James Chen

Beautiful Data Science Presentations

An introduction to making visually attractive PowerPoint slides.


The intended audience of this short blog post is someone who is interested in putting together presentations for 3 different purposes: sales kit, process flow, and analytics report. Of course, the key idea of this post is not limited to data science projects only, hence someone coming from outside of the field may find it useful as well.

We will be demonstrating the idea with a Data-as-a-Service project, where the input is a large collection of consumer surveys and output is a handful of personas that describe our target audience.

Business Proposal

The purpose of the business proposal, or sales kit, is to attract potential customers for our data science consulting service. They may be coming from different backgrounds and of different staff levels inside the marketing function of brands or marketing agencies. As a result, we would want to convey as our idea as simple as possible, through attractive visual designs.

We came across this album cover page by ODESZA (please be noted that this is for demo and educational purposes only, the album is copyrighted so we would need the owner’s permission to modify it):

Let us make it our primary visual by adding our service name to the picture, and modify it as our cover page.

Then we want to capture their attention by identifying possible job roles. They could be strategy planners, media purchasers, or creative designers. It is also suggested to highlight keywords with a different color.

Next, we will give them a reason to read further, by identifying problems that may pop-up in their daily jobs. Note that we can Google keywords such as “resource icon png white” to obtain small icons that fit into our points. We can also use the PowerPoint built-in function Arrange to align the spaces between words and icons to make the slides prettier while showing professionalism such as attention to details.

We need to tell the audience what our solution can deliver, in greater details. Again icons help us associate ideas better, and could also save audience time to understand our services.

The following part we need to explain a little more on how the service actually works, by identifying 4 key elements in our service. Note that it is also recommended to find icons of similar style. For example, the icons below all have thinner outlines. Although this could be time consuming, unless we want to create our own icons instead.

Finally, we need to give out a clear timeline on the execution plan, or our audience may find our service unrealistic. In additional, key milestones may give a better picture on how this service could actually be executed.

This summarizes the most important parts in a data science business proposal, and may generate significant interests for our audience so we can move on to the next stage.

Process Flow

The audience of this stage is more technical-savvy engineers or scientists. While they may be working with flow charts everyday, it is also worth mentioning that beautiful and neat icons can help their less technical audience understand complicated concepts better. A good resource we would like to recommend is McKinsey Insights, in which we can use their carefully designed exhibits as our presentation templates, such as this one:

Now, we can use this template and take relevant icons to show the process of our own service. The Eyedropper tool in PowerPoint is excellent in matching colors on our slides with inserted pictures or other objects, as shown below.

Once we get a better understanding of how our service works, we can actually build a PoC with it and come up with a demo report.

Analytics Report

This may be the most technical part of our presentations, as it has a lot to do with presenting the acutal results from our data science project. The key takeaway for the audience here is what the final output would look like. The audience here may be project stakeholders such as marketing head or brand lead in agencies. We will want to show them the input of our survey data as well as the output of our personas.

Open to project-based work.