove the corresponding channel credentials, account links, and pending choices. Your platform account and self-created content remain. Backups expire under our retention policy.
• Websites you audit: an audit fetches the pages of the URL you provide as a public visitor, which may leave a record of our request in that site’s server logs. If you turn on the AI Agent scheduled site check, the same website is fetched again at the frequency you choose (weekly or monthly).
• Meta (ad boosts): after you confirm a boost, we use the Meta Marketing API to send the post ID, goal, total budget and schedule, locations, age range and gender to your own Meta ad account to create the ad, then read its review and delivery status and results. We never upload customer lists, email addresses, or phone numbers.
• Hosting, email, and analytics providers: they process data only as necessary to provide the service and must apply reasonable confidentiality and security measures.
• Providers may process data outside Taiwan. We do not sell personal data and do not provide your workspace content to third parties for advertising.

10A. Threads Chat Agent

When you request a Taiwan news screenshot, an independent mobile browser reads a public TechNews or INSIDE article and captures one visible screen near its headline. The source URL and capture time are recorded, and the screenshot is saved to your media library. No personal login or full-article capture is used. Publisher popular lists and recent news are labeled separately, and missing view or comment counts are not estimated. Screenshots are offered for review and selection before confirmed publishing to Threads. Library images are retained until you delete them or close your account.

When you explicitly start public discussion discovery, the platform reads designated public discussion pages through an independent Browser Grid session without using your personal login. It stores available headlines, article and discussion URLs, public points and comment counts, observation times and selection reasons, matching your saved topics to support opinion drafts. It does not collect lists of visitors or retain complete discussion threads. Activity signals describe that source at the time observed, rather than popularity on Threads or the truth of the content. Discovery does not publish automatically; the topic and draft retention periods below apply.

You can set the personality prompt, topics, random daily posting range and hours, and separately opt into replies to new comments on your own posts. Enabling Chat Agent authorizes background publication of AI posts. Only separately enabled interaction publishes automatic replies, up to 20 per day. The agent selects suitable comments to answer and may skip simple agreement, thanks, or conversations that have naturally ended; it does not promise to answer every comment. Manual drafts and comments marked for a human are preserved. It sends no private messages, searches no other accounts posts and creates no leads. Personality, topics, brand profile, active public brand knowledge, recent posts, and the post and comment context required for replies are sent to the configured OpenAI or group AI service. You can pause the agent. Revoked authorization, an account change or platform suspension stops new sends. Activity is retained for 90 days; uncertain results are retained until resolved. This operates under your activation authorization, separately from replies confirmed one at a time.

Topic opinions reads a public URL you provide or an item you select from the platform RSS feed. It stores the source URL, the URL actually read, title, publisher, date, and available excerpt, rather than the full article. These source details, your personality prompt, and topics are sent to the same AI service to write an opinion draft. Popularity absent from RSS is not estimated. You must confirm the text, any image selected from your own media library, and the publishing time before it enters the existing social publishing queue. Apart from mobile news screenshots you explicitly request, original images from source articles are not downloaded or reposted. Unsubmitted drafts remain available for editing. Submitted or discarded drafts are removed after 90 days; older sources are also removed after 90 days once no draft references them. Existing retention rules continue to apply to social publication records and images in your media library.

10B. Meta ad boosts

Connecting: when you connect a Meta ad account under Social → Ad boosts, we use Meta authorization to list the ad accounts you can use (name, ID, currency, time zone, status). The list and the authorization are kept encrypted for at most ten minutes; after you choose one, we keep only that account and the encrypted authorization.

Costs: Meta bills your ad account directly. Easy Trust never handles the money and does not store your payment details. A boost is created only after you review the maximum spend and tick "Costs are paid by my Meta ad account"; on Meta the maximum is locked with a lifetime budget and an end time. The platform also sets per-boost and per-brand 7-day limits that customers cannot raise. Audience and reach estimates come from Meta and are not guarantees.

Records: so every charge can be traced, each confirmation, pause, resume, stop, connection, and disconnection, and every create, activate, and pause call the system makes to Meta, is written to an append-only log (time, actor, amount, confirmation hash, and result code). Authorization tokens are never logged.

Results analysis and boost suggestions: when you open a boost's details, we use that ad account's authorization to read the boost's aggregated results from Meta (spend, impressions, reach, clicks, and the number of likes, comments, shares, and link clicks), grouped by age and gender, placement, and region. These are figures Meta has already aggregated and contain no individual user data; they are held in server memory for at most three hours and are not written to the database. If the platform has enabled social post performance for you, "Suggested boosts" picks posts to boost from the likes, comments, and shares of posts you published through the service and from the results of your past boosts, and estimates roughly how many engagements a boost may get. The estimate is based only on your own boost history and is not a guarantee. A suggestion only pre-fills the boost dialog; nothing is created until you review the maximum spend and tick to confirm.

Platform-wide stop: if something looks wrong, platform administrators can pause every running boost at once. When the stop is lifted nothing resumes automatically; you press Resume yourself.

Revoking: you can disconnect in the workspace (stop running boosts first) or remove this app in your Facebook settings. When we receive a verified revocation from Meta we delete the authorization right away and mark running boosts as "Authorization revoked"; after that the service can no longer pause them for you, so manage them in Meta Ads Manager.

11. Retention and security

Data is retained while your account exists and for any period required by law. After termination we delete or de-identify it within a reasonable time, except where retention is necessary for legal obligations or to resolve disputes.

We apply access control, encrypted transport, password hashing, encryption of social tokens, rate limiting, and backups. No transmission or storage over the internet can be guaranteed absolutely secure.

If you enable browser push, encrypted subscription endpoints and keys are stored and used with your browser push provider for due-task reminders. Lock-screen notifications contain no lead names or comment text. You may disable this browser subscription at any time. Confirmed public/private reply text, schedules and delivery outcomes are retained with the source comments; clearing the source removes these records but cannot retract messages already sent.

AI Agent run records are kept for the latest 300 runs per account and for up to 180 days; older records are removed by background cleanup, and in-app notifications are shown for the latest 30 days. Turning off an Agent does not delete existing records; they are deleted together with the account. Records of tasks handed to the AI Agent service (time, status, progress text, IDs of the drafts produced, and failure codes) and logs of the AI Agent service’s data reads and writes (function name, result, and time, without content or credentials) are kept for 90 days after the task ends and are then removed by background cleanup; drafts created by the AI Agent are kept like your other content, until you delete them or the account is deleted. On the AI Agent service side, content is deleted by default once the result is delivered, within 24 hours at most. Weekly topics and improvement suggestions made by the AI Agent are kept for the latest 30 of each kind per account; those left unhandled for over 30 days are automatically marked as skipped; those you used or skipped are removed by background cleanup after 90 days, and all are deleted together with the account. Brand knowledge base drafts created by the AI Agent that nobody has confirmed or edited for over 30 days are deleted by background cleanup; other brand knowledge base drafts and planned times for content are kept like your other data, until you delete them or the account is deleted.

AI video production records (request, status, progress, failure reasons, and times) are kept while your account exists and are deleted together with the account; finished videos saved to your media library and short videos waiting for review are kept like your other content, until you delete them or the account is deleted.

Meta ad boost records, daily results, and the action log are kept while your account exists and deleted together with it; technical records of calls to Meta (only the action name and result code) are kept for 90 days. The ads authorization is deleted as soon as you disconnect or Meta revokes it.

A ChatGPT or Claude connection lasts up to 30 days and stops working when it expires or when you revoke it on the “Connect your AI assistant” page (www.easy-gogo.com/connect/chatgpt). Revoking only stops that assistant’s access; it does not delete submitted briefs, brand information you confirmed or changed through the conversation, or drafts already created. These are kept with your account, up to 100 briefs per account. You can disable saved brand information in the brand knowledge base and delete drafts in the content library; to delete briefs or other data, contact us as described in section 15.

Radar comments are kept for the latest thirty days and expired comments are removed by background cleanup. You may pause collection or clear a source. Clearing the source, removing the channel, or a verified Meta revocation/deletion notification removes that source’s comments, automatically created leads, analyses and follow-up records. Deleting one lead retains a hash of the source author identifier to prevent rediscovery; clearing the source also removes that exclusion. Analysis summaries and saved follow-up records remain until you clear the data or terminate the account. Backups expire under the existing retention policy.

12. Your rights

Subject to applicable law you may request access, a copy, correction or supplementation, cessation of collection, processing or use, and deletion of your personal data. Most account data can be viewed and edited directly in the workspace; to delete an entire account, contact us using the details below.

Where data is necessary to perform a contract, protect security, or meet a legal retention requirement, we may limit part of a request after explaining why.

13. Cookies and external links

We use a cookie to keep you signed in to the workspace, a cookie to remember your language preference, and a short-lived (ten-minute) cookie that secures social sign-in, Meta and Threads channel authorization, connecting Google Search Console, and connecting Google Business Profile. Analytics cookies are used only on public pages and only where analytics is configured. You can manage cookies in your browser, but disabling essential cookies makes signing in impossible.

When you choose to register from an audit result, an encrypted cookie retains the website URL and check summary for up to one day so you can continue after registration or social sign-in. The summary is saved in your workspace only after you sign in and confirm the website. We do not store the original HTML fetched for the audit.

The service may link to third-party or provider-owned websites, whose data handling is not covered by this policy.

14. Business development data and market statistics

In the platform admin console, Easy Trust uses government open data (such as the Ministry of Finance business tax registration and the Ministry of Economic Affairs company registration) and web pages that businesses publish themselves (official websites, public case studies by information-system vendors, job postings, and news) to compile business names, unified business numbers, addresses, registered capital, industry classifications, publicly listed company contact details, and statements about the information systems they use. This is used only by authorized platform staff for business development; data about individual businesses is not shown on public pages or in the customer workspace, and is not sold.

To summarize statements about information systems (ERP), relevant passages from public web pages are sent to an AI provider (OpenAI, or the group AI gateway configured by the platform and the providers behind it) for extraction; only the brand, usage, verbatim quotes, and source URLs are kept. Providers may process data outside Taiwan.

“Market research” in the customer workspace shows only statistics aggregated from the Ministry of Finance business tax registration data above that cannot identify any individual business or person (for example, the number and share of businesses in an industry in an area): numbers under 5 are not shown on their own, breakdowns are not shown for fewer than 20, and no individual business name, unified business number, address, or contact detail is shown. Official websites, contact details, and information-system data collected in the admin console are not used for these statistics. Businesses excluded at their request are no longer counted from the next statistics update.

A business or person concerned may contact us as described in section 15 to ask us to stop using or to delete the data. After verification we will exclude the business, stop collecting, updating, or contacting it, and keep only the minimum record needed to prevent it from being imported again.

15. Contact and updates

To exercise your rights or ask about personal data, email [email protected] or call 03-5827637.

Material changes to this policy will be announced on this page together with an updated effective date.