Whether or not you’ve simply uncovered a key information perception or have already walked by way of chart choice and storytelling, the problem is identical: How do you talk your message so your viewers understands it and takes motion?
On this tutorial, we’ll have a look at learn how to use Massive Language Fashions (LLMs) like ChatGPT, Claude, or Gemini to speak your findings, particularly while you’re brief on time, working throughout groups, or tailoring your message for various audiences.
As an instance the methods, we’ll use an instance from the Superstore dataset — a fictional retail firm that sells furnishings and expertise merchandise to shoppers.
In a earlier tutorial, we analyzed the connection between low cost ranges and revenue margins. The important thing discovering? When reductions exceed 30%, profitability drops sharply.
You don’t must be aware of that full evaluation; we’ll use this one perception all through the tutorial to point out how LLMs will help you talk clearly throughout totally different audiences. As a result of in most real-world situations, it’s not the chart that wants tweaking: it’s the message.
That is the place these instruments shine, supplying you with quick, versatile methods to form your message with out ranging from scratch, even when time is tight.
By the top of this tutorial, you’ll know learn how to use LLMs as highly effective writing assistants that may:
- Translate insights into plain language
- Rewrite messages for various tones and audiences
- Summarize chart findings shortly and clearly
- Brainstorm stronger slide titles, electronic mail blurbs, and govt summaries
You selected the precise chart. You designed it for readability. You informed a compelling story. Now let’s use an LLM to assist be certain that your viewers hears it, understands it, and acts on it.
The way to Write a Good Immediate
Earlier than we have a look at particular use circumstances, it helps to know how prompting works. Writing a transparent immediate is much less about giving directions and extra like collaborating with a junior teammate who’s quick, versatile, and keen to assist, however wants some path. If we merely say, “Summarize this chart,” we’d get one thing serviceable, however flat or generic. To get outcomes that sound extra assured and audience-ready, we have to give only a bit extra framing.
Let’s stroll by way of how to do this utilizing a key perception from our Superstore situation: the identical one we launched earlier about discounting and profitability.
The scatter plot exhibits how profitability declines as reductions enhance. The sample is evident: revenue margins stay pretty steady till a few 30% low cost, after which they start to fall sharply.
Now, let’s say we need to share this discovering with a regional supervisor who isn’t deeply aware of analytics or very snug with math. We might strive prompting an LLM in just a few alternative ways, and the outcomes we get will rely upon how effectively we arrange the context.
A obscure immediate like “Clarify this chart” doesn’t give the mannequin something to go on. It doesn’t know who the viewers is, what sort of tone to make use of, and even what sort of output we’re in search of. We’ll probably get one thing robotic, cluttered with irrelevant particulars, or framed in a means that doesn’t match our communication purpose.
Now evaluate that to a extra considerate immediate:
“I’m getting ready a brief replace for our regional supervisor. The chart exhibits that after reductions exceed 30%, revenue margins decline considerably. Are you able to write one clear, assured sentence that summarizes this discovering?”
Right here, we’re giving the mannequin:
- The fundamentals of what the info exhibits
- Who the viewers is
- What tone we wish
A typical LLM response is likely to be one thing like:
“Reductions above 30% persistently scale back revenue margins, signaling a have to reevaluate present pricing methods.”
It’s succinct, clear, and aligned with our supposed viewers. From right here, we are able to maintain iterating: ask for alternate options, strive totally different phrasings, or check variations for different communication codecs like slide titles or electronic mail blurbs.
Some of the helpful issues about working with an LLM is that this flexibility. You are not making an attempt to get the proper output on the primary strive — you are collaborating. If the tone isn’t fairly proper, you may ask for one thing extra informal. If the abstract is just too lengthy, you may say, “shorten this to a slide title.” The purpose isn’t to deal with the mannequin like a one-and-done reply machine. It’s to deal with it like a second pair of eyes, one that may assist you to rework and refine your message till it feels proper.
Within the subsequent part, we’ll have a look at actual before-and-after examples utilizing our Superstore situation. You’ll see how small immediate modifications could make a giant distinction, and how one can apply these identical methods to your personal information tales.
What LLMs Can Assist With
Beginning with the perception “When Superstore reductions exceed 30%, revenue margins begin to drop considerably,” let’s stroll by way of how this message may be tailored for various communication contexts. We’ll see how an LLM can function a versatile writing associate that helps us translate, reframe, and sharpen our message with out ranging from scratch every time.
To indicate how that works in observe, we’ll stroll by way of 4 variations of the identical perception, every tailor-made to a unique viewers and communication format: a crew Slack message, a slide title, a stakeholder abstract, and a marketing campaign headline.
A fast observe to the supervisor
We’re drafting a fast Slack message to our direct supervisor. It’s an FYI, not a proper report; we simply need to flag one thing necessary from our evaluation.
We write:
“Gadgets discounted above 30% are sometimes unprofitable.”
It’s technically true, however flat. We ask the LLM:
“Make this clearer and extra assured for a fast crew replace.”
It responds:
“Fast perception from our low cost evaluation: Merchandise with reductions over 30% persistently lose cash — could also be time to revisit our pricing thresholds.”
Why this immediate works:
- It units the purpose (“make this clearer and extra assured”)
- It offers context (“crew replace”)
- It invitations tone enchancment while not having a rewrite from scratch
The LLM response feels extra action-oriented and indicators a potential subsequent step, with out overstating the case.
A slide title for an all-hands assembly
We additionally need to share this perception with the entire firm on the month-to-month all-hands assembly. We have now one line on the prime of a slide. It must be crisp and clear. We might title the slide one thing like, “Reductions vs. Profitability”, however that lacks urgency and doesn’t assist the viewers shortly grasp the purpose.
We will get LLM help with a well-crafted immediate:
Present three variations of a slide title for our firm all-hands primarily based on the datapoint “Gadgets discounted above 30% are sometimes unprofitable.”
The LLM offers us sturdy choices:
“When Reductions Minimize Too Deep”
“Revenue Margins Disappear After 30%”
“The Hidden Value of Steep Reductions”
Why this immediate works:
- It requires the LLM to present a number of choices, permitting us to pick (or refine) the one which works greatest for us.
- It asks for a particular format (“slide title”) and a clear message (“reductions above 30% scale back revenue margins”).
- The LLM fills in tone and emphasis, serving to us discover which framing resonates greatest for a large viewers.
A abstract for regional administrators
We’re sending a brief electronic mail to regional administrators summarizing key findings from current pricing evaluation. They’re busy, so we have to sound polished and decisive.
We begin with:
“Discounting patterns are affecting revenue.”
Then we do this immediate:
“Write two sentences summarizing that reductions above 30% are sometimes unprofitable. Viewers is senior managers. Tone needs to be direct.”
The LLM affords:
“Our evaluation exhibits that product reductions above 30% persistently result in destructive margins. Adjusting low cost limits might assist protect profitability throughout classes.”
Why this immediate works:
- It defines the viewers (“senior managers”)
- Units a clear activity (two-sentence abstract)
- Specifies tone (“direct”)
These boundaries assist the LLM keep away from fluff and keep centered. Now we’ve a abstract that sounds executive-ready and a immediate we are able to reuse throughout different findings.
Serving to a colleague form a marketing campaign message
To see one other software, a advertising teammate has requested us for a fast abstract of the low cost evaluation to assist form a brand new promotion technique. They’re not in search of arduous numbers; they simply need just a few message concepts for the marketing campaign.
We inform the LLM:
“Recommend 3 brief headlines primarily based on the perception that 30% is the brink for worthwhile discounting.”
It replies:
“30%: The Candy Spot for Good Reductions”
“Go Large — However Not Too Large”
“Extra Isn’t At all times Extra: The place Reductions Damage Earnings”
Why this immediate works:
- It retains the duty open-ended (“recommend 3 headlines”) whereas nonetheless offering a key constraint (the 30% profitability threshold). This offers the LLM house to discover language variations with out veering off-message.
Now our evaluation is evident and able to help front-line messaging.
In every case, the LLM isn’t doing our pondering for us. We’ve carried out the arduous half: discovering the sample. What the LLM affords is a quick, versatile method to form our message for various conditions, tones, and audiences. It helps us say what we imply, with a bit extra precision and quite a bit much less effort.
With the precise framing, we are able to get outcomes that sound extra human and extra helpful.
What LLMs Can’t Do Properly
Massive language fashions may be highly effective writing assistants, however they’re not analysts, subject-matter specialists, or decision-makers. Similar to a well-meaning teammate, they will help you talk extra clearly, however they’re not all the time proper, and so they don’t all the time know what issues most.
Listed here are just a few necessary issues to bear in mind.
LLMs don’t know your information
They don’t perceive your small business context, and so they don’t validate the numbers you share. In case you ask a mannequin to investigate uncooked information, it might confidently generate a conclusion, even when that conclusion is wrong or primarily based on a misunderstanding of the enter. That’s why on this tutorial, we’ve centered on utilizing LLMs to talk findings, not generate them.
You’re the professional on what the info means. Use the mannequin to assist form the message, not uncover it.
LLMs may be obscure, overly generic, or factually off-base
Particularly when prompts are obscure or lack context, LLMs might return boilerplate responses that don’t match your voice or viewers. If one thing feels off, belief your instincts. Don’t hesitate to rephrase, make clear, or just begin once more.
Additionally, be cautious about utilizing any numbers the mannequin suggests. LLMs can confidently generate data that sounds believable however isn’t correct — a phenomenon usually known as “hallucination.” It’s normally greatest to offer key figures your self and deal with model-generated stats or comparisons as placeholders.
Keep away from delicate or proprietary data
Most public-facing LLM instruments retailer and be taught from inputs to enhance future efficiency. Don’t enter company-sensitive information, private data, or something you wouldn’t share externally. In case your group makes use of non-public or inner LLM instruments, comply with your crew’s tips round protected utilization.
Briefly: deal with the LLM like a sensible, quick exterior contractor who would not know your small business, your stakeholders, or your targets, except you inform it. The higher you body the duty, the extra helpful its responses shall be. And the extra critically you overview its output, the extra assured you will really feel placing these responses to work.
Recap and Takeaways
On this tutorial, we explored how Massive Language Fashions can help the communication facet of information storytelling. When you’ve carried out the evaluation and constructed your visuals, an LLM will help you:
- Translate insights into clear, plain language
- Alter tone and emphasis relying on the viewers
- Summarize or rephrase findings for various codecs
- Brainstorm sturdy messaging, from topic traces to slip titles
You already know your information. The LLM merely helps you specific what extra clearly and with much less friction.
Alongside the best way, we additionally lined learn how to write efficient prompts by pondering of LLMs like collaborators: Give context, set the purpose, ask for what you want, and iterate. You’re not aiming to get it excellent on the primary strive. As an alternative, it’s about shaping your message with extra readability, confidence, and management.
With considerate use, LLMs will help you bridge the final hole between information and choice by refining your message so it lands with the individuals who want it most.
Whether or not you’ve simply uncovered a key information perception or have already walked by way of chart choice and storytelling, the problem is identical: How do you talk your message so your viewers understands it and takes motion?
On this tutorial, we’ll have a look at learn how to use Massive Language Fashions (LLMs) like ChatGPT, Claude, or Gemini to speak your findings, particularly while you’re brief on time, working throughout groups, or tailoring your message for various audiences.
As an instance the methods, we’ll use an instance from the Superstore dataset — a fictional retail firm that sells furnishings and expertise merchandise to shoppers.
In a earlier tutorial, we analyzed the connection between low cost ranges and revenue margins. The important thing discovering? When reductions exceed 30%, profitability drops sharply.
You don’t must be aware of that full evaluation; we’ll use this one perception all through the tutorial to point out how LLMs will help you talk clearly throughout totally different audiences. As a result of in most real-world situations, it’s not the chart that wants tweaking: it’s the message.
That is the place these instruments shine, supplying you with quick, versatile methods to form your message with out ranging from scratch, even when time is tight.
By the top of this tutorial, you’ll know learn how to use LLMs as highly effective writing assistants that may:
- Translate insights into plain language
- Rewrite messages for various tones and audiences
- Summarize chart findings shortly and clearly
- Brainstorm stronger slide titles, electronic mail blurbs, and govt summaries
You selected the precise chart. You designed it for readability. You informed a compelling story. Now let’s use an LLM to assist be certain that your viewers hears it, understands it, and acts on it.
The way to Write a Good Immediate
Earlier than we have a look at particular use circumstances, it helps to know how prompting works. Writing a transparent immediate is much less about giving directions and extra like collaborating with a junior teammate who’s quick, versatile, and keen to assist, however wants some path. If we merely say, “Summarize this chart,” we’d get one thing serviceable, however flat or generic. To get outcomes that sound extra assured and audience-ready, we have to give only a bit extra framing.
Let’s stroll by way of how to do this utilizing a key perception from our Superstore situation: the identical one we launched earlier about discounting and profitability.
The scatter plot exhibits how profitability declines as reductions enhance. The sample is evident: revenue margins stay pretty steady till a few 30% low cost, after which they start to fall sharply.
Now, let’s say we need to share this discovering with a regional supervisor who isn’t deeply aware of analytics or very snug with math. We might strive prompting an LLM in just a few alternative ways, and the outcomes we get will rely upon how effectively we arrange the context.
A obscure immediate like “Clarify this chart” doesn’t give the mannequin something to go on. It doesn’t know who the viewers is, what sort of tone to make use of, and even what sort of output we’re in search of. We’ll probably get one thing robotic, cluttered with irrelevant particulars, or framed in a means that doesn’t match our communication purpose.
Now evaluate that to a extra considerate immediate:
“I’m getting ready a brief replace for our regional supervisor. The chart exhibits that after reductions exceed 30%, revenue margins decline considerably. Are you able to write one clear, assured sentence that summarizes this discovering?”
Right here, we’re giving the mannequin:
- The fundamentals of what the info exhibits
- Who the viewers is
- What tone we wish
A typical LLM response is likely to be one thing like:
“Reductions above 30% persistently scale back revenue margins, signaling a have to reevaluate present pricing methods.”
It’s succinct, clear, and aligned with our supposed viewers. From right here, we are able to maintain iterating: ask for alternate options, strive totally different phrasings, or check variations for different communication codecs like slide titles or electronic mail blurbs.
Some of the helpful issues about working with an LLM is that this flexibility. You are not making an attempt to get the proper output on the primary strive — you are collaborating. If the tone isn’t fairly proper, you may ask for one thing extra informal. If the abstract is just too lengthy, you may say, “shorten this to a slide title.” The purpose isn’t to deal with the mannequin like a one-and-done reply machine. It’s to deal with it like a second pair of eyes, one that may assist you to rework and refine your message till it feels proper.
Within the subsequent part, we’ll have a look at actual before-and-after examples utilizing our Superstore situation. You’ll see how small immediate modifications could make a giant distinction, and how one can apply these identical methods to your personal information tales.
What LLMs Can Assist With
Beginning with the perception “When Superstore reductions exceed 30%, revenue margins begin to drop considerably,” let’s stroll by way of how this message may be tailored for various communication contexts. We’ll see how an LLM can function a versatile writing associate that helps us translate, reframe, and sharpen our message with out ranging from scratch every time.
To indicate how that works in observe, we’ll stroll by way of 4 variations of the identical perception, every tailor-made to a unique viewers and communication format: a crew Slack message, a slide title, a stakeholder abstract, and a marketing campaign headline.
A fast observe to the supervisor
We’re drafting a fast Slack message to our direct supervisor. It’s an FYI, not a proper report; we simply need to flag one thing necessary from our evaluation.
We write:
“Gadgets discounted above 30% are sometimes unprofitable.”
It’s technically true, however flat. We ask the LLM:
“Make this clearer and extra assured for a fast crew replace.”
It responds:
“Fast perception from our low cost evaluation: Merchandise with reductions over 30% persistently lose cash — could also be time to revisit our pricing thresholds.”
Why this immediate works:
- It units the purpose (“make this clearer and extra assured”)
- It offers context (“crew replace”)
- It invitations tone enchancment while not having a rewrite from scratch
The LLM response feels extra action-oriented and indicators a potential subsequent step, with out overstating the case.
A slide title for an all-hands assembly
We additionally need to share this perception with the entire firm on the month-to-month all-hands assembly. We have now one line on the prime of a slide. It must be crisp and clear. We might title the slide one thing like, “Reductions vs. Profitability”, however that lacks urgency and doesn’t assist the viewers shortly grasp the purpose.
We will get LLM help with a well-crafted immediate:
Present three variations of a slide title for our firm all-hands primarily based on the datapoint “Gadgets discounted above 30% are sometimes unprofitable.”
The LLM offers us sturdy choices:
“When Reductions Minimize Too Deep”
“Revenue Margins Disappear After 30%”
“The Hidden Value of Steep Reductions”
Why this immediate works:
- It requires the LLM to present a number of choices, permitting us to pick (or refine) the one which works greatest for us.
- It asks for a particular format (“slide title”) and a clear message (“reductions above 30% scale back revenue margins”).
- The LLM fills in tone and emphasis, serving to us discover which framing resonates greatest for a large viewers.
A abstract for regional administrators
We’re sending a brief electronic mail to regional administrators summarizing key findings from current pricing evaluation. They’re busy, so we have to sound polished and decisive.
We begin with:
“Discounting patterns are affecting revenue.”
Then we do this immediate:
“Write two sentences summarizing that reductions above 30% are sometimes unprofitable. Viewers is senior managers. Tone needs to be direct.”
The LLM affords:
“Our evaluation exhibits that product reductions above 30% persistently result in destructive margins. Adjusting low cost limits might assist protect profitability throughout classes.”
Why this immediate works:
- It defines the viewers (“senior managers”)
- Units a clear activity (two-sentence abstract)
- Specifies tone (“direct”)
These boundaries assist the LLM keep away from fluff and keep centered. Now we’ve a abstract that sounds executive-ready and a immediate we are able to reuse throughout different findings.
Serving to a colleague form a marketing campaign message
To see one other software, a advertising teammate has requested us for a fast abstract of the low cost evaluation to assist form a brand new promotion technique. They’re not in search of arduous numbers; they simply need just a few message concepts for the marketing campaign.
We inform the LLM:
“Recommend 3 brief headlines primarily based on the perception that 30% is the brink for worthwhile discounting.”
It replies:
“30%: The Candy Spot for Good Reductions”
“Go Large — However Not Too Large”
“Extra Isn’t At all times Extra: The place Reductions Damage Earnings”
Why this immediate works:
- It retains the duty open-ended (“recommend 3 headlines”) whereas nonetheless offering a key constraint (the 30% profitability threshold). This offers the LLM house to discover language variations with out veering off-message.
Now our evaluation is evident and able to help front-line messaging.
In every case, the LLM isn’t doing our pondering for us. We’ve carried out the arduous half: discovering the sample. What the LLM affords is a quick, versatile method to form our message for various conditions, tones, and audiences. It helps us say what we imply, with a bit extra precision and quite a bit much less effort.
With the precise framing, we are able to get outcomes that sound extra human and extra helpful.
What LLMs Can’t Do Properly
Massive language fashions may be highly effective writing assistants, however they’re not analysts, subject-matter specialists, or decision-makers. Similar to a well-meaning teammate, they will help you talk extra clearly, however they’re not all the time proper, and so they don’t all the time know what issues most.
Listed here are just a few necessary issues to bear in mind.
LLMs don’t know your information
They don’t perceive your small business context, and so they don’t validate the numbers you share. In case you ask a mannequin to investigate uncooked information, it might confidently generate a conclusion, even when that conclusion is wrong or primarily based on a misunderstanding of the enter. That’s why on this tutorial, we’ve centered on utilizing LLMs to talk findings, not generate them.
You’re the professional on what the info means. Use the mannequin to assist form the message, not uncover it.
LLMs may be obscure, overly generic, or factually off-base
Particularly when prompts are obscure or lack context, LLMs might return boilerplate responses that don’t match your voice or viewers. If one thing feels off, belief your instincts. Don’t hesitate to rephrase, make clear, or just begin once more.
Additionally, be cautious about utilizing any numbers the mannequin suggests. LLMs can confidently generate data that sounds believable however isn’t correct — a phenomenon usually known as “hallucination.” It’s normally greatest to offer key figures your self and deal with model-generated stats or comparisons as placeholders.
Keep away from delicate or proprietary data
Most public-facing LLM instruments retailer and be taught from inputs to enhance future efficiency. Don’t enter company-sensitive information, private data, or something you wouldn’t share externally. In case your group makes use of non-public or inner LLM instruments, comply with your crew’s tips round protected utilization.
Briefly: deal with the LLM like a sensible, quick exterior contractor who would not know your small business, your stakeholders, or your targets, except you inform it. The higher you body the duty, the extra helpful its responses shall be. And the extra critically you overview its output, the extra assured you will really feel placing these responses to work.
Recap and Takeaways
On this tutorial, we explored how Massive Language Fashions can help the communication facet of information storytelling. When you’ve carried out the evaluation and constructed your visuals, an LLM will help you:
- Translate insights into clear, plain language
- Alter tone and emphasis relying on the viewers
- Summarize or rephrase findings for various codecs
- Brainstorm sturdy messaging, from topic traces to slip titles
You already know your information. The LLM merely helps you specific what extra clearly and with much less friction.
Alongside the best way, we additionally lined learn how to write efficient prompts by pondering of LLMs like collaborators: Give context, set the purpose, ask for what you want, and iterate. You’re not aiming to get it excellent on the primary strive. As an alternative, it’s about shaping your message with extra readability, confidence, and management.
With considerate use, LLMs will help you bridge the final hole between information and choice by refining your message so it lands with the individuals who want it most.
