You've done the analysis. The notebook works. The insights are solid. Now you need to share it with stakeholders.
So you start the ritual. You screenshot the chart and paste it into Google Slides. You format the table in a separate doc because Slides can't handle it properly. You copy the key numbers into an email, maybe add some context, and send everything off with a "let me know if you have questions."
Two weeks later, someone asks about a number in the report. You open the notebook to check. The number doesn't match what's in the slides. Did you update the analysis after you sent the report? Was it a typo when you copied? Did someone else change the underlying data?
You don't know. And that uncertainty is the real problem.
Every time you copy results from a notebook into a report, you're creating what I think of as information debt. Each screenshot, each pasted number, each manually formatted table becomes a liability. The connection between the claim and the evidence snaps the moment you hit Ctrl+C.
The first casualty is always data lineage. Once that chart becomes a PNG in a slide deck, nobody can trace it back to the code—or the data—that made it. If a stakeholder wants to verify a number, they'd have to find the right notebook, locate the right cell, hope the notebook still runs, and hope the data hasn't changed since you ran it. Most people won't bother. They'll either trust you blindly or quietly doubt everything.
Then there's drift. You find an error in your analysis, or new data comes in, so you update the notebook. You mean to update the report too, but you're busy with three other things. Now there are two versions of the truth circulating, and you might not realize it until someone catches the discrepancy in a meeting. That's a fun conversation to have with your VP.
Version control doesn't save you either. Git tracks your notebook beautifully—every change, every commit. But Git has no idea your slide deck exists. The report lives in Google Drive or SharePoint, completely disconnected from the analysis that produced it. When someone asks "what version of the analysis is this based on?" you're back to guessing.
This isn't a skills problem. Every analyst I know has experienced this. It's a tools problem. Notebooks are built for exploration. Slides are built for communication. The gap between them is where trust goes to die.
A connected report maintains a live link between the output and the code that generated it. This sounds like a small thing, but it changes the entire dynamic.
When you embed a chart in a connected report, that chart knows which cell produced it. The notebook version is recorded at the moment of embedding. If you update the notebook and re-run it, the report can show the fresh output. And if a stakeholder ever wonders where a number came from, they can click through to the source.
This is what Margin's briefs do. You don't copy a chart—you promote it. The artifact stays linked to its origin. The metadata that normally gets thrown away is preserved instead. Data lineage isn't something you have to document manually; it's built into the system. Anyone can trace a claim in your brief back through the notebook cell to the dataset that underlies it.
The workflow becomes much simpler. You do your analysis in a notebook, just like before. When you're ready to share, you create a brief that's linked to that notebook. Instead of screenshotting outputs, you insert them directly—the chart, the table, the key visualization. Then you share a link.
That's the whole process. One link. Full traceability. No export step. No format conversion. No hoping you remembered to update everything.
When you make changes to your analysis and re-run the notebook, the brief reflects those changes. When a stakeholder clicks on a chart, they can see it came from cell 14 in your customer_analysis notebook. The question "where did this come from?" finally has an answer that doesn't require archeology.
For an individual analyst, connected reports save time and reduce the low-grade anxiety of wondering if your reports are current. But the benefits compound significantly when you're working on a team.
Code review suddenly extends to reports. Instead of reviewing notebook code in isolation and hoping the report matches, reviewers can see the analysis behind every claim in the final deliverable. If something looks off, they can trace it back and understand why.
Updates become auditable in a way they never were before. You can see what changed in the analysis and when. If someone asks why the Q3 numbers are different from last week's report, you have an actual answer instead of a shoulder shrug.
Stakeholders start trusting the numbers more because they can verify them. This might seem like a small thing, but I've seen teams where every meeting starts with fifteen minutes of "let me double-check that figure." That time adds up. That skepticism corrodes working relationships.
New analysts onboard faster because they can trace conclusions back to methodology. Instead of asking a senior colleague to explain how something was calculated, they can click through and see for themselves. The institutional knowledge is embedded in the artifacts, not locked in someone's head.
You don't need to change how you do analysis. Keep using pandas and matplotlib and seaborn and whatever else you're comfortable with. The change happens after the analysis is done.
Instead of the old path—notebook to screenshot to slides to email—try notebook to brief to link. The analysis stays connected to the communication. That's the whole idea.
It sounds obvious once you see it working, but most tools treat analysis and communication as completely separate problems. They're not. They're the same problem: getting your work in front of people who can act on it, without losing the connection to the work itself.
Margin keeps notebooks, reports, and datasets in one place. Try it free and see how connected reports work.
The Convergence: How AI and Data Science Are Becoming the Same Discipline
AI isn’t replacing data science. The two fields are converging into one workflow—and you’ll need to understand both.
Pandas Visualization That Actually Communicates
Stop making charts for yourself. Here's how to make charts that stakeholders understand.