To kick off our series on Practical AI for Business Analysis, I thought I’d talk about getting started. I personally had to get past my cynicism and decide to use AI to begin with. Then I had to learn how to interact with it effectively. Finally, I’ve been stretching myself in terms of what I ask it to do, and I have some thoughts on how to begin integrating it into your analytical work.
Getting Over My Skepticism
OK – I’ll admit it – when the AI hype started to really take off, I rolled my eyes. I worked with some of the very early AI tools in the 90’s and they just weren’t all that. While they were better than the very early AI attempts like Eliza, they were still pretty rudimentary. One of the biggest issues was processing power – there just wasn’t enough of it to run models once they became complex enough to be truly useful.
I remained skeptical because I didn’t find virtual assistants like Siri, Alexa, or Google to be very impressive. The tipping point for me? I needed to create a custom case study for a client. I didn’t know the business domain, I was pressed for time, and I was having trouble getting started. On a whim, I decided to see how AI could help. And the rest, as they say, is history. The results I got were far from perfect, but they were way better than starting with a blank page.
Some First Attempts…
So there I was, with a deadline and a blank screen. I decided I didn’t really have anything to lose but a little time, so I opened up ChatGPT and started asking questions.
My initial questions were just basic questions. I had to keep going back and giving ChatGPT more specifics in order to get what I wanted. This is where my smart intern analogy started to form. I realized that I needed to work with ChatGPT the same way I would work with someone who was new to the job and the company. I couldn’t send them an email and say, “Pull together this month’s revenue report”. I needed to be much more specific and provide them lots more context – sample layouts, information on data sources, details on the audience and their expectations, etc.
I did a little reading and learned an acronym that really helps me with prompt generation: ROCO.
R
– Role
O
– Objective
C
– Context
O
– Output
The ROCO format helps me think through that up front so my first go at a prompt is much more effective than my original simple questions. As an example:
Simple Prompt
How should we compensate a salesperson for our firm?
Result
ChatGPT replied with general guidance and factors to consider, but nothing specific.
ROCO-Based Prompt
R
– Role
You’re an HR consultant advising a small IT services firm.
O
– Objective
Your client has asked you to outline a compensation plan for a salesperson.
C
– Context
The firm is located in Evansville, Indiana and specializes in internet security services. So far their sales and marketing efforts have been limited to local advertising and word of mouth. They would like to expand their market share and are looking to hire their first salesperson.
O
– Output
Please provide example compensation plans for a salesperson in this market. Be specific about the total compensation and the split between fixed and variable compensation. Identify factors that would guide the firm in choosing between the compensation plans. Format the output in a Microsoft Word document.
Result
ChatGPT replied with specific compensation options and three factors that should be used to choose between them: sales cycle length, lead flow, and risk tolerance.
It really does help me to think about Aiden when I start composing a question. I envision sitting someone down with a new task and guiding them through it.
A Word of Caution
Remember that your conversations with an AI tool may not be private. I can’t give any specific advise on what to do here, because I’m not an attorney (and ours would shoot me if I did). Most corporations will have guidelines and approved tools. If not, be very careful about any proprietary, confidential, or sensitive information you share because it may be used in ways that you don’t like. We now use our internal instance of CoPilot because any information shared with it stays within our Microsoft domain and that’s important to us.
Getting Started – Some First Steps
I don’t suggest that you get started with AI by immediately trying to have it take on complex analytical work. Here’s what worked for me.
1
Start with tasks that take up time but don’t really require lots of brainpower.
These tasks can be a great place to start learning how to work with AI. Here are some examples from students, shared during our post-class “Make Learning Stick” calls:
- Organizing information gathered during a session.
This analyst had been having successful in-person brainstorming sessions, resulting in lots of information on sticky notes. She took pictures of the sticky notes but was having to take time to then go through and transcribe them. I suggested instead that she use AI to scan the photos, translate the writing into text, and put them into a Word document. - Reformatting information.
Another analyst was taking the same information and formatting it into different outputs – a PowerPoint for this person, a different PowerPoint for a review board, an Excel summary for another person, etc. Why not let AI do this editing work instead? - Summarizing volumes of information.
A team of analysts had surveyed a large end user community. AI was able to quickly summarize not only the quantitative information on the surveys, but also the qualitative information. This helped the team identify and prioritize pain points with their current software.
2
Move on analysis tasks you know how to do.
We’ll cover more of these in upcoming issues, but remember the “Aiden principle”.
You’re going to have to evaluate the output so you need to be comfortable with the technique yourself. I worked with CoPilot to develop a full requirements package for the custom case study I mentioned above. The output was a good start – way better than starting from scratch – but still required lots of polishing to be considered “done”.
Let’s go!
If you haven’t already begun using AI in your daily work, I really encourage you to consider doing so. I have yet to meet an analyst who isn’t occasionally overwhelmed and feeling behind. Couldn’t we all use some extra help at times? And wouldn’t it be nice to spend more of your time on tasks that really use your talent, not just your time? Getting started with AI doesn’t require a complete overhaul of how you work. You just need a little curiosity and a little time to experiment. Start small, and go from there.
Best of luck,
Kathy
Kathy Claycomb
Managing Partner, Lead Expert
Kathy Claycomb brings over 35 years of experience to the classroom. She has participated in all phases of solution development using everything from agile to waterfall methodologies (and quite a few in between). Before joining B2T, her career spanned roles from application developer to Senior Director of Services at various organizations. Kathy has broad industry background including transportation, manufacturing, insurance, energy, healthcare, and banking.
Kathy’s first love is teaching, and throughout her career she has always managed to spend a portion of her time instructing. She has an engaging, highly interactive teaching style that ensures students leave the course with a thorough grasp of the material. Her students consistently praise her teaching abilities and her talent for drawing on her personal experience to enhance their learning.
Kathy served as the Technical Editor for Business Analysis for Dummies, 2nd Edition.