The Elegant Art Of Framing Problems Before Solving Them

How to think of problems?

Before we develop solutions to a problem, with AI tools or without it, we need a framework to define our problems correctly.

The reason for this is simple:

1. If the problem is stated incorrectly, then the solutions to it will go awry and will get you sub-optimal results
2. If the problem is not stated completely, the solutions too will be incomplete. That leads to sub-optimal results again.

This happens all the time in business…

Situation 1: I recollect a client who asked me for a database of academicians in South East Asia, once.

On questioning him a bit, I understood that he was trying to reach out to them to join his community for educationists.

The problem was reframed as:
What do I need to do to reach out to academicians in Southeast Asia, to convince them to join my community. The database was just a small part of the overall problem t0 be solved.

Situation 2: On another occasion, a client wanted to build a community on his website. The natural and obvious solution to this was to plug in a discussion board.

But the community tanked after a bit. No traction from customers.

In hindsight, the correct way to solve this problem would have been:

1. Creating a community takes a lot of hard work. Triggering discussions, moderating it, actively promoting the community, etc. The owner must be aware of this, motivated to do this, have the time on hand for this.
2. You need to pilot the idea quickly before proposing a full-blown solution. It’s so easy to create a Facebook Group, or a LinkedIn Group. Try the community idea there and when it becomes a good problem to solve, start it on your website.
3. If you short circuit the process of testing ideas, you will be left with a white elephant and a fat bill.

Situation 3: For years at Pigtail Pundits, we harbored the idea of starting a Community and even started on it, several times.

Why? Because we could create the system with WordPress, or Drupal, or some system that could build a community back then.

It took us a while to understand that we were on the wrong track and solving the wrong problem.

The issue was that most of our attempts failed. Creating a community involves a time and energy commitment that’s not easy.

You can’t be focussed on projects and services to clients on the one hand and nourish ambitions to creating a community on the other.

What happens in reality is that when you get sizeable projects to work on, the community idea gets derailed and projects take the centre stage. So attempts like these are doomed to fail.

The better way to solve the problem would have been to devote some resources to the idea. Test it for traction. Solve the problem in a better way when it becomes a good problem to solve.

So, how do we avoid wasting energy on the situations described above?

Is there a recipe for solving problems and evaluating ideas?

There are few questions we should ask

Before we get carried away by solutions that are traps.

1. Is this the right problem to solve?
   Live with the problem for a bit. Examine it from multiple perspectives. Try the 11 Thinking Filters technique to uncover facets about the problem that you have not yet considered.
2. Do we have the resources to dedicate to this problem, now and in the future?
   Immediate availability of resources could be a trap. Think long-term.
3. Is there a way to elegantly test or pilot the solution?
   Before we get carried away with idea and begin to invest time and money into this problem.

   For example, before you invest big in a Community, can you start with a community on FB, or LinkedIn, or Discord? Don’t start with a Circle subscription when the outcome is uncertain.

These questions and deliberations will help you with answers.

Given these scenarios, we also need to put some thought to framing problems correctly, and we must be able to do it up front.

While a framework does help account for all the factors to do with problem, framing the problem correctly still requires human insight.

Hence some of the questions above.

So how do we solve problems correctly?

Assuming that we do need to solve the problem after due deliberation, there are 3 key steps that together form the problem ecosystem.

It’s important to think of the problem holistically as an ecosystem, rather than just an isolated issue begging for an immediate solution.

Acme Inc., a mid-sized manufacturing company, has been experiencing a steady decline in sales over the past year.

The company’s leadership team is concerned about the impact on revenue and profitability, and they want to identify the root causes of the problem and develop effective solutions.

Let’s apply the problem-framing ecosystem to this scenario:

Step 1: Define the problem and determine the scope of the problem ecosystem

• Clearly articulate the specific problem statement (e.g. Acme Inc. is facing a steady decline in sales, negatively impacting revenue and profitability)
• Determine the boundaries and key aspects of the problem space that need analysis:
  • Sales performance: Analyze sales data by product, region, customer to identify patterns
  • Market conditions: Assess shifts in customer preferences, economic and regulatory factors impacting demand
  • Competitive landscape: Evaluate competitors’ strengths, weaknesses, share and strategies
  • Product: Review customer feedback, returns, quality issues to identify improvement areas
  • Pricing: Analyze pricing vs competitors and impact on sales
  • Marketing: Evaluate effectiveness of current campaigns, channels, messaging
  • Sales execution: Assess sales team skills, performance, processes in converting and retaining customers

Step 2: Identify all stakeholders impacted by or involved in the problem ecosystem

• Internal stakeholders:
  • Sales, Marketing, Product teams
  • Finance, Executive leadership
• External stakeholders:
  • Customers
  • Suppliers, Distributors, Channel partners
  • Competitors

Step 3: Thoroughly analyze the problem components and their interactions

• Break down the problem into its constituent parts and examine how they relate and contribute to the issue
• For Acme’s sales decline example, analyze:
  • Market trends, consumer behavior shifts impacting demand
  • Competitive dynamics – market share changes, competitor actions
  • Product quality issues, need for innovation based on customer feedback
  • Pricing strategy effectiveness and competitiveness
  • Gaps in marketing campaigns, channels, messaging, targeting
  • Sales team skills, processes, obstacles impacting conversions
  • Misalignment between customer needs/preferences and current offerings
• Identify the most critical factors based on data and stakeholder input
• Determine key focus areas and interactions to address

How do we use Chatbots to frame a problem?

This is not a one-step solution.

Secondly, you can mine secondary data from reviews and discussions on a topic from the web for analysis by the AI tool.

But Primary Research of your customers, or potential customers, is still the best way to find nuanced answers.

Do not substitute secondary research for primary research unless you must [don’t have resources for it, don’t have enough time for it].

So approach the solution too bit by bit.

Gather and summarize all the insights and then seek to solve the entire problem, or bits of it at a time, as appropriate.

For example, you can upload the findings and data as a PDF and then ask questions of the chatbot which will reference the PDF to answer your questions.

This can be iterative and it should be.

AI Recipe: Prompt for AI Chatbots

Notes

The RATS Framework is an elegant method I developed for writing prompts that provide context clues and structure to help GPTs frame and answer problems.

Before using RATS prompts, ensure you have clearly defined the problem and constraints using the Problem Statement or Product Brief Methods from Phewture.

If this this your first foray on Phewture, do familiarize yourself with the Basics of Prompting.

Begin Prompt

Complete the tasks below one step at a time. After each step ask for my permission before moving on to the next step. If a step takes more than one message to complete send “Continue” to proceed to the next message. Provide a complete detailed response without worrying about message length.

[Problem] =
State the problem with all the constraints [resources, stakeholders]

[Data] =
Individual sets of data/ findings from surveys, data from aggregators, competitor analysis, etc. can be put through AI to gain insights, or nuance the problem better.

Feed this data bit by bit and then analyze the overall situation after you have all the insights.

[Role] =
Act as a holistic and experienced business thinker [outline 5-7 facets for this role. Ask the AI tool to do this for you]

[Avatar] =
These could be individual stakeholders, or customers. Define the stakeholders and their mindset clearly so that you get a tailored solution for the problem.

[Task] =
Frame this problem/ challenge stated above with the inputs shared. Attempt the problem one step at a time, and ask my permission to proceed, before you attempt the next step.

[Topic] =
Broad area of the solution

[Structure] =
[This is structure of the task you want AI to solve, as below]

Step 1
Identify the specific problem to be solved and articulate it clearly

• Determine the scope and boundaries of the problem ecosystem
• Identify key stakeholders impacted by or involved in the problem ecosystem 

Step 2
Analyze the problem ingredients

• Break down the problem into its basic units or component parts
• Identify relevant variables, constraints, and parameters

Step 3

• Gather data and information related to each problem ingredient 
• Break down the problem into its constituent parts and examine how they relate and contribute to the issue

Rinse and repeat this process till you solve the problem elegantly given the constraints.

End prompt

AI Tools that you can use

The prompts on Phewture are AI tool agnostic.

They should work on GeminiPro, ChatGPT, Claude and Perplexity. Even on free tools such as llama3 by Meta which you can install on your desktop.

• ChatGPT and Gemini Pro can read Google Files by way of attachment
• Claude can read PDFs.
• Perplexity can read text files and PDFs.

Use these capabilities to upload your prompts and get the best out of these tools.

You will have to use the following prompt, if you follow the file upload method: Please complete the tasks outlined in the attached document.

Next Steps

1. To help you find your way around Phewture, I have put together a set of articles under Wayfinding. Do go through these and you’ll navigate like a pro through this stream of consciousness. 😄
2. The Learning Methods are exercises that I’d recommend if you wish to wrap your head around the possibilities of AI at work, or play.
3. Don’t forget to leave your comments below and share your joy with your friends on social media. Use the share icons below this post to gain some good karma.

Wish to train your in-house team on AI techniques?

Phewture offers AI-spurred training for teams. Do check out the Training Services.