Yusuf Tiamiyu ← All projects
Data cleaning and analysis

Tech layoffs, 2020 to 2023

I cleaned a public list of 2,361 tech layoff announcements and used it to answer five practical questions.

2,361layoff announcements cleaned and checked
7×more likely to close for early-stage startups
3×the size of the 2020 wave, in 2022 to 2023

The problem

Public datasets are messy. This one had duplicate rows, the same company spelled in different ways, missing industries, and dates stored as text.

Analysis on uncleaned data gives wrong answers, so the cleaning had to be careful, and every change had to be recorded so it could be checked.

What I did

What I found

The second wave was bigger

People laid off per month, with a much larger second wave from late 2022

From September 2022 to March 2023, 212,744 people were laid off, three times as many as in the 2020 wave. January 2023 alone was larger than all of 2021.

Young startups were most at risk

Share of layoffs that were full shutdowns, by funding stage

Startups at the earliest funding stages were about seven times more likely to shut down completely than larger, later-stage companies.

Big totals can mislead

Retail lost the second-most jobs, but only 4% of its workforce. Travel, crypto and fitness companies cut more than 20% of theirs. A big total often just means a big company.

Nigeria

Nigeria had 15 recorded layoff events, 12 of them between September 2022 and March 2023. Jumia accounted for almost half of the reported total.

What it means

Careful cleaning changes the answers. Matching companies by name alone would have merged three different companies called Bolt and given them the wrong industry.

For technical readers

MySQL (window functions, CTEs, data-quality checks), Python for the charts