Skip to main content

Customer Onboarding & Retention

Effective Onboarding of New Paying Customers: Retain Their Enthusiasm and Ensure Long-Term Satisfaction​

Acquiring a paying customer is an important success, but the real challenge and opportunity lies in having them stay with you long-term, actively use your AI expert, and continuously perceive its enormous value. The key to this is thoughtful and effective onboarding after purchase. Remember that for many of your customers, interacting with such an advanced and personalized AI may be a new experience, and therefore they need your guidance.

Quality Educational Materials -- Teach Them to "Fish" with Your AI​

Your goal is to equip customers with the knowledge and confidence to use your AI expert to the maximum.

Create an Information-Rich Onboarding Video​

Recommended length is 10-20 minutes. The video should be concise but packed with practical information.

What the onboarding video should contain:

  • πŸŽ‰ Warm welcome: Thank them for their trust and purchase.
  • πŸ’Ž Reiteration of key benefits: Briefly remind them how your AI expert can help them.
  • πŸ‘€ Introduction to BuddyPro's shared user profile
  • πŸ’¬ Practical demonstration of effective communication
  • πŸ“‹ Examples of typical tasks and topics: Inspire them with all the things they can ask their AI expert about in your field. Show the breadth of its capabilities.
  • πŸŽ™οΈ Using specific features: Demonstrate how to use voice messages (if relevant to your target audience), how to ask AI to remind about a topic in the future (verify current available features -- the user can tell AI e.g. "remind me in a week to discuss X").
  • βš–οΈ Setting realistic expectations
  • πŸ†˜ Where to find more help: Link to your community (if it exists), FAQ section on your website, or your support email.

Onboarding video placement:

  • πŸ’¬ Welcome message in Telegram: You can insert a video link directly into the welcome message you define in the Dashboard in the Messages section (described in PART C, point 12.1). Telegram automatically displays a video preview, which is very effective.
  • πŸ“š Member section: If you have a member section for the product, the video must be there.
  • πŸ“§ First email after purchase: Send a link to the video in the first welcome email.

Analogy for explaining the shared profile (you can use in the video or materials):

πŸ€– Analogy with a robotic teacher

"Imagine an advanced robotic teacher that has been programmed for school teaching of several different subjects. In mathematics, it uses mathematical aids, special analytical algorithms, and communicates in precise, logical language. In physical education, it uses sports equipment, applies training methods based on the physiology of movement, and motivates with an energetic, encouraging manner. In art class, it works with creative techniques, teaches according to aesthetic principles, and expresses itself in an inspirational, figurative language.

For each subject, the robotic teacher has a different educational module installed with unique expert knowledge, methods, and approaches. Nevertheless, it's the same robot with one central brain that remembers your preferences, strengths, topics you're dealing with, and areas where you need support, across all subjects.

This robotic teacher was designed with absolute emphasis on student privacy. Even though it knows students from all classes and remembers their individual needs, under no circumstances does it tell any living person -- not the principal, not its creators -- the content of personal consultations or the course of lessons. Its programming doesn't technically allow it.

This is exactly how your [your instance name] and other AI experts on the BuddyPro platform work. Each has its unique specialization and approach, but all share a unified AI brain that allows them to better understand you. The content of your conversations is encrypted and remains completely secure -- no physical person, including me as the creator of [your instance name], other experts with their AI with the BuddyPro brain, or even the creators of the BuddyPro platform, has access to the content of your profile."

Automated Email Onboarding Sequence​

Prepare a series of automated emails (e.g. for 7-30 days) that will be sent to new paying customers after purchase. The goal is to maintain their initial enthusiasm, gradually walk them through all the possibilities of the AI expert, and help them create a habit of regular use.

Proposed email content:

  • πŸ“§ Email 1 (Immediately after purchase)
  • πŸ“§ Email 2 (E.g. after 2 days)
  • πŸ“§ Email 3 (E.g. after 5 days)
  • πŸ“§ More emails (1-2x per week for several weeks)
  • πŸ“§ Last email in sequence: Summary, encouragement for further use, link to community or other resources.

Using Community for Mutual Support and Inspiration​

If you have a closed community for your clients (e.g. Facebook group, Discord server, forum), actively involve it in supporting the use of your AI expert.

  • πŸ’¬ Create a dedicated space: It can be a special channel or regular topic where users can exchange tips, tricks, and experiences with the AI expert.
  • ✨ Sharing "Aha-moments": Invite users to share their best discoveries, most valuable advice, or most interesting conversations they've had with AI.
  • ❓ Ask questions: Stimulate discussion with questions like: "What did you ask your AI Expert this week and how did it help you?" or "What's the biggest benefit the AI Expert has had for you so far?".
  • πŸ‘‘ Moderation and your involvement: You or your team can regularly share your own tips in the community, answer questions, and respond to member posts. Your activity is key.
  • πŸ† Power of social proof: Seeing how other community members successfully and enthusiastically use the AI expert is often the best motivation and inspiration for others.

Proactive Reminders and Maintaining Contact​

Your BuddyPro instance can be set up to remind users of itself from time to time. This feature should be used with sensitivity and respect for user preferences.

  • 🌐 Global setting for proactive messages
  • πŸ‘€ Individual user preferences
  • ⚠️ Use with sensitivity

Onboarding is a continuous process. Regularly evaluate how your customers are doing using the AI expert, collect feedback, and improve your onboarding strategies accordingly. The goal is for your AI expert to become an indispensable partner for your clients.

Informing Users About Message Balance​

Part of quality onboarding is also transparent communication regarding the message limit. We recommend informing new users right at the beginning about the option of how they can verify their current message balance in their subscription themselves.

End users can use the command at any time:

/messageBalance

After entering the command, Buddy will display the exact number of remaining messages.

This command is not displayed in the /help list.

This step increases user trust, reduces uncertainty about limits, and helps prevent situations where the user unexpectedly hits an exhausted message limit.

Automatic Follow-Up Messages from Buddy​

BuddyPro can automatically send follow-up messages if the user doesn't respond to Buddy's previous message. These messages are sent by default approximately 1-2 minutes after Buddy's last response.

The goal of follow-up messages is to:

  • πŸ“ˆ increase user engagement,
  • πŸ”” naturally remind of the ongoing conversation,
  • πŸ—£οΈ mimic human communication style.

Disabling Follow-Up Messages (For Instance Owner)​

The BuddyPro instance owner can globally disable this feature for all users using the command:

/shouldSendFollowUpMessages:false

After disabling, Buddy will not send automatic follow-up messages if the user doesn't respond after its reply.

Setting Initiated Messages from the End User's Perspective​

BuddyPro can by default initiate messages on its own (e.g. follow up on a previous response, remind of a conversation, etc.). However, if the user doesn't want Buddy to write to them on its own, they can explicitly set this behavior using commands.

Commands for End Users​

The end user can use the following commands:

/shouldMessageMe:false

Buddy will not initiate any new messages and will only respond to direct user queries.

/shouldMessageMe:true

Buddy can again initiate messages (e.g. follow-ups or reminders).

Important Clarification​

  • Buddy usually also responds to natural instructions like "don't write to me on your own",
  • The setting applies only to the specific user.
  • The functionality of these commands depends on whether the instance owner has globally enabled message initiation.

Tip for onboarding: Inform your users about this option during onboarding so they feel in control of their experience.

Image Generation and Improved Image Editing​

Buddy can now generate images. When editing images, you can now reference an image in the chat history -- no need to attach it again as an attachment to your instructions.

The Reply function in Telegram also works -- you can write your instructions for image editing as a reply to a specific image. Right-click on the image in the chat or long-press it on mobile, then select the Reply option.

Additional Indicative Costs for Advanced AI Features​

In addition to processing know-how (texts, audio, video, URL), BuddyPro can also use advanced AI features during communication with users, such as voice messages (Text-to-Speech) and AI image generation.

These features consume AI credits separately and are not included in the know-how processing costs listed above.

The values below serve as an indicative overview for better estimation of real operating costs:

Voice Messages (Text-to-Speech)​

Approximate price per one generated voice message depending on length:

The voice message (Text-to-Speech) feature in BuddyPro allows converting text responses to spoken word, which significantly enriches the user experience. This feature is very useful for situations where users prefer listening to reading or need hands-free interaction.

The pricing structure works based on the length of the generated voice message, with the price not growing linearly with length. As you can see from the table, the difference between 1 minute ($0.072) and 20 minutes ($0.15) is not twenty times, but only approximately double. This is because most of the costs are related to initiating the generation process itself, not to its length.

For practical use, it's important to know that:

  • Voice messages are generated automatically when a user sends a voice message -- BuddyPro responds with both text and voice. Note: automatic voice responses are only generated when the text response is shorter than 3,000 characters. For longer responses, only text is sent unless the user explicitly requests a voice reply.
  • The user can also explicitly request a voice response to a text message.
  • By default, BuddyPro uses the "echo" voice from OpenAI, but other voices can be set using the /setVoice:{voice} command.
  • You can also create a clone of your own voice using the /createVoiceClone command, which is approximately 2x more expensive than standard voices.

As for image generation, the price ranges between $0.03 and $0.15 per image depending on complexity, resolution, and style. This feature is useful for visual representations that can complement text explanations or provide creative content to users.

Both of these features are charged from your AI credits but are not counted as extra messages in user tariff limits -- the voice version of a text response counts as one message, just as it would be a text-only response.

AI Image Generation​

Approximate price per one image:

  • ~ $0.03 - $0.15 per image

The final price may vary depending on the AI model used, resolution, style, and overall output complexity.

Frequently Asked Questions (FAQ)​

How long should the onboarding video be?

Ideally 10-20 minutes. It should be concise enough to hold attention, but also contain all the key information and practical demonstrations.

How often should I send emails in the onboarding sequence?

The frequency can be higher in the beginning (e.g., the first three emails during the first week), and you can decrease the frequency later (e.g., 1-2 times per week). The entire sequence can last anywhere from one to four weeks, depending on how much information and tips you want to communicate over time.

What if my customer doesn't want AI to reach out to them on its own?

The customer can tell your AI instance directly that they don't want proactive messages (e.g., "Don't text me by yourself."). The AI should respect this wish. As an instance owner, you can also globally disable proactive addressing with /shouldInitiateMessages:false.