Stay updated with data visualization world - Sign up for my weekly newsletter (in Hebrew): 

Numbers Stories

My role: Storytelling, Infographics & Content Design

In recent years I have been working with Geektime's research team as an inforgraphic designer. I designed yearly reports covering the startup world, and  special reports on disrupting industries like Blockchain, insuretec and smart mobility.


Designing for these reports always starts with an Excel table and ends with an infographic that tells a story. Working hand-in-hand with the analysts, the process involves focusing on a compelling narrative, narrowing down the data, and choosing a structure that best describes the content.


I summarized some tips on how to do this process well in a piece I wrote for Geektime (follow the link to read).


Here I chose a few of the projects I worked on, the challenges they presented and the solutions I found.

1# Focusing on your narrative

Korean Blockchain Companies by Registration Country (ICO launches)

Korean Blockchain Companies by Registration Country (ICO launches)


South Korea is a hotspot of Blockchain and digital currency adoption. This is thanks to a unique combination of factors, including advanced digital culture, availability of superfast broadband and high unemployment rates among young people.


I was asked to visualize data about the countries where Korea-based companies launch their currencies, in a process called Initial Coin Offering (ICO).

The Original Data

Trying to figure out the best way to present this information, I jotted a few options. I presented them to the analyst, and we discussed what message each of them conveys.

Pie Chart

Focus: the majority of the companies launched in Korea itself. Pie charts are not effective if there are too many slices, so all the less important ICO locations are grouped together.

Columns Chart

Focus: distribution of ICO locations around the world. Korean companies launch in a large number of countries worldwide.

Seeing those options, the analyst and I realised the interesting story here is how Korean companies look to a large numbers of locations for ICOs. This is partly due to legal complications around crypto in Korea.

To further emphasise this outstanding situation, I ultimately chose to visualize the data using a map.

The Final Version - A Map 

Had I had more time, I would have presented a comparison to the distribution of ICO locations for companies based in other countries like the United States and Israel. This would have emphasised how outstanding the Korean situation is.

2# Squeezing data - When 2 become 1

Geektime's 2018 in Startups report covered the record breaking amounts of money raised by startups that year, including 17 companies who raised over a billion dollars each.

The report material included a list of those 17 companies, detailing where each was based and the sum of money it raised. It also included a table summarising the number of funding rounds and the amount raised per country.

Sometimes visualization is not about graphics: in this case the per country table needed a little simplification, emphasising the ranking of the countries by number of companies funded and amount raised.


The main question here was which of the two tables - which essential repeated the same data - we wanted to include and visualize. I asked the report author. The answer was both. The full list of companies was important, but the report also wanted to give a sense of where they came from.


I tried to sketch various options when the aha moment came: it would be best to combine the two table into one infographic. Not only would it be space saving and visually compelling, but also structure the data into one cohesive story.

The Result - 2 tables become one cohesive story

3# Making smart visual decisions

When an infographic has to present complex data, or multiple datasets, designing it involves many important visual decisions.


This challenge surfaced when I designed one of the infographics in Geektime's 2018 report, based on the data in the following table.

Israeli startup Investments 2018 - Table

Challenge 1:

Representing numbers in different orders of magnitude on the same page.

The table gave data on number of funding rounds and total funds raised. These are related numbers, but one is in millions of dollars and the other is in small numbers. Can I present them both in the same scale? Or should I use two similar charts with different scales? Or Is there a different solution?


There are two separate parameters here -  the number of funding rounds and the total money raised. Their values are very different. Using the same scale is very problematic, as the smaller bars virtually impossible to discern. Using different scales in confusing, as the bar sizes are much more visually obvious than labels.


The key point to solving this question is which comparison we wanted the readers to do, and which comparisons did not make sense. We wanted the reader to focus on comparing each parameter over the years and between the industries. The question of how the number of rounds is related to the total amount was less interesting.


I solved this by using completely different visual representation. Data that should to be compared is grouped together and visualised in the same way. Data that shouldn't be compared is separated both by location on the page and visual representation.


Challenge 2:

Two datasets with similar parameters, presented next to each other.

On the same page with the investment data, we also wanted to present mergers & acquisitions. Both investments and M&As are measure in the same parameters - number of instances, and sums in millions of dollars. It made sense to use the same structure for both, but I still needed them to be visually distringuished.  

Presenting two datasets with similar parameters - first draft

The solution here is color. Using different color to each dataset made it very clear that they are indeed two separate things.

Final design

You can also read some of the full reports here ...

Or read some tips I've summarized on how to do this process well in a piece I wrote for Geektime (follow the link to read).

2017 Geektime Annual Report2