Learning and development professionals have always struggled to keep up with the fast pace of business. They must produce a detailed needs analysis to create an effective training. However, many people feel they lack the time to do it consistently and thoroughly. What if generative artificial intelligence (AI) could help?

Integrating generative AI tools offers a promising way to streamline and enhance the needs analysis process. AI can help practitioners gather insights, strategize, and prepare for complex needs assessments, leading to more effective training interventions.

I teamed up with Dr. Jody Lumsden to experiment with ChatGPT 4.0 and Claude 3.5 Sonnett. Our goal was to find a strategy that saves time in the needs analysis process by using AI. Here’s what we discovered:

A Better Needs Assessment

The goal of conducting a needs assessment (also referred to as training intakes, up-front analysis, or root cause analysis) is to clearly define what’s wrong. What’s the business problem? And, how might training be part of the solution?

An effective needs analysis incorporates a performance consulting lens and explores what environmental and individual changes are needed to help solve the business problem.

Our experiment showed that L&D professionals can save time using generative AI and be better prepared to facilitate needs assessment conversations with stakeholders. AI tools can analyze lots of data, find patterns, and give recommendations based on evidence. This can help L&D professionals and business leaders work better together to identify performance gaps and solve problems, ultimately leading to more effective training solutions in less time.

The Experiment

Imagine you get a request for training. You instinctively recognize that a training is not going to be a complete solution or even the right solution. In the past, you struggled to talk about training needs with requesters. You also found it hard to seek mentorship to improve your ability to respond to training requests.

We used this common scenario and asked a series of needs analysis questions to two AI tools to see how helpful their insights and recommendations were. We reviewed the results, comparing the depth, accuracy, and applicability of the insights provided by each tool.

We also compared the different AI platforms side by side to see how they can improve the needs assessment process. Of course, AI is not a silver bullet, so we’ll share a few precautions to take while leveraging AI as well.

The Prompts

We selected eight questions to explore how various environmental and individual factors may be contributing to the business problem (and reason for the training request).

Initial prompt sharing the situation and context: (The following is an example.)

I am the lead instructional designer for the King County Metro System. We are the public transportation provider for the greater Seattle area. We employ 300 bus drivers and subway operators (our front-line staff) and another 100 management and administration staff members.

One of the front-line staff managers asked me if our team could update the safety precautions training program. We normally update this training every two years. This request is coming much sooner than anticipated.

I would like some help preparing for the meeting with the manager to gather possible reasons for the training request, problems that might be going on with the front-line staff, and some ideas around possible outcomes and goals for the program. Can I ask you a few questions to help me generate ideas to present to my manager?

Prompt exploring possible reasons for the training request:

What might be some things happening in the organization that could be driving this request for updates to the safety protocols training program?

Prompt asking for possible data that could be used to explore if the training did, in fact, solve the problem:

For each of the examples you referenced above regarding possible reasons the front-line staff manager is requesting training, which included:

• An incident or accident

• Changes in regulations or industry standards

• Introduction of new equipment or technology

• Feedback from the front-line staff

• Changes in operational conditions

Can you provide suggested data we could use for each example to verify if the training solved these possible problems/changes?

Prompt to explore other possible actions to address the possible problem:

What are some possible actions the front-line manager might have already taken to address these issues outside of training the front-line staff?

Prompt to explore other contextual and environmental factors that could play a role in the training request:

What other organizational changes might have taken place that could also influence the need for changes in a safety protocols training?

Prompt to explore possible outcome metrics:

If the safety protocol training is successful and front-line staff are following the protocols with near 100 percent fidelity, what might be some business results that we see?

Prompt to explore specific data artifacts to use for the possible outcome metrics:

For each of the possible business results you referenced in the question above, what might be some data we could collect to show that the training made a positive contribution to those business outcomes?

Prompt to summarize all the information generative AI provided into a useful tool to incorporate in your initial training intake or needs analysis conversation:

Can you take the information provided in this conversation and create a checklist I can use when I talk with my front-line manager to have them select: the reasons for the training request, the possible actions already taken, possible organizational changes that have taken place, possible business results, and also include the example data we could collect for each area?

The Result

The outcome of our AI-assisted needs analysis was transformative. By using the AI tool, we achieved significant time savings and streamlined a traditionally labor-intensive endeavor. The AI-driven approach led to a comprehensive checklist that can be easily adopted by both practitioners and clients. By standardizing the needs analysis process, and ensuring that multiple critical factors are considered, it provides a reliable and efficient method for conducting assessments.

Here’s a checklist that Claude 3.5 provided from our AI-assisted needs analysis experiment. You can use this tool when meeting with stakeholders to gather details about the training request, for example, actions already taken, organizational changes, potential business results, and example data to collect.

Note: The items included in this checklist are specific to the public transit training scenario. Customize your own checklist by inputting the exact details of what your organization does and a description of the training request to support your intake conversations.

Reasons for Training Request:

[ ] Incident or accident

[ ] Changes in regulations or industry standards

[ ] Introduction of new equipment or technology

[ ] Feedback from front-line staff

[ ] Changing operational conditions

[ ] Other: ____________________________

Actions Already Taken:

[ ] Incident investigations

[ ] Policy and procedure review

[ ] Equipment or facility inspections

[ ] Safety committee meetings

[ ] Communication and reminders

[ ] Individual coaching or counseling

[ ] Temporary measures

[ ] Collaboration with other departments

[ ] Other: ____________________________

Organizational Changes:

[ ] Restructuring or reorganization

[ ] New leadership or management

[ ] Changes in union agreements or labor contracts

[ ] Implementation of new technologies or systems

[ ] Expansion of services or routes

[ ] Changes in ridership patterns or demographics

[ ] Mergers or acquisitions

[ ] Changes in regulatory or compliance requirements

[ ] Evolving industry best practices

[ ] Other: ____________________________

Potential Business Results:

[ ] Reduced accidents and incidents

[ ] Improved employee safety

[ ] Decreased liability and legal costs

[ ] Lower insurance premiums

[ ] Increased customer satisfaction and ridership

[ ] Enhanced reputation and public trust

[ ] Improved employee morale and retention

[ ] Compliance with regulations and standards

[ ] Cost savings

[ ] Competitive advantage

[ ] Other: ____________________________

Example Data to Collect:

• Accident or incident rates (pre- and post-training)

• Severity of accidents or incidents

• Compliance audit or inspection results

• Proficiency assessments or tests

• Observation or monitoring data

• Staff feedback or survey results

• Ridership or passenger counts

• Public perception or social media sentiment

• Employee satisfaction or engagement surveys

• Employee turnover or retention ratesRegulatory citations or fines

• Cost data (repairs, replacements, legal fees, insurance premiums)

• Market share or benchmarking data

• Awards or recognitions

• Other: ____________________________

How Do You Use the Checklist?

If your natural inclination is to schedule a meeting and just run through the checklist, that might be a misguided action. Generative AI helped us brainstorm ideas for the training requester. These included potential individual and environmental factors, along with success metrics to track the program’s effectiveness.

However, we still must thoughtfully prepare the best way to use that data. Here are some other things to consider before you meet with the training requestor:

1. What is your relationship like with this person?

2. Do you know them well?

3. Have you had many needs analysis/training intake conversations before?

4. Do you know the level of knowledge the training requestor has regarding the individual and environmental factors that might play a role in the training request?

5. What expectations do you and the training requestor have regarding how the training intake/needs analysis conversation will go? Is your requestor prepared for the list of factors you’d like to explore, or will this come as a surprise?

6. Using the information provided by generative AI.

In our example, ChatGPT 3.5 offered more information than requested. We wanted a simple list of ideas, but ChatGPT 3.5 anticipated we might need advice on data collection and analysis. While useful, this detail is too much for a needs analysis checklist or conversation. Avoid copying everything provided by generative AI—some information might be confusing or add unnecessary complexity.

A great intake conversation with your stakeholder leaves no stone unturned. L&D professionals can’t possibly be experts in every single factor affecting a training request. Your stakeholders won’t have all the answers either. They also might lack institutional and market knowledge, as well as other insights that influence business problems and solutions.

AI insights empower both requesters and L&D professionals to explore all problem and solution variables.

Check out additional resources to simplify and streamline your needs analysis and measurement and evaluation processes by these authors:

• Needs Analysis on a Shoestring by Dr. Jody Lumsden

• Measurement and Evaluation on a Shoestring by Dr. Alaina Szlachta