Rylvo Audience Hub: Unified End-User Management with Segments, Insights, GDPR Compliance, and CSV Import/Export

Every time someone sends a message to your AI bot, a relationship is formed. That person on WhatsApp asking about your product. The customer on your website widget who needs support. The employee on Slack querying your internal knowledge base. Each one becomes part of your audience. But here is the problem that most businesses face: all of this valuable user data is scattered across disconnected systems, fragmented dashboards, and isolated databases.

You end up with WhatsApp users in one tool, web widget visitors in another, and no way to see that the same person might have interacted with three different bots on two different channels. You manually track consent on spreadsheets. You guess at engagement patterns because your analytics do not talk to each other. And when a user asks to be forgotten under GDPR, you are scrambling to find every place their data lives.

Rylvo Audience Hub was built to solve exactly this problem. It brings every end-user across every bot and every channel into one unified, powerful workspace. But it is not just a user list. It is a complete audience intelligence platform that scales effortlessly to hundreds of thousands of users, gives you deep insights into how people engage with your bots, and keeps you fully compliant with GDPR, CCPA, and other privacy regulations — all without writing a single line of code.

In this comprehensive guide, we will walk through every feature of the Audience Hub, explain how it works under the hood, show you real-world use cases, and help you understand why it is the best AI audience management platform for enterprises that take their conversational strategy seriously.

What Is the Rylvo Audience Hub and Why Do You Need It?

The Rylvo Audience Hub is your central command center for everything related to the people who interact with your AI bots. Think of it as a CRM that was purpose-built for conversational AI from day one. Traditional customer relationship management tools like Salesforce or HubSpot were designed for human sales teams tracking leads through pipelines. They do not understand the unique dynamics of bot conversations: the multiple channels, the automated data collection, the consent flows, and the real-time engagement patterns.

The Audience Hub replaces this fragmented approach with a unified, server-driven interface that handles the complete lifecycle of every end-user who talks to your bots:

Discovery — Instantly see who is talking to your bots, when they last interacted, which channel they prefer, and how many conversations they have had. No more guessing or cross-referencing multiple dashboards.

Organization — Tag users with custom labels, build intelligent audience segments based on behavior and attributes, and group people in ways that make sense for your business. Want all WhatsApp users who have had at least five conversations and reached the "purchase intent" checkpoint? Two clicks.

Analysis — Understand engagement patterns with real-time analytics. See where users drop off in your onboarding funnel. Know which channels drive the most active users. Monitor profile completion rates to optimize your data collection flows.

Compliance — Track granular consent scopes for every user. Process GDPR right-to-be-forgotten requests with one click. Maintain complete audit trails for SOC 2 and regulatory inspections.

Operations — Import your existing contact database via CSV. Export any segment for analysis in BigQuery, Snowflake, or Excel. Run bulk actions on thousands of users at once.

Unlike legacy CRMs that treat bot users as an afterthought, the Audience Hub understands that conversational AI is different. A single person might interact with your sales bot on WhatsApp, your support bot on your website, and your internal HR bot on Slack. The Audience Hub connects these identities automatically and shows you the complete picture.

The Six Workspaces of the Rylvo Audience Hub Explained in Detail

The Audience Hub is organized into six focused workspaces. Each one is designed for a specific operational need, and together they give you complete control over your bot audience. Let us explore each one in depth.

1. People Workspace — Your Complete End-User Directory

The People workspace is the beating heart of the Audience Hub. When you first open it, you see a clean, paginated table showing every end-user who has ever interacted with any of your bots, on any channel. This is where you go when you need to find someone, understand their history, inspect their profile, or take action on groups of users.

Server-Driven Pagination That Scales to Hundreds of Thousands of Users

One of the biggest challenges with user directories is performance. If you have ten thousand users, a traditional web application might try to load all of them into the browser at once. This creates massive lag, crashes on mobile devices, and makes the interface unusable. Even worse, traditional offset pagination — where you load page 1 with users 1-50, page 2 with users 51-100 — becomes slower with every page because the database has to scan and discard all the rows before your target page.

The Rylvo People workspace solves this with cursor-based pagination. Here is how it works in simple terms: instead of saying "give me page 5," the system says "give me the next 50 users after the last one I saw." The cursor is an opaque base64-encoded token that contains the sort value and document ID of the last user on the current page. This means page 1,000 loads exactly as fast as page 1, because the database never has to scan through previous pages.

The page loads 50 users at a time by default, with infinite scroll support. As you scroll down, the next chunk loads seamlessly. There is zero client-side lag because the server does all the heavy lifting. Whether you have 500 users or 500,000 users, the experience remains snappy and responsive.

14-Channel Filtering for Multi-Platform User Tracking

Modern AI bots do not live on a single channel. Your customers might start on your website widget, move to WhatsApp for quick questions, and occasionally reach out via Instagram. The People workspace supports filtering by 14 different channels:

• Widget — Users who interacted through your embedded web chat
• WhatsApp — Users who messaged your WhatsApp Business API number
• Slack — Workspace members who used your Slack app
• Telegram — Users who found your bot on Telegram
• Discord — Community members on your Discord server
• MS Teams — Corporate users on Microsoft Teams
• Messenger — Facebook Messenger conversations
• Instagram — Direct messages and story interactions
• Email — Users who engaged via email-based bot flows
• SMS — Text message interactions
• Voice — Voice assistant and IVR callers
• API — Programmatic integrations and webhook users
• Custom — Any other channel you have configured

Simply select a channel from the filter dropdown, and the table instantly refreshes to show only users from that platform. This is incredibly powerful for understanding channel preferences. Are your enterprise customers mostly on MS Teams while your e-commerce buyers prefer WhatsApp? The data is right there.

Consent Scope Filtering for Privacy Compliance Visibility

Every user in your Audience Hub has a consent scope that defines how much personalization they have agreed to. The People workspace lets you filter by these scopes so you can quickly audit your privacy posture:

• None — The user has not granted any personalization. Their data is kept to an absolute minimum. These users see generic bot responses without any history or context.
• Short-term only — The user allows session-level personalization. The bot remembers them during the current conversation, but nothing is stored long-term. When they come back tomorrow, it is a fresh start.
• Personalize — The user agrees to long-term personalization. The bot collects profile information, remembers past conversations, and tailors responses based on historical data.
• Full — The user consents to complete data collection, including profile building, analytics tracking, and usage data for improving the bot over time.

Being able to filter by consent scope is essential for compliance. If you are running a promotional campaign that requires personalization, you can quickly see exactly how many users have granted the necessary consent level. If a regulator asks for a count of users with minimal data collection, you have that number in seconds.

Forgotten Status and GDPR Workflow Management

The People workspace includes a forgotten status filter with three modes:

• Exclude forgotten — Show only active users (default view)
• Show all — Include both active and forgotten users
• Only forgotten — Show users who have invoked their right to be forgotten

This last mode is crucial for GDPR compliance. When you switch to "Only forgotten," you see a complete queue of users awaiting data deletion. The retention sweeper processes these on a schedule, but being able to inspect the queue gives you visibility and control. You can verify that forget requests are being honored, check timestamps, and even inspect individual records before deletion completes.

Real-Time Search by External ID or Display Name

Need to find a specific user? The search bar at the top of the People workspace searches by external ID or display name in real time. Type a phone number, email fragment, or name, and results filter instantly. This is invaluable for support teams who need to look up a specific customer quickly.

The Inspector Drawer — Deep-Linked User Profiles

Clicking any row in the People table opens the inspector drawer — a slide-out panel that shows the complete user profile without navigating away from the list. The drawer displays:

• Profile overview — External ID, display name, channel, consent scope, and forgotten status
• Conversation metadata — Total turns, first seen date, last seen date, and retention period
• Tags — All custom labels applied to this user
• Bot profiles — Per-bot profile data showing what each bot has collected about this user
• Consent history — A timeline of consent changes and operator overrides

The URL automatically updates when you open the inspector, creating a deep link like /dashboard/audience/people?user=john_doe_123. You can copy this URL, share it with teammates, or bookmark it for quick access later. Hit the browser back button, and the drawer closes smoothly.

Bulk Operations on Selected Users

Select multiple users by clicking checkboxes, and a bulk action bar appears at the top of the table. From here, you can perform powerful operations on your selection:

Export to CSV — Download all selected users with their complete data fields. The CSV includes external ID, display name, channel, consent scope, turn count, tags, forgotten status, first seen date, and last seen date. This is perfect for importing into Excel for ad-hoc analysis, loading into BigQuery for warehouse storage, or sharing with business intelligence teams.

Bulk Tag Management — Add or remove tags on multiple users simultaneously. For example, select all users from WhatsApp who have had more than ten conversations, add the tag vip_customer, and now you have a ready-made segment for targeted broadcasts.

Send to Broadcast — Seed selected recipients directly into a bot's broadcast composer. Choose any bot from a dropdown, and the system stages your selected users as the recipient list. Click Continue, and you are taken to the broadcast composer with everything pre-filled.

Forget Users (GDPR) — Process right-to-be-forgotten requests in bulk. A confirmation dialog explains that this will revoke consent and schedule personalization data for deletion. Once confirmed, the users immediately move to the forgotten queue, and the retention sweeper handles the rest.

2. Segments Workspace — Intelligent Audience Slicing and Targeting

Segments are one of the most powerful features in the Audience Hub. They let you define reusable filters that group users by behavior, attributes, or engagement patterns. Once created, segments power broadcasts, exports, and analytics across the entire Rylvo platform.

Think of segments as saved search queries that update automatically. Instead of manually filtering the People table every time you want to message your VIP customers, you create a segment once and reuse it everywhere.

The Eight Segment Types Explained with Real-World Examples

Rylvo supports eight distinct segment types, each designed for a specific targeting need:

All Users Segment — This is your baseline. It includes every end-user in your organization who has interacted with any bot on any channel. Use this when you want to broadcast a system-wide announcement, run an org-wide survey, or export your complete user base for backup or migration purposes.

Active Within Days Segment — Target users who have interacted with your bots within a specific time window. For example, create a segment of users active in the last 7 days to send them a "We miss you" re-engagement message. Or target users active in the last 30 days for a monthly newsletter. The day count is fully configurable.

By Channel Segment — Filter users by their primary communication channel. Want to send a WhatsApp-specific promotion? Create a "By Channel" segment set to WhatsApp. Need to notify all Slack workspace members about a new feature? Create a Slack segment. This is especially useful because messaging content and tone often need to adapt to the channel.

Minimum Conversations Segment — Identify power users and highly engaged audience members. Set a threshold like "at least 5 conversations" to find your most active users, then reward them with exclusive content or early access. Or set "at least 1 conversation" to exclude completely inactive signups from a campaign.

Checkpoint Achieved Segment — This is where bot conversation intelligence really shines. Checkpoints are milestones you define in your bot's conversation flow. Examples include: email_collected, phone_verified, demo_booked, purchase_completed, onboarding_finished. Create a segment of users who reached demo_booked to send them a reminder before their demo. Or target users who completed onboarding_finished with a satisfaction survey.

Field Value Segment — Target users based on specific profile data they have provided. This segment supports three operations:

• Is set — The field exists and has a non-empty value. For example, "users who have provided their email address"
• Equals — The field exactly matches a value. For example, "users whose country field equals 'United States'"
• Contains — The field contains specific text. For example, "users whose company name contains 'Inc'"

Completeness Segment — Profile completeness measures how much of a bot's collection schema a user has filled out. If your bot collects name, email, phone, and company, a user who provided all four has 100% completeness. A user who only gave their name has 25% completeness. Create segments like "completeness >= 80%" to find users who are nearly fully profiled and might be ready for a sales call.

Compound Segment — This is the powerhouse. Combine any of the above segment types using AND or OR logic for sophisticated multi-dimensional targeting. For example:

• AND compound: "Active within 30 days" AND "By Channel = WhatsApp" AND "Checkpoint = purchase_completed" — gives you recent WhatsApp buyers
• OR compound: "By Channel = Email" OR "By Channel = SMS" — gives you all users reachable by either email or text

Compound segments support unlimited nesting, so you can build incredibly precise audience slices for any business need.

How Segment Resolution Works Under the Hood

The segment resolver is a critical piece of Rylvo's architecture. It is the single source of truth for turning a segment definition into an actual list of users. Here is why it is special:

Firestore, the database powering Rylvo, has limitations on composite indexes. You cannot easily create an index that covers "channel = WhatsApp AND checkpoint = email_collected AND completeness > 50%." Most platforms would struggle with this query or require expensive index management.

Rylvo solves this through denormalized flags. Every time a user achieves a checkpoint or fills a profile field, the system automatically sets a flag on their profile document. For example:

• User verifies their email → flags.checkpoint_email_verified = true is set
• User provides their phone number → flags.field_phone_set = true is set
• User completes onboarding → flags.checkpoint_onboarding_finished = true is set

These flags live in a flat map on each profile document. The segment resolver can query any flag with a simple equality filter, which Firestore handles efficiently without requiring custom composite indexes. This means segments resolve in milliseconds, even across large user bases.

When you preview a segment, the resolver returns up to 5,000 matching users. Each result includes the user's channel, display name, conversation turn count, and profile completeness score. This preview is what powers the broadcast composer, the export panel, and the segment preview in the People workspace.

3. Profiles & Schema Workspace — Master Your Data Model

As your organization scales and you deploy more bots, data consistency becomes a serious challenge. One bot might collect "email" as a field. Another bot might use "email_address." A third might use "user_email." Before you know it, you have three different field names for the same piece of information, your analytics are broken, and your exports have missing columns.

The Profiles & Schema workspace was built to prevent exactly this problem. It gives you a global, read-only view of every collection schema across every bot in your organization.

The Field Catalog — Your Org-Wide Data Dictionary

The field catalog is a flattened table showing every field key used by any bot in your workspace. It is sorted by popularity, so the most commonly used fields appear first. For each field, you can see:

• Field key — The exact identifier, displayed in monospace for clarity
• Data types — What types are assigned (string, number, boolean, date, etc.)
• Used by — A list of every bot that uses this field, with clickable links to jump to that bot's configuration
• Required somewhere — A checkmark if any bot marks this field as required

This single view immediately surfaces naming inconsistencies. If you see "email," "email_address," and "user_email" all in the catalog, you know you need to standardize. The catalog makes it easy to identify which bots need updating and which field name should become your organizational standard.

Per-Bot Schema Cards

Below the field catalog, the workspace displays a grid of cards — one for each bot in your organization. Each card shows:

• Bot name — Clickable link to the bot's Collection tab for editing
• Fields — All configured fields with their types, descriptions, required status, and whether they have custom validators
• Checkpoints — All conversation milestones defined for this bot, including prerequisites

Checkpoints deserve special mention. A checkpoint is a milestone in a conversation flow that marks a specific achievement. For example, a sales bot might have checkpoints like lead_captured, qualification_complete, demo_scheduled, and deal_closed. Each checkpoint can have prerequisites — "deal_closed" might require that qualification_complete and demo_scheduled were already achieved. The schema cards visualize these dependencies so you can understand your conversation architecture at a glance.

Why This Matters for Analytics and Integrations

Clean, consistent data models are the foundation of reliable analytics. If your export contains "email" for some users and "email_address" for others, your data warehouse cannot match records properly. If your segment filters look for country but half your bots use country_code, your targeting will miss users.

The Profiles & Schema workspace turns this invisible problem into a visible, actionable dashboard. You spot drift early, standardize field names, and ensure that every bot in your organization speaks the same data language.

4. Consent & Privacy Workspace — GDPR Compliance Made Simple

Privacy regulations like GDPR in Europe, CCPA in California, and LGPD in Brazil have made user consent management a legal requirement, not a nice-to-have. The Consent & Privacy workspace gives you complete visibility into your users' privacy choices and the tools to handle their rights efficiently.

Understanding the Four Consent Scopes

Rylvo implements a four-tier consent model that gives users granular control over how their data is used. Each tier is more permissive than the last:

None — Maximum Privacy When a user selects "None," the bot operates in a completely stateless mode. It does not remember previous conversations, does not collect profile data, and does not store any personalization information. Every interaction is treated as the first interaction. This mode is appropriate for users who want maximum privacy or for regulated industries where data collection is restricted.

Short-Term Only — Session Memory Users with "Short-term only" consent allow the bot to remember context during the current conversation. If they ask "What did I just order?" the bot can reference the previous turn. But when the session ends, everything is wiped. The next time they return, the bot starts fresh. This is ideal for users who want helpful responses but do not want long-term data storage.

Personalize — Long-Term Memory Users who grant "Personalize" consent allow the bot to build and maintain a profile over time. The bot collects fields defined in the collection schema, remembers conversation history, and uses this information to provide tailored responses. If a user previously mentioned they prefer email over phone calls, the bot remembers that. This is the standard consent level for most business bot interactions.

Full — Complete Data Collection "Full" consent enables everything in Personalize plus analytics and improvement data. The bot can use conversation patterns to improve its responses, and your team can analyze aggregated behavior to optimize conversation flows. This is the highest consent level and should be clearly explained to users so they understand what they are agreeing to.

The Consent Dashboard — Health at a Glance

The top of the Consent & Privacy workspace shows a row of tiles displaying the total count for each consent scope. At a glance, you can see:

• How many users have not granted any consent
• How many are using session-only mode
• How many have agreed to personalization
• How many have granted full data collection

This dashboard helps you monitor consent health over time. If you see a sudden spike in "None" consents after a privacy policy update, you know the change confused users. If "Full" consents grow steadily, your trust-building efforts are working.

GDPR Forget Queue — Managing Right-to-Be-Forgotten Requests

When a user invokes their right to be forgotten under GDPR Article 17, they appear in the forgotten users queue. This queue is a critical compliance tool that shows:

• External ID — The user's unique identifier, with a clickable link to inspect their record
• Channel — Where the user originally interacted (WhatsApp, Widget, etc.)
• Last consent scope — What level of consent they had before requesting deletion
• Forgotten at — The exact timestamp when the forget request was processed

Being able to inspect the queue gives you several important compliance capabilities:

1. Verification — Confirm that forget requests are being received and processed
2. Audit evidence — Show regulators that you have a systematic process for handling deletion requests
3. Error detection — Spot users who should have been forgotten but were missed due to a system error
4. Timeline tracking — Ensure requests are processed within your committed timeframe

The retention sweeper runs on a configurable schedule (typically daily) to purge forgotten users' personalization data. It removes profile information, conversation history, and collected fields while preserving the minimal audit trail required for legal compliance. The queue shows you exactly what stage each user is at in this process.

Upcoming Audit Export Feature

A full audit export is planned that will generate a CSV of every consent change, operator override, and forget action over any date range you specify. This export will include:

• Timestamp of each action
• User external ID (where permitted)
• Action type (consent granted, consent revoked, operator override, forget processed)
• Consent scope before and after
• Operator identity (for manual overrides)

This feature is designed specifically for SOC 2 Type II audits, GDPR Article 30 records of processing activities, and general regulatory compliance documentation.

5. Insights Workspace — Real-Time Audience Analytics and Funnel Tracking

The Insights workspace transforms raw user data into actionable business intelligence. Instead of running expensive database queries every time you open the page, Rylvo uses a smart caching strategy that makes analytics instant while keeping data fresh.

How the Insights Caching System Works

Here is the challenge: computing analytics over a large user base in real time is expensive. If you have 50,000 users, calculating an onboarding funnel requires scanning every user record, checking their consent status, looking up their profiles, counting checkpoints, and calculating completion percentages. Doing this on every page load would hit Firestore rate limits and make the interface unusable.

Rylvo's solution is a cached snapshot system:

1. When you first open Insights, the system checks for a cached snapshot in Firestore
2. If the snapshot exists and is less than 6 hours old, it serves the cached data immediately
3. If the snapshot is missing or stale, the system computes fresh analytics and stores them as a new snapshot
4. You can trigger a manual recompute at any time by clicking the "Recompute" button

This means the page loads instantly, even for massive organizations. The "stale" indicator tells you when the data is more than 6 hours old, so you always know how fresh your numbers are.

The Complete Analytics Dashboard

The Insights workspace includes seven powerful analytics views:

Totals Dashboard — Five key metrics at the top of the page:

• Total end-users across all bots and channels
• Total collection profiles (users who have provided structured profile data)
• Active users in the last 7 days
• Active users in the last 30 days
• Total forgotten users awaiting deletion

These numbers give you a health check for your audience in under three seconds.

Onboarding Funnel — This is arguably the most valuable analytics view. It visualizes your user's journey through five stages:

1. Known end-users — Everyone who has ever interacted with a bot
2. Gave consent — Users who granted at least "Short-term only" consent
3. Has a collection profile — Users who have provided structured data through a bot's collection schema
4. Reached a checkpoint — Users who achieved at least one conversation milestone
5. All required fields collected — Users with 100% profile completeness

Each stage shows the absolute count and the drop-off percentage from the previous stage. This immediately reveals where users are falling out of your funnel. If 10,000 users are known but only 3,000 gave consent, you have a consent friction problem. If 3,000 gave consent but only 500 have profiles, your collection schema might be too aggressive or poorly timed.

New Users by Day — A 30-day bar chart showing daily user acquisition velocity. Each bar represents one day, with height proportional to the number of new first-seen users. A trend line at the bottom shows the date range, and the total count is displayed prominently. This chart helps you identify growth spikes (maybe after a marketing campaign) and spot declining acquisition early.

Channel Mix — A ranked list showing where your users come from, with absolute counts. For example:

• WhatsApp: 4,532 users
• Widget: 2,891 users
• Slack: 1,204 users
• Email: 876 users

This helps you prioritize channel investment. If 60% of your users are on WhatsApp but you are spending most of your development effort on the web widget, you might want to rebalance.

Consent Breakdown — A distribution chart showing how many users fall into each consent scope. Monitor this over time to ensure your consent flows are effective and users are not defaulting to "None" because the prompt is confusing or pushy.

Completeness Distribution — A histogram showing how far users get through your collection schemas, bucketed into six ranges: 0%, 1-25%, 26-50%, 51-75%, 76-99%, and 100%. If most users are stuck in the 1-25% bucket, your schema might be asking for too much too soon. If a healthy percentage reaches 100%, your data collection flow is well-designed.

Field-Fill Rates — A ranked list of all your collection fields showing what percentage of profiles have a non-empty value for each field. This is incredibly useful for schema optimization. If "company_name" has a 12% fill rate but "email" has an 89% fill rate, you know most users are willing to share their email but hesitate to provide company information. You might move "company_name" to a later checkpoint or make it optional.

Performance at Enterprise Scale

When your organization exceeds 10,000 users, the insights computation samples the dataset rather than scanning every record. This keeps response times under control while still providing statistically meaningful results. The UI clearly flags when a report is "based on a sample," so you are never misled. For critical decisions, you can always trigger a full recompute.

6. Imports & Exports Workspace — Bridge Your Existing Data

Most organizations already have user data somewhere: a CRM like Salesforce, a customer database, an email marketing platform, or a spreadsheet. The Imports & Exports workspace makes it easy to bring this data into Rylvo and to extract Rylvo data for use in other systems.

CSV Import Wizard — Four Steps to Migration

The import process is designed to be safe, predictable, and repeatable. Here is exactly how it works:

Step 1: Upload Your CSV File Drag and drop any CSV file up to 15 MB onto the upload area. The system immediately parses the file and shows you:

• Number of data rows detected
• Number of columns found
• A preview of the first few rows

If the file is larger than 15 MB, the system prompts you to split it into smaller batches. This prevents timeout issues and ensures reliable processing.

Step 2: Map Your Columns The system analyzes your CSV headers and attempts to auto-map them to Rylvo fields. You see a mapping interface with five target fields:

• External ID (required) — This is the unique identifier for each user. It might be a customer ID from your CRM, a phone number, or an email address. Every row must have an external ID.
• Channel — Where this user primarily interacts. Options include widget, whatsapp, slack, telegram, discord, msteams, messenger, instagram, email, sms, voice, api, or custom.
• Display Name — The user's human-readable name.
• Tags — Custom labels to apply, separated by commas, pipes, or semicolons.
• Consent Scope — The user's initial consent level: none, short_term_only, personalize, or full.

You can override any auto-mapped column manually. If the system guesses wrong, simply select the correct header from the dropdown.

Step 3: Preview the Import Plan Before committing, you see a complete preview showing:

• New users — How many rows will create brand new user records
• Updates — How many rows will update existing users (matched by external ID)
• Errors — How many rows have problems (missing external ID, invalid channel, etc.)

A detailed table shows the first 100 rows with their predicted action (create, update, or error) so you can spot issues before they affect your data.

Step 4: Commit the Import Click the import button, and the system processes all rows transactionally. Because the import is idempotent — keyed by external ID — re-running the same file never creates duplicates. Existing users are updated with new information, and new users are created. This makes it completely safe to run periodic syncs from external systems.

Segment Export for Data Warehouses and Analysis

The export side of the workspace lets you resolve any saved segment (or your entire audience) and download it as a CSV. The export includes:

• External ID
• Display name
• Primary channel
• Conversation turn count
• Profile completeness percentage
• Last seen timestamp

Each export is capped at 5,000 users to ensure fast response times. For larger exports, you can either export in segments or use the API directly. The exported CSV is ready for immediate import into:

• Google BigQuery — For large-scale analytics and SQL queries
• Snowflake — For enterprise data warehousing
• Microsoft Excel — For ad-hoc analysis and reporting
• Tableau or Looker — For business intelligence dashboards
• Python Pandas — For data science and machine learning workflows

Technical Architecture: How the Audience Hub Scales to Enterprise

The Audience Hub is not a simple user list backed by basic database queries. It is built on architectural decisions that specifically address the challenges of managing large bot audiences at scale.

Cursor-Based Pagination: Why Offset Pagination Fails

Most web applications use offset pagination, where page 1 shows rows 1-50, page 2 shows rows 51-100, and so on. This works fine for small datasets but falls apart at scale because the database must scan and discard all rows before your target page. Page 100 requires scanning 5,000 rows and throwing away 4,950 of them. As tables grow, every subsequent page gets slower.

The Audience Hub uses cursor pagination instead. The cursor is an opaque token that encodes the sort value and document ID of the last item on the current page. To get the next page, the system asks for "the next 50 items after this cursor." The database jumps directly to that position using its index, without scanning any previous rows. This means page 1,000 is exactly as fast as page 1, because the database never has to look at earlier pages.

The cursors are base64-encoded JSON for tamper resistance and are stable even if the underlying data changes between page loads.

Denormalized Flags: Solving Firestore's Index Limitations

Google Cloud Firestore, the database that powers Rylvo, has powerful querying capabilities but also limitations. One of the biggest is composite index management: if you want to query by channel AND checkpoint AND completeness, you need a composite index covering all three fields. Managing indexes for every possible combination becomes unwieldy and expensive.

Rylvo solves this through denormalized flags on each user profile document. Instead of querying "find users where checkpoint_email_verified is true," the system stores flags.checkpoint_email_verified = true in a flat map. This means every segment predicate becomes a simple equality query on a single field, which Firestore handles efficiently with automatic single-field indexes.

When a user achieves a checkpoint or fills a profile field, the bot runtime automatically updates the corresponding flag. This happens transparently — bot builders never need to think about it. But the result is that complex segments resolve in milliseconds without any index management overhead.

Cached Insights with Smart Staleness Detection

Computing analytics over a 100,000-user organization involves multiple collection scans, aggregation calculations, and histogram generation. Running this on every page load would be prohibitively expensive and would quickly exhaust database read quotas.

The Insights API computes a complete analytics snapshot and stores it as a single document in Firestore. This document is refreshed automatically every 6 hours and can be manually refreshed on demand. When you open the Insights page, you see the cached data instantly — no loading spinners, no waiting.

The system also detects staleness. If the cached snapshot is older than 6 hours, a subtle "stale" badge appears, inviting you to recompute. For most daily monitoring, 6-hour-old data is perfectly adequate. For critical decisions, one click triggers a fresh computation.

Collection-Group Queries for Cross-Bot User Profiles

In Rylvo's data model, each user has a subcollection of profiles — one profile document per bot they have interacted with. This means a user who has talked to three different bots has three profile documents, each stored at /organizations/{orgId}/endUsers/{userId}/profiles/{botId}.

Collection-group queries let the system search across all these profile documents in a single query, scoped to your organization. Want to find all users whose "sales_bot" profile has completeness > 80%? The system fires one collection-group query and gets the results in a single round trip. Without collection groups, this would require querying each user individually — an N+1 problem that would be impossibly slow at scale.

Idempotent Import with Transactional Safety

The CSV importer is designed for safety and repeatability. Every import operation is keyed by external ID, which means:

• If a row's external ID already exists, the user is updated, not duplicated
• If a row's external ID is new, a user is created
• If the same CSV is imported twice, the result is identical to importing it once

This idempotency is critical for production workflows. You can schedule nightly syncs from your CRM to Rylvo without worrying about creating duplicate records. The system also validates every row before committing, so bad data never corrupts your user base.

Privacy-First Design: Compliance Built In from Day One

The Audience Hub was architected with privacy compliance as a foundational requirement, not an afterthought. Every feature considers the user's right to control their data.

Granular Consent Control

The four-tier consent model (None, Short-term only, Personalize, Full) gives users meaningful control over their data. It is not a binary "accept all or leave" choice. Users can opt for session-only memory if they want helpful conversations without long-term tracking. They can upgrade to full personalization if they trust your brand. This granularity actually increases consent rates because users feel in control.

Automated Right-to-Be-Forgotten Processing

GDPR Article 17 gives users the right to erasure. The Audience Hub makes honoring this request a one-click operation. When you forget a user:

1. Their consent is immediately revoked
2. Their profile data is queued for deletion
3. The retention sweeper removes personalization data on schedule
4. An audit log entry records the action for compliance evidence

The system preserves only the minimum information required for legal compliance (such as proof that the request was honored), ensuring you meet regulatory requirements without retaining unnecessary data.

Role-Based Access Control

Not everyone in your organization should be able to view user data or process forget requests. The Audience Hub respects Rylvo's role-based permission system:

• Viewers can see the People list, segments, and insights but cannot modify data
• Editors can add tags, import users, and create segments
• Admins can process forget requests and access the full consent audit trail

This ensures that user data is only accessible to people who genuinely need it, reducing the risk of unauthorized access or data breaches.

Configurable Retention Policies

Every organization can configure its own data retention policies. You might decide to delete inactive users after 90 days, purge conversation logs after 30 days, or retain profile data indefinitely for active customers. The retention sweeper enforces these policies automatically, so you never have to manually clean up old data.

Real-World Use Cases and Industry Applications

The Audience Hub is versatile enough to serve businesses across many industries. Here are detailed examples of how different organizations use it.

E-Commerce and Retail

An online fashion retailer uses Rylvo bots on WhatsApp, Instagram, and their website widget. The Audience Hub lets them:

• Track customers across all three channels, discovering that 30% of WhatsApp users also use the web widget
• Build a segment of "users who reached the checkout checkpoint but did not complete purchase" and send them an abandoned cart recovery broadcast
• Use the onboarding funnel to discover that 40% of users drop off at the consent prompt, leading them to redesign the consent flow with clearer language
• Export high-value segments (completeness > 80%, active within 30 days) to their email marketing platform for coordinated campaigns
• Process GDPR forget requests from European customers through the forget queue

Healthcare and Telemedicine

A telemedicine provider uses Rylvo bots for patient intake, appointment scheduling, and post-visit follow-up. They rely on the Audience Hub for:

• Strict consent tracking, ensuring HIPAA-aligned data handling where "Full" consent is never assumed
• Monitoring the forget queue to ensure patient data removal requests are processed within 48 hours
• Using the field catalog to standardize patient data collection across intake bots, ensuring "date_of_birth" is collected consistently rather than having variants like "dob" and "birth_date"
• Building segments of patients who completed the "prescription_renewal" checkpoint for automated refill reminders
• Analyzing the channel mix to discover that elderly patients prefer SMS while younger patients use the web widget

SaaS and Technology Companies

A B2B SaaS company uses Rylvo bots for onboarding new users, providing in-app support, and handling billing inquiries. Their Audience Hub workflow includes:

• Importing their existing user base from PostgreSQL via CSV, preserving customer IDs as external IDs
• Tagging power users (minimum 20 conversations, reached "advanced_feature_discovery" checkpoint) as champions
• Monitoring the onboarding funnel to identify that most users drop off at the "API_key_generated" checkpoint, leading to improved documentation
• Exporting monthly active user segments to their data warehouse for churn prediction modeling
• Tracking consent scopes to ensure users who selected "Short-term only" do not receive persistent personalization they did not agree to

Enterprise Customer Support

A multinational enterprise uses Rylvo bots for tier-1 support across Slack, MS Teams, and email. Their support team uses the Audience Hub to:

• Filter the People workspace by channel to see which platforms different business units prefer
• Build segments for VIP customers (identified by custom tags imported from their CRM) and route them to premium support bots with faster response times
• Use the insights funnel to measure how many support inquiries reach the "satisfied" checkpoint (collected via post-resolution feedback)
• Export weekly active user data to their BI platform for support volume forecasting
• Process GDPR forget requests from former employees whose support history should no longer be retained

Marketing Agencies and Consultancies

A digital marketing agency manages bot campaigns for multiple clients using Rylvo. They leverage the Audience Hub to:

• Create compound segments for A/B testing: Group A gets "active within 7 days AND channel = WhatsApp AND checkpoint = product_viewed" while Group B gets the same but with "channel = Instagram"
• Monitor channel mix across clients to advise on platform strategy
• Use field-fill rates to recommend schema optimizations to clients
• Export segment data to create custom reports showing campaign reach and engagement
• Ensure all client data complies with GDPR through the consent and forget queue dashboards

Getting Started with the Rylvo Audience Hub: A Step-by-Step Guide

Ready to explore your audience? Here is exactly how to get started:

Step 1: Open the Audience Hub

Navigate to /dashboard/audience in your Rylvo workspace. You will land on the People view, which shows all end-users who have interacted with your bots.

Step 2: Explore Your User Base

Start by scrolling through the People list. Notice the pagination loading seamlessly as you scroll. Try the search bar to find a specific user by name or ID. Click on any user row to open the inspector drawer and see their complete profile, conversation history, and tags.

Step 3: Filter by Channel

Open the filter panel and select a specific channel like "WhatsApp" or "Widget." See how the user count changes. This gives you an immediate sense of where your audience is concentrated.

Step 4: Check Your Consent Distribution

Switch to the Consent & Privacy workspace. Look at the consent tiles. Are most users on "None," "Personalize," or "Full"? This tells you how trusting your audience is and whether your consent flows need improvement.

Step 5: Create Your First Segment

Go to the Segments workspace and click "Create Segment." Start simple: choose "Active within days" and set it to 30. Give it a name like "Active Users — Last 30 Days." Save it. Now you have a reusable segment you can use in broadcasts and exports.

Step 6: Analyze Your Onboarding Funnel

Open the Insights workspace and look at the onboarding funnel. Where is the biggest drop-off? Between "Known" and "Gave consent"? That suggests your consent prompt needs work. Between "Has profile" and "Reached checkpoint"? Maybe your conversation flow is too long before the first milestone.

Step 7: Import Legacy Contacts (Optional)

If you have existing user data in a CSV, go to Imports & Exports. Upload the file, map the columns, preview the results, and commit the import. Your existing contacts are now part of your Rylvo audience.

Step 8: Review Your Data Model

Visit the Profiles & Schema workspace. Look at the field catalog. Do you see multiple variants of the same field? Make a note to standardize them in your bot configurations.

The Audience Hub is available to all Rylvo organizations with appropriate role permissions. No additional setup is required — it works automatically with data already collected by your bots.

Frequently Asked Questions About the Rylvo Audience Hub

How many users can the Audience Hub handle? The Audience Hub is designed to scale to hundreds of thousands of users per organization. The cursor-based pagination system ensures the People workspace stays fast regardless of user count. Insights sampling kicks in at 10,000 users to maintain performance while preserving statistical accuracy.

Is my user data secure? Yes. All data is stored in Google Cloud Firestore with enterprise-grade encryption at rest and in transit. Access is controlled through Firebase Authentication and Rylvo's role-based permission system. Only users with explicit permissions can view, edit, or delete audience data.

Can I import users from my existing CRM? Absolutely. The CSV import wizard supports files up to 15 MB and automatically maps common column headers. The import is idempotent, so you can safely re-run the same file or schedule periodic syncs without creating duplicates.

What happens when a user requests to be forgotten? When you process a forget request, the user's consent is immediately revoked and their personalization data is queued for deletion. The retention sweeper removes profile data, conversation history, and collected fields on schedule while preserving the minimal audit trail required for compliance.

Can I export segments for use in other tools? Yes. Any segment (or your entire audience) can be exported as a CSV with fields including external ID, display name, channel, turn count, completeness, and last seen date. Exports are capped at 5,000 users for fast response times.

How fresh are the analytics in Insights? Analytics snapshots are computed every 6 hours and cached for instant loading. A "stale" indicator appears when data is older than 6 hours. You can trigger a manual recompute at any time with one click.

Do I need to configure anything to start using the Audience Hub? No. The Audience Hub works out of the box with data already collected by your bots. As users interact with your bots, they automatically appear in the People workspace.

Can I segment users by custom fields I collect? Yes. The Field Value segment type lets you filter users by any field in their profile, with operations including "is set," "equals," and "contains." Combined with compound segments, this enables incredibly precise targeting.

What channels are supported? Rylvo supports 14 channels: Widget, WhatsApp, Slack, Telegram, Discord, MS Teams, Messenger, Instagram, Email, SMS, Voice, API, and Custom. The Audience Hub can filter and segment users by any of these channels.

Is the Audience Hub GDPR compliant? Yes. The Audience Hub includes granular consent tracking, a forget queue for right-to-be-forgotten requests, role-based access control, configurable retention policies, and audit trail logging. It is designed to help you meet GDPR, CCPA, and other privacy regulation requirements.

Conclusion: Transform Scattered Bot Interactions into a Cohesive Audience Strategy

The Rylvo Audience Hub is more than a user directory. It is a complete audience intelligence platform that gives you the visibility, control, and compliance capabilities you need to manage bot interactions at any scale.

With six specialized workspaces — People, Segments, Profiles & Schema, Consent & Privacy, Insights, and Imports & Exports — you get everything from real-time user discovery to deep analytics, from intelligent segmentation to automated privacy compliance. The enterprise-grade architecture ensures it stays fast whether you have 500 users or 500,000.

Whether you are an e-commerce brand tracking customers across WhatsApp and Instagram, a healthcare provider ensuring HIPAA-aligned data handling, a SaaS company optimizing onboarding funnels, or an enterprise support team routing high-value customers to premium bots, the Audience Hub provides the tools you need.

Stop managing your bot audience in spreadsheets and disconnected dashboards. Start exploring your audience today at /dashboard/audience and discover what complete conversational intelligence looks like.