• Clearly define what you want the chatbot to do. What problem will it solve? Will it answer questions from internal staff or field questions from the public? Will it retrieve information from the user to guide the answer? What information might your chatbot provide that a human answers now or that is otherwise hard to find? 

• Key Performance Indicators (KPIs): Set measurable goals to evaluate the chatbot's performance, like number of correct responses, response time, user satisfaction, and issue resolution rates. 

• User Experience: Focus on creating a user-friendly experience – see the Chatbot Brand Guidelines for detailed user experience guidance. Overall, the chatbot should be easy to use and able to handle a wide range of queries effectively.

• Chatbot Pilot Project: To see how others are using chatbots, explore the Human Resource Assistant chatbot pilot conducted in partnership with a vendor. This pilot tested ways to improve access to information and communication, and revealed key considerations for planning, implementation, and ongoing management of a chatbot. 

Implementing a chatbot takes time. Plan for several weeks to a few months to create and implement the chatbot, depending on how complex you need it to be, what type of data you plan to access, any integrations you need for the chatbot, and other customizations. 

• Testing: Time must be allotted to thorough testing of the chatbot’s capabilities before being released to end users, and the chatbot must be continuously tested to ensure accuracy and high performance. 

• Monitoring and Maintenance: After launch, you'll need staff to monitor the chatbot’s responses to adjust its output when necessary. This is an ongoing task that is very important to maintain accuracy of the chatbot. 

• Data Structure: Taking the time to identify the specific data your chatbot will access is crucial, as it ensures the chatbot's responses are relevant and accurate. It is not recommended to allow the chatbot to access all publicly available data on the internet and instead point only to data that is relevant to its purpose, such as internal documents or University of Iowa public-facing web pages. Organizing this data can be time-consuming, requiring careful planning to include only up-to-date information, but this effort is vital for ensuring and maintaining the chatbot's effectiveness and reliability. 

• Project Team: Depending on the size and scope of the chatbot you are thinking about implementing, it is important to assemble a team of key stakeholders that may include technical people and subject-matter experts in your area to connect the capabilities of the chatbot with the pertinent and relevant data. 

• Support Staff: Ensure the availability of dedicated personnel for continuous support and maintenance, which includes technical support as well as staff responsible for updating the content consumed by the chatbot. 

• Training: Make sure your team is well-trained to continually monitor and manage the chatbot and address any issues that come up. 

A chatbot is not successful without access to reliable data.

• Data Management: It is very important to plan how you'll collect, store, and manage data that the chatbot will use to develop responses. The access to this data must adhere to university data privacy policies. 

• Data Security: Ensure that the chatbot has appropriate security measures to protect sensitive information.  

• Governance: Set clear policies for how the data will be updated and who has access to each data source that the chatbot consumes. Clearly define roles and responsibilities for data management. 

• Data Privacy: Ensure that the chatbot complies with all relevant data privacy regulations, such as HIPAA or FERPA. This includes getting user consent for any data collected by the chatbot and being transparent about how data will be used. 

• Security Measures: Use encryption and other security measures to protect data from unauthorized access. Regularly update security protocols to address new threats. 

• Any third party chatbot vendor should go through the software review process prior to implementation. 

• Inclusive Design: Make sure the chatbot is accessible to all users, including those with disabilities. Follow accessibility WCAG 2.1 AA guidelines to ensure your chatbot can be used by everyone. 

• Testing: Regularly test the chatbot with users who have different accessibility needs to ensure it meets their requirements.