About SiteGPT

SiteGPT is an AI chatbot platform that helps you create AI chatbots that are trained on your company's knowledge base.

People come to SiteGPT for 3 use cases:

1. Reducing Load on customer support.
2. Lead Generation through automated chatbot.
3. To create an internal-facing chatbot for the team members to use.

Reducing Load on Customer Support

Most of our customers come to SiteGPT because they get 1000s of support tickets every month – many of these tickets are repetitive FAQ type of questions that can easily be answered by an AI chatbot. This is the main use-case why they come to SiteGPT. They want to put an AI chatbot on their website that answers all their users' questions 24x7. If the user asks a question that is more complicated or needs human intervention, they can easily escalate the conversation to human support. This way, the support team can focus on answering more important questions and leave the rest to the chatbot.

Lead Generation through automated chatbot

Lead generation is another use-case why customers use SiteGPT. They want to put an AI chatbot that automatically collects them leads. They can then contact those leads at a later point of time and get new business through them. This is still in the works and not many people use SiteGPT for this use-case. But in the future, this has the potential to become a bigger use-case for SiteGPT as it helps businesses generate actual revenue.

Internal Facing chatbot for team members

This is another use-case where customers create chatbot only for internal use. These companies often have 1000s of internal documents, SOPs and processes that they use internally. They do not want to make these chatbots public facing and these users often use our Google Chat and Slack integrations so that their team members can directly chat with the chatbot from their own communication platforms that they use for their business. The chatbot helps the employees find information about any of these internal processes quickly by just chatting with the bot. They don't have to search through 1000s of documents each time to get a simple answer.

​

From this, it is clear that the top 2 use-cases people use SiteGPT for are:

• Reducing Load on Customer Support
• Internal Facing chatbot for team members

​

So our ICPs will be based on these two use-cases.

​

ICP 1

• Team size of 5 - 50 with 1 or many customer support agents
• Web traffic of 10,000+ visitors per month
• Get 1000+ support tickets per month
• Main features that this ICP uses:
• • Analytics dashboard to see how many conversations are happening each day, number of positive and negative conversations, number of escalations, etc.
  • Teams feature – invite all their agents and other stakeholders to SiteGPT to manage the chatbot.
  • Escalate to Human – If the chatbot cannot answer the question, they will escalate to human support and agents will take over.
  • Q&A – Some times, the chatbot may not be able to answer some of the questions properly. This feature is there to override the bot answers so that next time if someone asks the similar question, the new answer will be used to answer it.
• Apps used -
• • Chat Solutions - (Zendesk, Intercom, Crisp, Freshdesk etc)
  • Email Automation tools (Mailchimp, HubSpot etc)
• JTBD
• • This ICP has a lot of support tickets that they get every month.
  • 80% of the support tickets are simple questions that can be easily found on their existing knowledge base.
  • The only 20% of the tickets are actual questions which require human intervention.
  • So the main JTBD for this ICP is a functional one – To reduce load on customer support by answering all the usual questions (that can be easily found on the existing knowledge base) so that their support agents can focus on more important questions. They will save time and also be more productive.

ICP 2

• Team size of 25+
• Have atleast 1000+ documents/internal web pages
• Apps used -
• • For Internal communication - (Slack, Teams, Google Chat etc)
• JTBD
• • When I use SiteGPT, I want to be able to find the information easily in seconds, so that I can spend my time on productive work.

​

Onboarding Teardown

Please go through the entire onboarding teardown of SiteGPT in the below attached PDF file. I have attached compressed pdf version as it is not allowing me to upload the actual one. So, i am also mentioning the drive link of my actual PDF. If you feel the quality is much less in the compressed version, please refer the drive link.

Onboarding Teardown - SiteGPT-pages-deleted-compressed.pdf

Activation Metrics

Activation Metric 1

Hypothesis

The average daily activity of the user over first 30 days should be greater than 30 min.

Reasoning

If the user is spending 30 min daily for the first 30 days on the platform, that means user realizes the value of the product.

Metrics to track

• Daily Activity time

​

Activation Metric 2

Hypothesis

User downloading the chat history from the dashboard at least twice in the first 2 weeks.

Reasoning

The purpose of the user downloading chat history from the dashboard is to analyse the chatbot performance and to see if there are any new set of questions asked by visitors where the bot is giving the best answer to them. By doing this, user can take a decision on how to improve the existing knowledge base so that next time that same question is asked by anyone else, the bot should respond better. (or) If there is need to tweak the prompt so that bot could respond better.

By doing this exercise regularly, user will become a kind of Pro User, he will spend more time on the platform and he will realise the real value of the SiteGPT and overtime, the chatbot will get perfect in terms of knowledge it possess.

Metrics to track

• Dates of chat history downloads
• Frequency of chat history downloads

​

Activation Metric 3

Hypothesis

User adding/updating 10 Q&As from the chat history page over first 10 days.

Reasoning

The purpose of Q&As is to give additional FAQ type knowledge to the chatbot. If there are some core questions about something for which the answer has to be same all the time, for those cases Q&As are perfect. There are 3 types of Q&As -

• Manually Added Q&As (can be added and updated from Q&A page)
• Revising Bot Answer Q&As (can be added from Chat history and updated from either Chat history or Q&A page)
• Negative User Feedback Q&As (can be added from Chat history and updated from either Chat history or Q&A page)

The main thing that we want to track here is how many Q&As user is adding from the chat history page. i.e., Revising Bot Answer or Negative User Feedback.

When user goes through chat history and if the user feel like bot has not answered up to the mark for any common question which is very important, they can Revise the answer from Chat history by converting it to Q&A. Or if visitor was not satisfied with the bot answer and he gave a thumbs down on that answer, it will be auto added to Q&A and user can see all those in chat history.

By keeping the hypothesis in mind, if the user achieves the said hypothesis, that means he is actively looking at bot responses and correcting them and if does 10 of them in 10 days that means he is able to realize the value of the product.

Metrics to track

• Number of unique creates of Revise Answers
• Number of unique updates of Revise Answers
• Number of unique updates of Negative User Feedback

Here we wont track number of creates of Negative User Feedback because those are automatically added by SiteGPT itself whenever user gives a negative feedback, so we only track updates on those.

​

Activation Metric 4

Hypothesis

Minimum of 3 active team members (agents) added in the first 2 weeks.

Reasoning

SiteGPT has multiple types of team members (ranging from Super Admin to Agent). Here we will be only tracking number of Agents. Agents mostly have access to chat history. If at all they are spending time on SiteGPT, it would be 95% on chat history page itself. If the user has added min 3 of their team members as agents, then it shows that they have gained confidence on the product and then added 3 agents. This shows both commitment to the product and also shows that they have realised the value of the product and that is why they have 3 members.

Metrics to track

• Number of agent member invites
• Number of agent member accepts
• Daily activity time of each agent

Activation Metric 5

Hypothesis

Adding atleast 1 integration in the first 7 days.

Reasoning

SiteGPT has various integrations like Google Chat, Messenger, Crisp, Slack, Zendesk, Freshchat etc. If user is integrating one of these channels with SiteGPT that means that they are showing the commitment to use the product.

Metrics to track

• Integration creation
• Integration update

​

​