2026-09-24 06:06:57
This post is part of the SWD + AI series—practical guidance for using AI as a thought partner across the various stages of your data storytelling work. If you’re just joining us, start with the first three installments: start with context, craft a story, and choose an appropriate visual. Explore all of our AI resources.
You’ve chosen your visuals. Now comes the work of actually designing them—and making sure they communicate clearly, direct your audience’s attention to what matters, and hold together as part of a cohesive presentation.
This is where many people spend the bulk of their time, and where the gap between “good enough” and genuinely effective communication is most visible. Whether it comes from a charting tool or AI, the first version of a graph is rarely ready to present. It needs to be decluttered, focused, and given the words that make it meaningful. A graph without a takeaway title makes your audience guess at the point. A slide crowded with data makes them work too hard to understand it. Getting this step right is what separates a presentation that informs to one that moves people to act.
It’s also a step where AI—given the right context and direction—can do a lot of the heavy lifting. Rather than building each slide from scratch, you can describe what you need, share your data, and let AI produce a first version to react to. It won’t be perfect, but it will be a starting point to work with—and getting there in minutes rather than hours changes what’s possible.
In this post, we’ll use AI integrated directly into PowerPoint—so the output is editable slides, not images. That’s an important distinction from the previous post: you’re not evaluating prototypes anymore, you’re building the real thing, and AI is helping you refine it in place.
Effective graph and slide design comes down to a few essential principles: removing what doesn’t belong, creating visual order through alignment, directing attention to what matters, and using words to make the message clear.
Declutter. Every element on a graph or slide contributes to cognitive load—the mental effort your audience spends processing what they see. Anything that doesn’t add informative value is clutter, and clutter makes your visuals feel more complicated than they are. The goal isn’t minimalism for its own sake; it’s reducing the extraneous so the essential stands out. In practice, this often means removing chart borders, gridlines, unnecessary tick marks, redundant labels, and default formatting that the tool added because it always does (not because it’s what serves your audience). When you take away what doesn’t belong, what remains gets stronger.
Align. When elements on a slide are misaligned—text starting at different points, graphs floating at inconsistent positions, titles not anchoring the content below them—the slide feels unintentional even if the viewer can’t articulate why. Consistent alignment creates a visual structure that your audience navigates without thinking about it. When design is thoughtful, it fades to the background; when it isn’t, your audience feels the burden.
Focus attention. A clean, well-aligned visual still leaves your audience to decide for themselves what matters—and they may not choose what you intended. The next step is to actively guide where they look. The most powerful tool for this is color used sparingly. When everything is the same color (or everything is a different color), nothing is emphasized. When one element is highlighted against a neutral field, the people look there immediately. Size, position, and contrast work similarly: make important things larger, place them where the eye lands first, and use visual weight to create hierarchy. The goal is to make the right thing obvious without having to say “look here.”
Use words wisely. Even the best designed graph needs words to make its point. Every graph needs a descriptive title that tells the audience what they’re looking at. Every slide needs a takeaway title—a sentence that answers the question “so what?” before the audience has to ask it. Axes should be titled. Key data points should be annotated where they add meaning. But more isn’t better; every word that isn’t earning its place is adding noise. The discipline here is the same as everywhere else in SWD: include what serves the audience, cut what doesn’t.
One more practical consideration: if your organization has a standard template or brand guidelines, start there. Colors, fonts, and layouts that align with your brand aren’t just aesthetic choices—they signal credibility and consistency to your audience. Before applying any of the principles above, make sure you’re working within your company’s visual identity. If you’re not sure what that looks like, check with your communications or design team.
These principles work together. Apply them with your audience in mind—what do they need to see, and what might get in their way? When design is working, it’s invisible: your audience isn’t thinking about the slide, they’re thinking about the message.
This is where having a shared foundation in SWD principles becomes especially useful. In earlier posts, we gave AI a few sentences of context before diving in. For this step, we recommend going further: share the SWD + AI primer with your tool before you begin. This free PDF download gives AI a grounding in SWD principles—including specific guidance on how to approach graph and slide design—and produces noticeably better results than prompting without it.
The workflow here is different from the previous posts in an important way that I mentioned earlier: rather than using AI as a standalone thought partner, we recommend using AI integrated directly into PowerPoint. This means the output is editable slides, not images. Tools like Copilot in PowerPoint and Claude or ChatGPT for PowerPoint all support this kind of integrated workflow.
You can come to this step with a draft slide you’ve already built and ask AI to help you refine it. Or you can describe what you want and let AI generate a first version to react to. Either way, AI can often get you 90% of the way there quickly—producing something clean, structured, and on-message that would have taken much longer to build from scratch. The remaining 10% is where your judgment comes in: the final tweaks to color, emphasis, wording, and alignment that make the difference between a good slide and a great one. Some of those refinements you’ll direct AI to make; others you’ll make directly yourself.
Before getting to the prompt and example, let’s review some potential pitfalls.
AI may get decluttering wrong—it can add to many labels, callouts, and explanatory text, or strip away context your audience needs to understand the visual. Don’t equate clean with clear. Make sure every element earns its place, and retain the context necessary to understand what’s being shown.
AI may misuse color—it often suggests using multiple colors to distinguish categories, not understanding that strategic, sparing use of a single color is far more powerful for focusing attention. Watch for suggestions that add color complexity rather than reduce it.
AI will describe, not recommend—left unprompted, AI tends to produce neutral, balanced responses. Push it to tell you whether your takeaway title is doing its job and whether the message is clear.
AI-generated slides need your eye—especially when AI produces something that looks polished, apply your own SWD judgment before accepting it. Is it free from clutter? Is where to look clear? Are the words right? Does it tell the audience what to see and why that matters?
Be mindful of what you share—avoid including sensitive data or personally identifying information in your prompt.
Before you begin, share the SWD + AI primer with your AI tool and ask it to use those principles when helping you design graphs and slides. Then proceed with the following.
Here is what I’m working on: [share your draft slide or graph, or describe what you want to create and provide the data]
My audience is: [briefly describe]
What this slide needs to communicate: [state your takeaway in a single sentence]
How it will be used: [for example, one slide in a live presentation, a standalone graph in a report]
Please help me create or refine this through the lens of clear, simple data communication.
With the primer shared and the prompt ready, let’s walk through an example.
If you’ve been following this series, you’ll recognize this scenario. I’m a People Analytics Manager at a mid-sized consulting firm, working on a presentation to recommend a change to our hybrid work policy. In previous posts, I identified my audience and formed a Big Idea (post 1), planned the story and developed a narrative arc (post 2), and chose the visuals I wanted to build (post 3). Now it’s time to design the slides.
For this step, I used Copilot in PowerPoint—which means the output is editable slides rather than images I’d need to rebuild from scratch. Before diving in, I shared the SWD + AI primer with Copilot using the suggested opening on the primer’s first page, giving it a grounding in SWD principles to work from.
After Copilot confirmed it would use the foundation learned from the primer, I followed with this prompt and a table of my summarized data:
After doing a bit of “thinking,” Copilot asked me which visual direction should frame this high-stakes policy finding, outlining the options: Editorial Ivory and Red, Crisp White and Coral, Warm Paper and Ink, along with an option to enter my own specifics via text. I responded with my company’s standard font (Monserrat) and image of our color palette.
Here is the slide it created:
Copilot’s initial slide was a strong starting point: structured, on-brand, and closer to presentation-ready than anything I could have produced from scratch in the same time. But it wasn’t finished. The next step was to iterate, and I found it useful to divide that work into two categories: changes that require thinking and reconsideration, and changes that require craftsmanship and fine-grained execution. The first category is where I kept working with Copilot; the second is where I took over myself.
For changes where I wanted AI to reconsider the communication (not just move objects around), I kept the conversation going in the prompt window.
The most important issue was message hierarchy. The slide had three competing statements: a headline, a subtitle, and a takeaway at the bottom all making similar points. That’s too much repetition, and it dilutes the impact of each. I asked Copilot:
Review the title, subtitle, and takeaway at the bottom. They feel repetitive. Recommend a clearer hierarchy that communicates one primary takeaway, with supporting text only where it adds useful context.
This is a perfect AI task: there are multiple legitimate ways to solve it and I wanted its thinking, not just execution.
I also pushed on visual emphasis. The story is primarily about early-tenure employees, but the navy lines for leadership were fairly prominent, competing for attention. I asked:
The main story is the decline among early-tenure employees across all three role types. How would you strengthen focus on that pattern while keeping the other tenure groups available as context?
Finally, I asked it to pressure-test the headline itself. The graph shows performance ratings, not support directly—and “hybrid work is widening the support gap” is a stronger claim than the data strictly supports. I asked:
Review the headline against what the data actually supports. Is “support gap” too strong given that the data measures performance ratings? Suggest alternatives that preserve the intended message without overstating the evidence.
Here are Copilot’s responses:
I confirmed the outlined changes and indicated I wanted the primary slide title to focus on the decline in early tenure performance after the hybrid work policy was introduced. After Copilot made the changes, the slide looked like this:
Once the conceptual decisions were made, I stopped prompting and took over myself. Detailed design work—exact font sizes, spacing, alignment, line weights, label positions, panel widths and margins—is faster to do directly than to describe to AI. The awkward label positioning on the red data points, for example, is exactly the kind of thing where I could spend five prompts trying to get Copilot to do something I can fix in a matter of seconds.
That said, I did turn back to Copilot for a few specific tasks where it genuinely saved time. Changing the y-axis scale across all three panels to run from 2.5 to 5.0—with consistent trailing zeros on whole numbers—was fast for Copilot to execute and would have meant a lot of clicking through graphs and menus for me. I also asked it to repeat the y-axis labels on the right of the panel group to facilitate comparisons, and to remove the x-axis tick marks. One thing to watch: when Copilot made this last change, it undid some of the changes I’d already made manually, so I had to ask it to restore those—and explicitly tell it not to touch what I’d done myself.
The work I kept for myself was largely about language. AI writes polished copy, but just because it’s polished doesn’t mean it’s right. The words it chose were reasonable starting points, but they weren’t exactly what I wanted to say. Reading every word critically and rewriting where needed is an important step that only you can do. Beyond the language, I removed the extraneous elements, adjusted label positioning and spacing, modified the size and weight of some text, colored the headline to tie visually to the relevant data, added light gray background shading to the Collaborative/client-facing panel, and changing the gridlines on that panel to white. (Yes, I usually remove gridlines, but in this case I wanted to keep them to make it clear that the same y-axis range applies across all three graphs, and so people can easily estimate the non-labeled data points.)
After about ten minutes of this work, here is my resulting slide:
This same approach can work for text-based and concept slides—those that carry your message through words, diagrams, or a simple visual rather than a graph. Share the primer, describe what the slide needs to communicate, and ask AI to create a first version. As with the data slide, expect to do the conceptual refinements in the prompt window and the fine-grained language and design work yourself. To reiterate: the language AI chooses will often be close but not quite right—read it critically and rewrite where needed.
If you find an example slide you want to emulate—a layout, a structure, a visual approach that works well—you can share this image with AI alongside your content brief and ask it to build something similar. This is particularly useful for data slides where the design logic is already solved and you just need to apply it to your own data (Alex explored this approach for a recent makeover). This is a powerful shortcut that can save significant time while keeping you in control of the final result.
Before wrapping up, I wanted to try one more thing: asking Copilot to build out the rest of the deck from the takeaway titles we developed in the story planning post. Rather than designing each slide one at a time, I shared all eleven titles and asked Copilot to create a full draft deck. It wasn’t perfect, but it got me further faster—similar to what we saw with its initial drafts of the slides—and gave me something concrete to continue to refine.
Below is the slide sorter view: a title slide, followed by the eleven takeaway titles we crafted previously, each on its own slide and ready to be built out—including slide 7, which we designed in this post. In a matter of minutes, Copilot turned a list of titles into a structured deck. The story arc is visible, the narrative is in place, and the hard thinking is already done. From here, I can work through the remaining slides in the same way I did with slide 7: prompting Copilot with specific guidance to build a first version for each, iterating on the conceptual decisions in the prompt window, then taking over myself for the language and detailed refinements. What remains is the craft—which is the part I truly love—and I have a clear path to get from here to the completed deck.
Don’t forget: if you haven’t already, download our free SWD + AI Primer to help apply SWD principles when working with AI.
2026-09-19 04:34:02
Yesterday, a participant in one of our workshops asked a question that I’ve been thinking about ever since. She works with researchers and statisticians, and the idea of telling a story with data can make some of them uncomfortable. Doesn’t telling a story mean choosing a perspective? And doesn’t choosing a perspective introduce bias? Wouldn’t it be more objective to simply show the data and let people draw their own conclusions?
I understand the instinct. I suspect the word story is actually part of the problem. It can sound like we’re asking people to embellish the facts, manufacture drama, or cherry-pick evidence to support a predetermined conclusion. That isn’t what I mean when I encourage people to tell stories with data.
But I also think “story doesn’t work for me” can become an easy excuse. It lets us focus on all the reasons storytelling can’t work in our particular situation instead of thinking openly about how it could. Storytelling with data isn’t a rigid formula. It needs to flex—to the audience, the setting, the type of work, and what the evidence can reasonably support. The story a researcher tells may look quite different from the story a salesperson tells, and it should. The challenge is figuring out what storytelling means in your situation, not dismissing it because one version doesn’t fit.
That brings me back to bias. Let’s distinguish between bias and perspective. We should absolutely work to minimize biases that distort our research or analysis. But that’s different from pretending we can eliminate perspective from the process altogether.
Long before you decide how to communicate your findings, you’ve made choices. What did you decide to study? What did you measure? How did you measure it? Which questions did you ask? What data did you collect? What did you analyze? Which comparisons did you make? Even the decision to put a particular graph in front of someone—and not the dozens of other graphs you could have made—is a choice.
There is no magical moment when we can simply “show the data” without having influenced what someone sees. The alternative to story isn’t objectivity.
I don’t think having a perspective is a bad thing—quite the contrary.
If you’ve done the research or analysis, you’ve likely spent considerably more time with the data than the people you’re communicating it to. You’ve explored it. You’ve asked questions of it. You’ve seen what surprised you and what didn’t. You’ve investigated anomalies. You know its limitations. You understand the context surrounding it.
Why would you discard all of that knowledge at the moment you communicate?
I’ve argued for years that our audience shouldn’t have to guess what we’re trying to tell them. Back in 2014, I wrote about the importance of leading with story: if you’ve done the analysis, you’re in a unique position to interpret the data and help others understand what it means. A few years later, in a post titled so what?, I made a similar argument about our responsibility as communicators to make the point clear rather than leaving our audience to figure it out themselves.
I believe that even more strongly today. When you don’t offer your perspective, you don’t eliminate interpretation—you outsource it. Your audience will make sense of the data somehow. They may arrive at the same conclusion you did, but they may also miss something important. They may misunderstand what they’re seeing. Someone could even selectively use your data to advance a conclusion the evidence doesn’t support.
“Just showing the data” doesn’t protect it from bias, misinterpretation, or nefarious use. In some cases, your silence may make those things more likely.
Perhaps the idea of telling a story still makes you uneasy. Fine. Start with something smaller: What perspective can you bring?
Given everything you know about the data, what do you see? What deserves attention? What might someone reasonably conclude? What decisions or conversations could this information inform?
There doesn’t have to be a single definitive answer. This is one of the places where we can flex what we mean by story. In exploratory or research settings, perhaps you surface several perspectives: Here’s one way to interpret what we found. Here’s another possibility. Here are the questions I think these findings raise. You can give people a useful starting point without pretending the data supports a certainty that it doesn’t.
Bringing perspective also doesn’t mean overstating what the data can tell us. Quite the opposite. Part of your responsibility is to communicate uncertainty, limitations, caveats, and plausible alternative explanations. You don’t want your work to have more impact than the evidence warrants.
But fear of having too much impact isn’t a good reason to aim for none.
I think this is ultimately where the discomfort with story can lead us astray. “My role is simply to inform” sounds admirably objective. Sometimes, though, it can also be a way of avoiding the harder work of deciding what the information means and helping someone else make sense of it.
You can be rigorous and have a point of view. You can tell a story and acknowledge uncertainty. You can direct attention and show people the evidence they need to challenge your interpretation. These aren’t contradictions. They’re part of communicating responsibly.
So if your immediate reaction is storytelling doesn’t work in my situation, resist it. Ask a different question: What could storytelling look like in my situation? Maybe it’s a strong recommendation. Maybe it’s a clearly articulated finding. Maybe it’s several possible interpretations and the evidence supporting each. Maybe it’s simply directing your audience toward what you believe deserves their attention and explaining why.
Flex the concept to fit the situation. Bring your knowledge, context, and perspective to the data. Then give your audience what they need to question it, challenge it, build on it, or act on it.
You won’t eliminate perspective by refusing to tell a story. You’ll simply leave the story to someone else.
2026-09-03 19:15:31
There’s a lot to learn when it comes to communicating effectively with data. You can improve your charts, strengthen your narrative, get better at understanding your audience, or sharpen your presentation skills. But you probably don’t need to work on all of those things right now. A useful question to ask yourself is:
What skill would make the biggest difference in my work today?
We found ourselves thinking about this recently as we looked back across 100 episodes of the storytelling with data podcast. Rather than simply asking our team to pick their favorite episodes, we asked them, “When would you recommend this episode to someone?”
The answers revealed five different areas to focus on, depending on what you’re trying to improve right now.
Maybe you’re relatively new to data storytelling. You know you want to communicate more effectively, but you’re still figuring out what “good” looks like. At this stage, focus on the fundamentals: choosing appropriate charts, reducing clutter, using color intentionally, directing attention, and making your message clear.
Don’t wait until you feel like you know enough to start creating. In one of our early podcast episodes, Andy Cotgreave offers advice we still come back to: create, create, create. Practice gives you a way to discover what you don’t know yet. Your early work might make you cringe later—and that’s evidence that you’re improving.
Here are the podcasts episodes that we recommend at this stage:
Once you can make a decent graph, you might be wondering, “How do I turn a collection of graphs into something people want to pay attention to?” This stage is where you start thinking more about narrative, structure, tension, flow, and the human element of your communication.
It also means getting comfortable with iteration. Your first attempt doesn’t have to be your final one. One of our favorite metaphors from the podcast comes from Cole’s son Avery, who described an early draft as a “sloppy copy.” Data stories need sloppy copies, too. Get the ideas out, move them around, test them, get feedback, and refine.
Here are the podcasts episodes that we recommend at this stage:
Knowing how to create an effective data story doesn’t automatically tell you what story you should tell. At this stage, the questions get bigger: What does my audience need? What should I show? What can I leave out? What decision am I trying to enable? How should I adapt when I have limited time, space, or resources?
This is also where slowing down can help you move faster. Before jumping into an analysis, ask questions. Before presenting everything you found, consider what your audience needs to know.
Here are the podcasts episodes that we recommend at this stage:
Now that you’ve made your data understandable, the next stage you might be working on is to increase your influence on your audience. Your audience gets the data perfectly well; they just aren’t doing anything about it.
Influence requires us to think beyond the graph itself. We have to consider how people interpret information, what they care about, what might make them resistant, and what will help them feel confident taking the next step. It’s especially valuable to get feedback at this stage. The work makes perfect sense to you because you created it. Seeing it through someone else’s eyes can reveal assumptions, confusion, or unintended interpretations you would never catch on your own.
Here are the podcasts episodes that we recommend at this stage:
Finally, your slides are solid, your story is clear, your recommendation makes sense, but you need to get better at delivering it. This stage is about enhancing your presentation skills. Cole makes a strong case for practicing saying your story out loud. Stand up. Think about your posture. These small changes in how you deliver your message can shift how both you and your audience experience it.
Here are the podcasts episodes that we recommend at this stage:
Circling back to where we started, don’t overwhelm yourself by trying to improve everything at once. Focus on the skill that will make the biggest difference in your work right now. These five areas aren’t necessarily a linear progression. You might be highly skilled at creating graphs but uncomfortable presenting them. Or perhaps you’re a confident presenter who needs to become more strategic about what you show.
So instead of asking, How do I get better at data storytelling? try making the question smaller:
What skill would make the biggest difference in my work today?
Start there by diving into the playlist that’s most useful to you right now. Also, we often publish new podcast episodes, and you can follow the show on your favorite podcast player so you never miss an episode: Spotify, Apple Podcasts, Overcast, or on the web.
2026-08-26 21:07:52
During a recent virtual public workshop, a participant raised an interesting question. Paraphrased, it went something like this:
"I can completely see the value of what you're teaching, but how do I tell a story that my audience doesn't want to hear, or doesn't want to focus on right now?"
My initial reaction was surprise. Why would we share something our audience isn't interested in? After a few follow-up questions, it became clear that the person they were communicating to was expecting an update framed in a particular way. However, the exploratory analysis and resulting findings pointed in a different direction than the agreed-upon strategy.
I realized the challenge wasn't necessarily that the audience didn't want to hear the story. More likely, they were expecting a different one. When that's the case, acknowledging those expectations before introducing new findings helps your audience make the transition from what they thought they would hear to what the data is actually saying.
This is a difficult situation. You need to understand why the audience holds those expectations, while still surfacing what the analysis shows. With that in mind, those were the thoughts I began with in my reply:
Understand audience constraints. Resistance isn't always a sign someone is ignoring the data; they may have priorities or constraints that weren't visible during the analysis. If you can, talk with your stakeholders directly to understand their thinking. If you can't, ask for input from others in similar roles. They may be able to help you understand why the audience is so invested in the story they expected. Understanding that context helps you proceed thoughtfully, with your audience's perspective in mind, and may even turn a skeptic into a supporter before you're in the room.
Get clear on your message. This is where taking the time to form a Big Idea—a single sentence that captures your point of view and, in particular, what's at stake for your audience—helps. Thinking through the benefits or risks depending on how they respond to your recommendation is what makes it feel relevant and clear. (For more on the Big Idea, including how AI can act as a thought partner to help compose your idea, check out this post.)
Consider your framing. If the topic is controversial, think about how to position the message constructively, and whether to soften any recommendations. Rather than "we've seen this in the data, so you must take this action," try something closer to: "we'd like to share some insights, outline the options, and get your thoughts."
Separate findings from recommendation. An audience can accept what the data shows while still disagreeing with the proposed action. Keeping the two distinct creates space for a more productive conversation.
So how do you tell a story that your audience doesn't want to hear?
By first making sure it's a story they understand, care about, and see as relevant to the decisions in front of them. If the content has been tailored to the specific audience and gives them a clear sense of why the message matters—whether in terms of opportunity, risk, or impact—even a difficult message is more likely to be received in the right spirit.
If this scenario sounds familiar, fellow data storytellers Alli and Ryan dig into a related challenge on a recent podcast episode:what to do when a leader asks you to find data to prove them right. It's a useful listen if you're facing something similar.
The goal isn't to avoid an uncomfortable truth, but to make it clear, relevant, and useful in service of the organization's broader success.
2026-08-18 01:55:48
This post is part of the SWD + AI series—practical guidance for using AI as a thought partner across the various stages of your data storytelling work. If you’re just joining us, start with the first two installments: start with context and craft a story. Explore all of our AI resources.
Now that we have the story planned, it’s time to start developing the content that will support our message and narrative. When data is part of that, a good first step is choosing a visual that aids in comprehension. The right graph makes your point immediately clear. The wrong one makes your audience spend their mental energy decoding the graph instead of understanding your message.
This is where people sometimes stumble. They use the first chart that comes to mind—or simply carry forward the one they used during exploratory analysis. But a graph that works for exploring data isn’t necessarily the best for communicating it. Your audience and takeaway should drive the choice. By this point, you’ve already done that work: you know your audience, you’ve planned your story, and you’ve written takeaway titles that tell you exactly what each graph needs to show. Let those sentences guide your design.
A handful of graph types meet most everyday needs—here are the ones we use most at SWD (from storytelling with data: before & after, Wiley 2025):
Bar charts compare across categories. Dot plots and slopegraphs emphasize change between two points. Line graphs show change over time. When in doubt, familiar works: your audience shouldn’t have to learn how to read a graph before they can understand your message. Use something less familiar only when it reveals an insight that would otherwise be difficult to see. (For more on when to use these and other common visuals, check out the SWD chart guide.)
Choosing an effective visual is rarely a straight line from first attempt to finished graph. You try one form, realize it emphasizes the wrong thing, try another, get closer. This iterative process—experimenting with different views of the same data to find the one that best serves your message—is one of the most valuable things AI can accelerate. The outputs are often rough and may take some back-and-forth to get right. Still, rather than spending time building a chart only to realize it doesn’t work, you can prototype options quickly, evaluate them against your takeaway, and commit to a direction before investing in the final build.
If you’re following this series in order, you already know what you want to show—your takeaway titles from the storyboarding step outline this. If you’re coming to this post independently, or you’re still in exploratory mode, AI can help you figure out what’s worth visualizing first. Share your data and ask it to surface patterns or trends worth highlighting. Once something interesting emerges, pause to articulate the takeaway before moving into prototyping. Either way, that articulation step is key—it’s what will turn a prototype into a purposeful visual.
From there, the workflow is straightforward: tell AI what you want to communicate, share your data, and ask it to suggest options. You don’t need to clean or aggregate the data first—that’s part of what AI can handle. For each prototype, ask AI to explain the tradeoffs: what each option makes easy to see and what it obscures. You’re not looking for AI to make the decision; you’re using it to quickly surface options you can evaluate with your own judgment and knowledge of your audience and goals.
Once you’ve chosen a direction, build the chart in your tool of choice—either yourself or with the help of AI features within your tool. We’ll explore the latter in more detail in the next post in this series. Either way, it’s important that you remain responsible for the accuracy of what’s shown.
Before getting to the prompt and example, let’s review some potential pitfalls.
AI optimizes for the data, not the message—without clear direction, AI may try to visualize everything you give it or suggest a chart that represents the data accurately but doesn’t communicate your point. Lead with your takeaway, and be explicit about which data matters to the story.
AI may suggest unfamiliar chart types—it may recommend something technically interesting but unfamiliar to a general business audience. Push back if it suggests something your audience is unlikely to recognize immediately.
AI-generated prototypes aren’t finished work—use them to evaluate direction and spark ideas, then build the final chart yourself. When you build it from your own data, you also control the accuracy of what’s shown.
Be mindful of what you share—avoid including sensitive data or personally identifying information in your prompt. If your raw data contains details you’d rather not share, aggregate into a summary table or anonymize it before passing it to AI.
A note on tools: chart rendering capability currently varies significantly across toolsand account tiers. For the planning and thinking steps in this series, any tool can assist you. For visual prototyping, however, you’ll get better results with tools that can render actual chart images. Even then, it sometimes takes multiple prompts to get actual images in the output, rather than text-based approximations of the charts.
In my testing for this article, Copilot and the free version of ChatGPT struggled to render chart images for visual comparison, while the free versions of Gemini and Claude generally produced stronger results. These capabilities are evolving rapidly, so your experience may differ as the tools improve.
The paid tiers generally produced the strongest results overall, so if you have access to one, it’s worth using for this step. If not, the free version of Gemini is currently your best bet among the tools we tested. If you want to maximize your options, you could even copy and paste your prompt across multiple tools to generate a set of approaches to choose from or iterate upon.
If you’re continuing in the same AI conversation from one of the previous steps (start with context, craft a story), your context is already established and you can jump straight to the prompt below. If you’re starting a new conversation, take a moment to briefly orient AI: describe your audience and note the key message you’re trying to communicate visually. A few sentences should suffice.
I’m working on a data visualization for a presentation and want to explore chart options. I’ll share the context and my data. Act as a thought partner to help me identify and prototype effective visual options.
My audience is: [briefly describe]
I want to show: [state your takeaway in a single sentence]
How it will be used: [describe the context—for example, a single slide in a live presentation, a standalone graph in a report or email—and the goal, such as informing, persuading, or prompting a specific action]
Here is my data: [paste your data, aggregated table, or describe the dataset]
Please generate 2–3 charts that could work well for this message and audience. Render the graphs so I can evaluate them visually. For each, briefly explain what it makes easy to see, and what it might obscure.
Before making suggestions, ask me any questions that would help you give better input.
Note: if the output looks code-like or uses text and symbols to approximate the charts rather than rendering them visually, follow up with: Can you render the graphs as actual images so I can compare them visually?
Let’s look at an example.
If you’ve been following this series, you’ll recognize the scenario. I’m a People Analytics Manager at a mid-sized consulting firm. My team has completed a thorough analysis of the company’s hybrid work policy—examining performance ratings, in-office attendance patterns, collaboration network data, and attrition trends. We have a recommendation: move from the current three-days-in-office, two-days-remote policy for all employees to a differentiated approach based on role and team type.
In the first two posts, I worked through the context and story planning stages. I identified my audience and what’s at stake, formed a Big Idea, built a storyboard, and developed a narrative arc with takeaway titles for each planned slide. Now it’s time to start creating the actual content—and for several of the slides, that means choosing and building effective graphs.
I’ll work through two of those graphs here, each supporting a different point in the story. For the first—showing how early-tenure attrition has increased since we implemented the hybrid policy—I used Gemini. As mentioned, this tool had the best output across the free tiers that I tested (the others were Copilot, ChatGPT, and Claude). For the second—showing how the policy is affecting different employee groups in opposite ways—I used ChatGPT Plus. I followed the same prompting approach across both. In practice, you would likely continue with the same tool; I’m varying which I partner with here to give you a general sense of the output and how different tools handle this task.
I gave Gemini the general prompt shared earlier, with the following specifics:
My audience is: a leadership team with divided opinions and stakes in the outcome
I want to show: early-tenure attrition has spiked since we implemented the hybrid work policy
How it will be used: this will be a single graph that is part of a larger live presentation; the ultimate goal is to persuade the leadership team to move to a new differentiated policy
I shared a data table that summarized attrition rates pre- and post-hybrid policy by role type and tenure.
Gemini asked a few clarifying questions before proceeding—useful for orienting the tool, though by this point in the process I already had clear answers to all of them. It posed questions about the proposed differentiated policy (whether I’d be advocating flexibility based primarily on tenure, role type, or both), where the expected pushback would come from, and whether the leadership team prefers traditional graphs or if something less familiar would be acceptable.
After answering the questions, Gemini gave me three options: a slopegraph, grouped bar chart, and dumbbell graph. It explained that it chose gray to signify the Before Policy baseline and the Mid and Senior tenure segments where no dramatic change occurred, and red to highlight the Early tenure attrition post-policy, where the change was most dramatic.
These were accompanied by a table that explained for each why it’s powerful (what it makes easy to see) and what it may obscure (trade-offs). For example, for the slopegraph it stated, “It shows that all early-tenure groups were impacted, but highlights that the Collaborative group had the most extreme shift. The ‘story’ is instantly visible.” For trade-offs, it shared that the grouped bar chart “design can become very visually heavy and busy with 18 bars,” and that “if the audience is not familiar with dumbbell plots, they might need a moment to understand that the dots represent ‘Before/After’ points.”
Gemini recommended the slopegraph for this situation. It also shared the following to help think through how each might work in a live setting:
Gemini gave me some decent options here. I prefer the first two; the dumbbell, though interesting, feels unnecessarily complicated, and would take a lot of explaining before we could focus on what the data is showing. I agree with Gemini, that the slopegraph in particular makes the change between before and after the policy easiest to see (both where things have been stable in the higher tenured groups and where it clearly has not for the early tenure employees). The familiarity point about the grouped bar chart is worth keeping in mind.
If I needed a quick and dirty view for my own use, either of these would give me a useful starting point. Given the high stakes in my situation, I’ll want to recreate and customize the design for my audience. As I think ahead to the live presentation, I can imagine starting with overall attrition by tenure in a familiar grouped bar chart. From there, I could transition to a slopegraph, using the movement from one form to the other to make sure my audience knows how to read the slightly less familiar visual. Once that structure is established, I can move to the panel of three slopegraphs showing the breakdown by role type. I’ll recreate the visuals for this so I can have full control over the design details. I’ll use AI to help with it in the next post.
In the meantime, let’s look at some options for another important visual in my presentation.
Next, I worked with ChatGPT Plus. I gave it the general prompt with the following specifics:
My audience is: a leadership team with divided opinions and stakes in the outcome
I want to show: the hybrid policy is hurting the employees who need support most
How it will be used: this will be a single graph that is part of a larger live presentation; the ultimate goal is to persuade the leadership team to move to a new differentiated policy
I shared a data table that summarized performance ratings before and after the hybrid policy by role type and tenure.
ChatGPT’s questions were more analytically focused than Gemini’s—asking about sample sizes, confidence intervals, and cohort definitions. It also asked how much to editorialize in the graphic itself (not yet: I’ll do that in a later step). I answered the questions and asked it to proceed.
Here are the options ChatGPT suggested and the tradeoffs for each:
ChatGPT also offered “One additional idea I’d seriously prototype—a 2x3 small-multiple slopegraph.” It originally shared a prototype that was a little messy (see below). Note the narrative it outlines to accompany it—I found this a useful reminder of how the graph fits into the broader story, even if the visual itself needed work.
When I asked it to clean up the image, it confirmed I was okay with a mockup rather than a data-perfect chart, then shared the following:
Like it did for attrition, the slopegraph makes the areas of change stand out among the mostly flat lines. Looking back at ChatGPT’s initial suggestions, I had thought the diverging bar chart showing change from baseline would be workable—but this small-multiple slopegraph is clearer. We get the absolute numbers in addition to the change, whereas the diverging bars only showed the latter. Using the same graph type as the attrition chart has another advantage: by this point, my audience will already know how to read it, reducing the cognitive load for interpreting this data.
More broadly, this exercise reinforced a few things I’ve learned about working with AI. AI prototyping for visual choice is still imperfect. Getting usable chart images sometimes takes more back-and-forth than you’d like. This should improve with time. Even today, though, it’s often faster than sketching by hand or iterating directly in a graphing tool, and it surfaces concepts you might not have considered—particularly if you’re still building your repertoire of visual approaches. A few things worth trying: ask for more options if you want a broader set to evaluate, share a rough sketch with your tool if you have a specific idea in mind, or bounce between tools to see how different ones handle the same data.
The visuals I’ve chosen here are starting points, not finished graphs. In the next post—designing effective graphs and slides—I’ll return to some of these and work through how AI can help refine them: cleaning up clutter, focusing attention, and making them presentation-ready.
In the meantime, if you want to go deeper on using AI for data storytelling, watch the recording of our recent live event, where Simon and I share additional tips and examples.
2026-08-12 00:12:22
One of our top tips for explanatory communications is to use color sparingly and purposefully to help your audience understand your data and message. Color should be an explicit choice, not something your software applies by default, whether that's a graphing tool or an AI assistant generating your first draft. These tools can build a chart in seconds. They might even add highlighting on their own. But they don't know which data matters most to your audience. That call is still yours. Used thoughtfully, color is often one of the quickest ways to improve a graph.
If you struggle with how to do this in Excel (or with charts embedded in Microsoft Word or PowerPoint), this post will walk you through the steps.
Let’s consider an Excel graph, which shows year-over-year (YoY) change in sales for cat food brands from a pet food manufacturer. This example comes from exercise 4.2 of storytelling with data: Let’s Practice! You can download the data to follow along.
A good way to start is with shades of grey and then use color only where you want your audience to focus. A greyscale chart gives you a neutral foundation and serves as a reminder that more work needs to be done before it's presented, making it easier to be deliberate about where color belongs.
Begin by removing Excel’s default colors and replacing them with a shade of grey. Right-click on the chart and choose Format Data Series.
In the Format pane on the right, go to Fill & Line (the first tab with a paint bucket icon) and set the fill to a neutral grey.
If your graph had multiple series, you’d repeat as needed for the other bars until the full chart is in grey.
Tip: once you’ve created a clean greyscale version, save it as a reusable template so you don’t have to repeat those steps.
Let’s assume we want to call attention to the brands that declined year over year. To focus on these brands, we could choose to only color those bars. For accessibility purposes, we’ll avoid red and instead choose orange, a hue that reinforces the negative sentiment.
To color only the declining bars in Excel, click once on the bar you want to color. Then, click again to select just that specific bar (this avoids coloring all the bars at once). Right-click and choose Format Data Point. In the Format pane, go to Fill & Line (paint bucket icon). Under Fill, choose Solid fill and select the desired color.
Repeat for each declining brand, leaving the increasing brands in grey. Or, as a shortcut, you can simply click on the next bar and use Cmd+Y on Mac or Ctrl+Y on Windows to repeat formatting.
Coloring just the negative values tells your audience exactly where to look. Now imagine that within the declining brands, we want to specifically highlight the two brands that decreased the most: Fran’s Recipe and Wholesome Goodness. We could make those orange and everything else grey, but that would lose the signal for brands with lower sales this year versus last year. Instead, let’s vary the color's intensity to draw extra attention to the two biggest drops.
There are at least three ways to do this:
Color the two bars a darker orange and the others a lighter orange.
Change the transparency for the bars you want to deemphasize.
Overlay a transparent white shape on the other bars to create a stronger visual emphasis on the two bars at the top.
To manually vary the color in Excel—the first option—follow the same steps above but select a lighter or paler orange for the declining brands that are not Fran’s Recipe and Wholesome Goodness. Alternatively, you could keep the same color hue and adjust the transparency within the Fill menu (second option).
For the transparent shape, or the third option, insert a rectangle object (this shape works well for bar charts) over the other decreasing brands.
Then format the rectangle with a white fill, no outline, and the desired transparency.
By varying the color intensity, we can highlight all the declining brands while bringing greater focus to the largest negative changes.
Imagine we need to share this graph for an audience to consume independently. If we are not able to walk through the information with them, we can pair the designed chart with explanatory text to make the point clear. A text annotation next to the data, with key words bolded and colored the same orange, visually ties the words to the data, making the takeaway quick to see and understand.
Color should support your message, not distract from the data. The best charts use a restrained palette, purposeful highlighting, and clear words to guide the audience to the intended takeaway.
For more Excel how-to’s, check out our tutorial library. And if you want to see these same concepts applied to a line chart, look at our related post on emphasizing a data point.