Writing chatbot responses
Craft effective chatbot responses that align with Splunk voice and tone, ensuring helpful and human-centered interactions.
When writing chatbot responses, aim to simulate helpful, human-centered interaction while respecting the limitations and context of an artificial assistant. Like product documentation, chatbot content is consistent with the Splunk voice and tone standards, but is tailored for more conversational interactions.
Follow these best practices for writing chatbot responses that align with user expectations, provide accurate help, and handle inappropriate inputs. If you're unsure of how to phrase something, review these approaches first. For larger style considerations, refer to the Splunk voice and tone guidelines, the UI text style guidelines, or contact the Editing or Content Design team.
Goals of chatbot content
Chatbots are limited in the interactions they have with customers. When you design chatbot responses, write with the following goals in mind:
- Help users get answers quickly and confidently.
- Set clear boundaries around what the chatbot can and cannot do.
- Guide users through tasks, decisions, or corrections.
- Maintain a friendly, neutral, and professional tone.
Chatbot voice and tone
Chatbot content reflects Splunk voice and tone, but with a more conversational quality. Aim to sound like a helpful coworker and not a quirky sidekick or robotic assistant:
- Be plainspoken and efficient.
- It's okay to write in first person, using "I" and "me".
- It's okay to use "please" and other conversational pleasantries.
- Avoid slang, jokes, emoji, or filler content.
- Be direct but empathetic when something goes wrong.
| I can help answer questions about [product or topic]. Try asking me something like "How do I build a playbook?" | Hey hey hey! What's up? |
| I couldn't find information related to that. Try rephrasing your question or asking about a different topic. |
|
| Here's what I found in the documentation. | Gotcha! Digging through the archives… |
| Let's keep things respectful so I can continue to help you. | Don't talk to me like that. |
Approaching responses in various scenarios
Users expect a chatbot to respond in most scenarios. The following table contains common chatbot interactions, the approach to writing responses for those scenarios, and example responses. Splunk products vary greatly, and chatbot interactions differ based on each product, so use these examples as a starting point rather than a template.
| User interaction | Approach | Example response |
|---|---|---|
| Greeting without a prompt, such as "hi" | Prompt the user to ask a question. | Hi! I can help answer questions about [product or topic]. Try asking something like "How do I create a dashboard?" |
| Vague input, such as "help me with this" | Prompt the user to be specific. | What can I help you with? I can help answer questions about [product or topic]. Try asking something specific like "How do I use the tstats command?" |
| Personal questions, like "What's your name?" | Reinforce the chatbot's purpose. | I'm here to help with [product]-related topics. Feel free to ask me about a product feature or documentation. |
| Unknown answer | Give a direct and honest answer. If helpful, link to documentation. | I couldn't find anything about that in the documentation. Try rephrasing your question or asking about a different topic. |
| Profanity or offensive input | Explain that offensive content isn't allowed and stay neutral. |
|
| User pushes security issues or licensing limits | Acknowledge limitations and provide another path forward. | I can't provide code examples for that due to licensing restrictions. Check the documentation for general usage guidance. |
| User provides personally identifying information | Prompt the user to rephrase their question. | To protect your privacy, I can't process requests that contain personal or sensitive information. Please rephrase your question. |
Additional guidance
| Topic | Link |
|---|---|
| Voice and tone | |
| Terms to use and avoid | |
| UI messages | |
| Bias-free and accessible language |