This experiment emerged from a client session I led on storytelling with data. During our discussion, it became clear they were actively encouraging the use of AI to boost productivity and effectiveness. I saw a strong connection between data storytelling and AI, which inspired the experiment.

The core question was: Can AI enhance how I tell stories with data? I wanted to explore whether AI could help me articulate my thoughts, structure my narrative, suggest relevant data, tailor messages to specific audiences, and apply established storytelling frameworks. My hypothesis was that this would lead to clearer, more audience-specific, and ultimately higher-quality stories that would support better discussion and exchanges of perspectives, and ultimately better decision making and risk management in businesses.

Hypothesis:

My hypothesis was essentially “Leveraging AI tools such as CoPilot can improve how I tell stories with data to support communication, decision making and risk management in organisations”. There were a couple of nuances within this around “improve” e.g.

• Using AI (CoPilot) will improve how I tailor data stories to specific audiences
• It will improve how well I apply engaging story structures to create the narrative or “Talk track” of the story (assuming a presentation)
• It will improve the choices I make about what data to include by suggesting ideas that I wouldn’t have considered independently
• As a result of all of the improvements, my data stories will be more engaging, clearer in terms of the ask of the listener, and more influential in terms of gaining support and commitments to action.

Process:

These are the steps I took:

1. I dictated an initial talk track of the data story into MS Word. This was very free flow and top of mind, but allowed me to get all of the ideas in my head on to a piece of virtual paper. It was messy, unstructured and totally unsuitable other than as initial source material. All the information was 100% fictional.
2. I copied and pasted the raw text into CoPilot and added the prompt “Edit for conciseness” in front of the copied text.
3. I reviewed the output and edited it for accuracy.
4. I used the prompt “Structure the transcript using a story structure of current state, crisis, resolution” and review the output.
5. I prompted CoPilot to adapt the story for specific data preferences that might be associated with a specific leader in a specific role (in real life I would do this based on my knowledge of working with someone and knowing their preferences e.g. needing a lot of detail, wanting the big picture, needing to understand all the different options). The prompt I used was “Adapt the story for a commercial leader who likes to see data and needs to make a decision on the basis of the story”. I reviewed the outputs.
6. I prompted CoPilot to identify key metrics e.g. “Can you suggest key metrics to highlight”? Then I reviewed and edited the output.
7. I followed some of CoPilot’s suggested prompts: “Please help me visualize these metrics”, and “Please provide examples of how to present this data”.

Output:

Here is the output of the data story once it had been edited for conciseness by CoPilot. I want to stress that I made up this data story, and all the data is fictional. The output is included so you can follow along, but feel free to skip to “Was it useful / What did I learn?”

Quarterly Business Review: Sales Overview & Market Segments
I’ve been asked to present data on our sales performance this past quarter, comparing it to previous periods and examining three key market segments: direct-to-customer sales, café partnerships, and mobile coffee vans.
1. Direct-to-Customer Sales
Originally launched during COVID via our website, this channel has seen a steady decline.
Peak monthly sales were only 20–25 units; now we’re averaging around 10.
It’s currently costing us more to maintain than it generates.
Recommendation: Consider discontinuing this channel or partnering with niche online retailers to scale more effectively.
2. Café Partnerships
We’ve lost two major café clients, accounting for ~20% of revenue, due to feedback on flavour profile and rising prices.
Remaining partners are buying irregularly, and several contracts are up for renewal.
Rising costs (minimum wage, energy, production) continue to impact margins.
Outlook: Declining demand and revenue expected; limited short-term recovery.
3. Mobile Coffee Vans
This segment shows promise, with increased interest from vendors at beauty spots and festivals.
Our local, low-footprint coffee appeals to this market.
Sales are ad hoc rather than contract-based, making revenue unpredictable.
Recommendation: Focus relationship-building efforts here to grow this segment.
Summary & Recommendations
Overall, revenue is declining while costs are rising.
Prioritize mobile van partnerships for growth.
Monitor café partner feedback but avoid over-investing in flavour profile changes without broader evidence.
Exit or restructure the direct-to-customer channel to reduce losses.

Whilst this is an effective summary of the data, it doesn’t read like a story and is more suitable for a report structure. I’d give it a 3/10 as a data story as it’s not engaging.

Here is the CoPilot output once it had been structured using the story structure of current state, crisis, resolution:

Current State
Our business operates across three main sales channels: direct-to-customer via our website, partnerships with local cafés, and supply to mobile coffee vans. Each channel has played a role in our growth, with varying levels of success and stability.
Direct-to-Customer: Launched during COVID to offer locally ground premium coffee at home. Initial uptake was modest, peaking at 20–25 monthly purchases.
Café Partnerships: Historically strong, with around 50 local coffee shops supplied, including two major clients.
Mobile Coffee Vans: A growing segment, especially at beauty spots and festivals, with positive feedback on our local and eco-friendly credentials.

Crisis
This quarter has revealed significant challenges across our channels:
Direct-to-Customer: Sales have steadily declined, now averaging just 10 per month. The channel is no longer financially viable and is costing us money to maintain.
Café Partnerships: We’ve lost two key clients (20% of café revenue) due to flavour profile concerns and rising prices. Remaining partners are buying irregularly, and several contracts are up for renewal. Rising costs (wages, energy, production) are squeezing margins.
Mobile Vans: While interest is growing, purchases are inconsistent and not contract-based, making revenue unpredictable.
Overall, revenue is declining, and costs are expected to rise again within six months.

Resolution
To navigate these challenges, we propose the following actions:
Refocus on Mobile Coffee Vans: Strengthen relationships in this promising segment, leveraging our local and sustainable brand appeal.
Exit or Restructure Direct-to-Customer Sales: Consider removing the website channel or partnering with niche online retailers to scale more efficiently.
Monitor Café Partnerships: Track contract renewals and feedback, but avoid overreacting to isolated flavour concerns that conflict with broader positive feedback.
Prepare for Cost Increases: Strategically plan for rising operational costs to protect margins.

This represented a good step to better for me, although it still reads a bit like a report. It wouldn’t take a lot of work to make clear the ask of the audience. I’d probably give it a 6/10 as a data story.

Here is the output once it had been adapted by CoPilot for an archetypal leader – in this case a commercial leader who likes to see data and needs to make a decision on the basis of the story.

Quarterly Review: Sales Performance & Strategic Direction

Current State
We operate across three key sales channels:
Direct-to-Customer (DTC) via our website
Launched during COVID to offer premium, locally ground coffee at home.
Peak monthly sales: 20–25 units; now averaging ~10 units/month.
Channel is operating at a loss due to low volume and maintenance costs.
Café Partnerships
Previously supplied ~50 local cafés.
Two major clients (20% of café revenue) did not renew contracts.
Remaining cafés buy irregularly; 5 contracts up for renewal.
Price sensitivity and flavour profile concerns cited.
Mobile Coffee Vans
Growing interest from vendors at beauty spots and festivals.
Strong appeal due to local sourcing and low carbon footprint.
Sales are non-contractual and ad hoc, making revenue unpredictable.

Crisis
Revenue Decline: All three channels show signs of instability, with café and DTC segments in decline.
Cost Pressures: Rising minimum wage, energy costs, and production expenses are squeezing margins.
Customer Feedback Conflict: Flavour profile concerns from cafés contrast with positive feedback from mobile vendors and tasting events.
DTC Channel: Currently unsustainable and not delivering ROI.

Resolution & Recommendations
To protect revenue and position for growth, we recommend:
Exit or Restructure DTC Channel
Shut down the website or partner with niche online retailers to scale efficiently.
Reduces operational losses and redirects focus to higher-potential areas.
Prioritize Mobile Coffee Vans
Invest in relationship-building and explore ways to stabilize demand (e.g., subscription models or seasonal contracts).
Leverage brand strengths: local sourcing, sustainability, and flexibility.
Monitor Café Segment
Track contract renewals and feedback but avoid major product changes based on limited data.
Consider targeted promotions or pricing strategies to retain smaller partners.
Prepare for Cost Increases
Forecast impact of rising costs over the next 6 months and adjust pricing or sourcing strategies accordingly.

Decision Points for Leadership
Should we sunset the DTC channel or seek a partner?
How much investment should we allocate to mobile van partnerships?
Do we adjust our pricing strategy to retain café partners or accept attrition?

I really liked the clear titling and the decision points for leadership, and how these were suggested as questions that would resonate and be within the role, responsibilities and priorities for someone in the role of a commercial director. This probably took it to a 7/10 for me, but it now feels like it’s missing critical data and comes across as vague.

Here is the output for following CoPilot’s prompt around “Can you suggest key metrics to highlight”

🔹 Current State Metrics
Monthly DTC Sales Volume: Peak of 20–25 units → now ~10 units/month
Café Partnerships: 50 cafés supplied last year
Mobile Van Partners: Number of active vans (if available), frequency of orders

🔹 Crisis Metrics
Lost Revenue from Café Clients: % of total café revenue lost (e.g., 20%)
Contract Renewal Risk: Number of café contracts up for renewal (e.g., 5)
Cost Increases:
% increase in minimum wage
% increase in energy and production costs
DTC Channel Losses: Monthly operating cost vs. revenue

🔹 Resolution Metrics
Mobile Van Growth Potential:
% increase in inquiries or new partners
Customer satisfaction scores or feedback highlights
Projected Cost Savings from exiting DTC channel
Revenue Forecast Scenarios:
With vs. without café renewals
With increased mobile van engagement

Following this, CoPilot suggested refining the suggested data for the leader archetype from before (in this case a Commercial Leader focused on data driven decision making):

🔹 Current State Metrics
Monthly DTC Sales Volume: Peak of 20–25 units → now ~10 units/month
Café Partnerships: 50 cafés supplied last year
Mobile Van Partners: Number of active vans (if available), frequency of orders

🔹 Crisis Metrics
Lost Revenue from Café Clients: % of total café revenue lost (e.g., 20%)
Contract Renewal Risk: Number of café contracts up for renewal (e.g., 5)
Cost Increases:
% increase in minimum wage
% increase in energy and production costs
DTC Channel Losses: Monthly operating cost vs. revenue

🔹 Resolution Metrics
Mobile Van Growth Potential:
% increase in inquiries or new partners
Customer satisfaction scores or feedback highlights
Projected Cost Savings from exiting DTC channel
Revenue Forecast Scenarios:
With vs. without café renewals
With increased mobile van engagement

I found this really useful as it addressed my concern that I was missing key data. Whilst it doesn’t improve the data story, it does give me clear steps that I could follow to get to an 8 or even 9/10.

Was it useful / what did I learn?

I found this process incredibly useful. Writing concisely usually takes me a long time—I often have clear insights and connections in my head, but struggle to structure them in a way that is easily understood by others. It typically requires multiple reviews and iterations for me to get there.

Using AI as a partner helped shortcut this process. I could dictate my thoughts, feed them into Copilot, and get a concise, structured version without needing to perfect it upfront. What would normally take me half a day took just 30–45 minutes.

Adding a story structure brought clarity and polish. Conciseness took my data story to 3/10 in my opinion, but adding story structure elevated it to 6/10, which is a significant improvement. Whilst this was useful from a time perspective, and a powerful improvement. I found the next improvements added more value in terms of telling a persuasive story. The first was tailoring to a specific audience – this helped me think differently about how I was going to craft my story, shifting from the story I wanted to tell, to a story that a listener wanted, and perhaps needed to hear in the context of their role. Following up the prompts about metrics was, however, a significant leap for me as it felt like having a knowledgeable partner guiding me on the things I was missing – the holes in my thinking. Would I have got there on my own? Probably, because I have experience in undertaking research and analysing data, but it would have taken longer. However, it was impressive how using AI could help a more novice analyst to identify the kinds of data that would be useful. In short, using AI to support the ways that you put together data stories could create an option for more novice data users to create powerful data stories, whilst developing their analytical skills.

A challenge that went through my mind was that if someone could use AI to create a talk track through the steps above, could they simply feed it into an AI video generator and do away with the need for a presentation. On the basis of such a video, could they achieve the purpose of telling a data story (often supporting discussion around a course of action, or gaining commitment to action) faster. For me that’s an exciting idea. It’s technically feasible, but would it work in the same way to gain diverse insights before committing to action. I don’t know.

In summary, I think the hypothesis was proved. AI can significantly improve the speed and quality of data storytelling, especially when audience needs are clearly defined, but there is greater potential available in terms of who tells a better data story – humans or AI.