Data Science storytelling or data science presentation is one of the most underrated skills a data scientist can have. It is barely touched in most data science courses and programs. The importance of describing a compelling story about the results of an analysis is critical for a data scientist or data analyst. In fact, as a data scientist, more often than not, you will be working with people that do not know data. Consequently, you should be able to convey your story in a manner that is adequate to whoever is listening.
In this article, I will not go through any type of technical way of solving a data-related problem. However, I will try to review several data storytelling habits that would help you better present your analysis.
What is the problem of data presentations (storytelling) in Data Science?
Talking to the client is not the easiest of things. In fact, most of my data scientist colleagues do not particularly like talking to clients. It is, believe it or not, somewhat of a meme.
The result of someone who does not know how to present is a boring presentation filled with numbers. The result of having a boring presentation is a potential loss of clients. Indeed, if a client did not understand what you presenting in a report or during a presentation, he may feel that he did not derive enough insights to make a decision. Consequently, it may render your analysis useless even though you had some valuable insights
The importance of narrating a compelling story is critical to the success of the analysis work you did. Indeed, it is essential for the person you are communicating with to understand the main points of your analysis. Although you can impress them with your knowledge through your complicated algorithms, the results of all said algorithms are not always what should be reported to answer a person’s business question.
What are the goals of data analysis storytelling?
When you prepare your report or presentation, there should be 3 main goals you should attempt to achieve:
- Answer the business questions/ requirements presented at the beginning of a project.
- Illustrate knowledge, information, and insights derived from the analyzed data.
- Narrate a story in a simple way that contains valuable and relevant result-focussed recommendations.
Keeping those goals in mind during your data analysis presentation will allow you to drive value by delivering and presenting a compelling analysis. In fact, how you deliver a message extracted from the data can even be more important than what the data says.
How do you ensure a positive data science presentation?
Understand that if you go beyond what is common knowledge during the report/presentation, you present a complicated concept, or even if you show negative business performances, your presentation is likely to be disputed. We are all humans after all, we like to be correct even if we are wrong. People need to UNDERSTAND WHAT YOU ARE SAYING. I cannot emphasize that enough. Therefore, it is your job as the data analyst/data scientist to present the data in the most humanistic way possible. Indeed, you have to bring a context to your analysis and understand how your results affect human emotions, motivation, or organizational behavior.
It is not advisable to blast people with numbers, 30 pages reports, and 100 pages full of irrelevant charts and graphs. Do not make the mistakes of presenting just data visualizations. After 10/12 slides, people will stop listening. Therefore, you have to construct a narrative through the results of your analysis, or even better, tell a story. Tell your clients what they should remember from your analysis and how they should use your insights to make decisions. You have to create a piece that is so valuable that they come back to it, just like your favorite book.
How do you deliver a data science presentation/ data analysis report?
In this section, we are going to go through the process you should take to deliver a compelling data analysis presentation (data science presentation). You have to understand that every data analysis report will be different. However, you can follow these simple guidelines to create a compelling story from your data analysis. Consider applying the following guidelines in your next report.
1- Always answer WHY your client should care about the data/insight you are reporting
Before creating the report/presentation, ask the question WHY SHOULD THEY CARE? Indeed, knowing the why is primordial to telling a compelling story. So you have to identify why the story or section you are presenting has occurred and why it is important to present it. If you cannot answer the why then consider scrapping said part from your data science presentation.
Remember, business people are busy people and they always need context when reporting your data analysis results. You should give them a reason as to why they should care about every section you are presenting. If you can’t answer the why then that insight is probably a filler and should be removed.
2- Present challenges and bring solutions
Introduce the challenges you are providing insights for and explain the costs of not taking action. Furthermore, you have to frame the business issue that the analysis or part of the analysis solves and suggest recommendations on how it may be fixed or improved if possible. This will eliminate any confusions that may arise.
3- Explain the general data settings.
For your interlocutor to get a sense of the data set, do not forget to mention errors, date range, omissions… You always have to let people know in advance what you are talking about. For example, you can say something like “the conclusions are based on the year X. We did not consider any value above Y. ” This way, you shareholder knows under what circumstances you draw your conclusions. Again, for this step, keep it general as mentioned in the previous point. Your client does not need to know what Python function you used to omit certain values. Unless he/she asks, of course.
4- Humanize the data.
Try to use examples and scenarios to give context and humanize the data. Especially if you are explaining complicated concepts. Indeed, creating aliases, settings and abstractions allow you to not only simplify the problem you are presenting. But as well to reduce the risk of offending your stakeholder if you find any negative performance in the data.
5- Emphasise important events.
Do not be afraid to stop and emphasize the important events to help you demonstrate your data science presentation. For example, if you used external data that had an important impact on the analysis/results, do let your stakeholder know how that piece of information affects them. Consequently, emphasizing on important events goes a long way to help clarify and simplify your analysis.
6- Speak in simple terms.
You are not Shakespeare. Do not need to impress anybody with marvelous and complicated vocabulary. Keep your words simple. Speak to help people understand. Do not confuse them because certain jargon can be extremely confusing. I did not take a class in college because it contained “Stochastic Processes“. Why didn’t they simply use “Random Processes”? So, remember, try to make your presentation and writings as simple as possible.
7- Use Pictures.
One of the most famous sentences about pictures. A picture is worth a thousand words. Use pictures to illustrate your points. There are multiple types of graphs that you can use in various situations. They vary from bar charts, histograms, trendlines, scatter plots, and many more. The only thing to be careful, do not create too many useless and repetitive charts. I have written a guide on charting that can help you with creating and illustrating your results through charts.
8- Share your thoughts and actions for a successful or improved project.
Do not be afraid to tell your shareholders what is required for a project to be successful or a result of analysis to be usefull. Use clear verbose and descriptive nouns. For instance, if you need data from somewhere, tell them, “To get this level of accuracy, we need daily access to Xs data through an API“.
9- Communicate the cost of not taking an action
Tell them the negative aspects of not taking any actions towards the implementation of a solution. Indeed, you have to state how important a piece of analysis is, and contrast the financial repercussions of doing nothing against doing something. For example, say that you find a relation between the website speed and the page load time. You found that a series of images is causing them. What you can do here, you can be like “leaving the page like this make you lose x$ amount of money, whereas hiring a web developer to improve the page can cost you y$. But in the long run, I am t% confident that making the changes will help you gain x$ after z year“. This sentence can obviously apply in many scenarios
10- Share your Recommendations during your data science presentation.
Always finish with a series of recommendations that can bring value to the person you are presenting to. Remember you are generating value, ergo since you have worked on a project for so long, you are the most suited to give recommendations on potential improvement. Do not be scared if you are not the most knowledgeable or experienced person in the room.
More often than not, you will be working with people with more experience than you in a particular field. However, the theory learned in your training can be applied to any type of data, and you should be able to give them insights that can help said-experts, to make better decisions. Do not forget the previous point. Your recommendations should be made clear, straightforward, and backed with data. You should be able to give a logical answer to any questions regarding something you recommend.
To wrap up.
These are the things to keep in mind to improve your data science presentation. Data science presentations, or data science storytelling, are not easy, especially for people that are not used to it. So, next time you have a presentation, keep the above recommendation in mind and, do not be afraid to try them. They should make your presentation more compelling, and you will be able to get your point across.
If you are scared that your presentations or report is too long or you feel like your shareholders are not getting enough value from your presentation, try giving some of these techniques a try next time you present, and let me know how it went. Do you have any additional tips on how to help people with their data presentation, share them in the comment section down below.
If you made this far in the article, thank you very much.
I hope this information was of use to you.
Feel free to use any information from this page. I’d appreciate it if you can simply link to this article as the source. If you have any additional questions, you can reach out to malick@malicksarr.com or message me on Twitter. If you want more content like this, join my email list to receive the latest articles. I promise I do not spam.
If you liked this article, maybe you will like these too.
Are Data Science Jobs in Demand?
Why is Machine Learning important? [in 2021]
Why Data Science is Important?
How machine learning is used in Cybersecurity? [in 2021]