Science Communication at Work

01

The last mile

Data work has a "last mile" problem. You can do everything upstream perfectly — clean the data, fit the right model, evaluate it honestly — and still fail completely at the final step of getting a busy, non-technical person to understand and believe it enough to change what they do. That last mile is where most analytical value leaks away.

The mental shift is to treat communication not as a write-up you bolt on at the end, but as part of the work itself — something you design for from the first question. The goal of a data presentation is not to show what you did; it's to change a decision. Hold that, and every choice below follows.

02

Start with the audience

The first rule is the one analysts break most: it's not about you or your work — it's about them and their decision. Before a single slide, ask who is in the room, what decision they're making, what they already know, and what they care about. An executive, a fellow analyst, and a minister's office need three completely different versions of the same finding.

For senior decision-makers specifically, the rule is less, not more. They don't want every data point from your analysis — they want the three or four findings that bear directly on the choice in front of them, with the implications spelled out. The technical depth you're proud of belongs in a backup appendix or a linked dashboard, available if they ask, invisible if they don't.

03

Lead with the recommendation

Academic training teaches you to build up to a conclusion: method, then results, then finally the answer. In a workplace this is exactly backwards. Lead with the recommendation, not the methodology. Decision-makers need to know what to do and why, up front — they'll ask for the details if they want them.

04

The narrative arc

Humans are wired for stories, not spreadsheets — a narrative is remembered and acted on where a table is forgotten. The most reliable structure for a data story is situation → complication → resolution:

• Situation — the shared context everyone agrees on. "Here's where things stand."
• Complication — the tension: the problem, change, or risk the data reveals. "But here's what's happening."
• Resolution — your recommendation: what to do about it. "So we should…"

This arc creates a small amount of tension and then resolves it, which is what holds attention and motivates action. It turns a pile of charts into a story with a beginning, middle, and end — and a point.

05

Answer the 'so what'

The most useful question to ask of every chart, number, and slide is brutally simple: so what? Every metric you show must connect to an implication. If you report that response times rose 8%, the very next sentence has to say why that matters and what should change because of it — otherwise you've handed the audience homework, and they won't do it.

This is the difference between reporting and communicating. A report states facts; communication states what the facts mean for the decision. Relentlessly converting "here's a number" into "here's what this number means for you" is the habit that makes analysis land.

06

Visuals that decide

A good chart shows an obvious pattern that needs no explanation; a bad one makes the audience do work. The guiding principle is clarity over flair: pick the visual for the decision being made, not for how impressive it looks.

• Eliminate clutter — strip out anything that isn't carrying meaning (chart junk, redundant grid-lines, decorative 3-D).
• Direct attention — use colour, labels, and annotations sparingly, and only to highlight the one thing you want them to see. If everything is emphasised, nothing is.
• Don't mislead — truncated axes, dual axes, and the wrong chart type can distort the story; accuracy is part of the message.

One chart that makes the point cleanly beats a dashboard of twelve that bury it. The full toolkit is the subject of the dashboards page; the principle here is that a visual is an argument, and a cluttered argument fails.

07

Earning trust

A finding is only as persuasive as it is credible, and credibility is fragile. Before you present, the numbers have to be right — a single error someone in the room can spot will sink the entire analysis, however sound the rest of it is. So sanity-check everything, and reconcile your figures against whatever sources the audience already trusts.

Trust also comes from honesty about uncertainty. Stating the limitations and the confidence in your result — the discipline from the statistics page — builds credibility rather than undermining it. Decision-makers are wary of analysts who sound too certain; being straight about what you don't know is what makes them believe what you do.

08

Driving to action

The presentation isn't the finish line — a decision is. The final discipline is to make sure the conversation ends in execution, not just agreement. That means closing with a clear, specific recommendation, and where you can, naming what happens next: who does what, by when. A discussion everyone nods at and nobody acts on was a failure of communication, not of analysis.

09

Where it shows up in my work

10

Refresh in 60 seconds

Practical guidance on this page draws on current industry writing about data storytelling for decision-makers (ThoughtSpot, ClicData, and others), alongside the UniMelb subject.