Over the past few months, I’ve seen a number of fear-mongering pieces in the press about how the recession is driving tech layoffs and how tech employees (and engineers specifically) are losing their leverage as a result. Here are a few recent examples:
- The balance of power is shifting in the tech industry (CNN)
- The hot market for tech employees is ending, giving the power back to companies like Meta and Google to set and reduce pay (Business Insider)
- Snap’s Web3 team among 20% workforce reduction after first single-digit-growth quarter since going public (Fortune)
Articles like these drive engineers to speculate on Blind and on Reddit, in the internet equivalent of hushed whispers, about whether they’re next. Unfortunately, none of the pieces I’ve seen have bothered to draw a distinction between “tech” and “software engineering,” and many even use these terms interchangeably (like the CNN piece above).
The problem is that “tech” can mean anyone working at a tech company. You’re an engineer? Of course, you’re tech. You do ops? Great, you’re tech. You do marketing? You, too, are tech! These are all critical roles, and what I take umbrage with isn’t the decision to label non-engineers as tech employees. It’s deliberately misleading your audience by implying that “tech” refers to engineers specifically.
I don’t like imprecision, and I really don’t like fear-mongering. So, we at interviewing.io decided to dig into the data to see if engineers do indeed have a reason to fear. The TL;DR is that while, yes, tech layoffs have affected engineers somewhat, they are one of the least affected departments. While the existence of layoffs, in concert with eng hiring slowing down in general, does mean that engineers are losing some leverage in the market, it’s nowhere near as dire as the press makes it out to be.
The best source for layoff data
At some point in the last 2.5 years, you’ve probably visited layoffs.fyi. It was launched by Roger Lee in February 2020, just when concerns that this COVID-19 thing might affect the economy went mainstream. The site does exactly what it sounds like – it tracks layoffs at tech companies. Every time a company conducts a public round of layoffs, it gets added to a growing list. Each entry includes a layoff count, and a small subset of entries include a list of actual people who were let go, as well as some info about them (name, LinkedIn, geography, title, and so on). These layoff lists are usually in a Google doc but sometimes in something more fancy, like this dedicated website for Shopify layoffs.1
As far as I know, even with its limitations, layoffs.fyi is the best source for layoff data, and because the individual layoff lists actually break out layoffs by function, it’s the natural place to go if you want to disambiguate “tech” into specific departments.
So that’s where we looked. Before we talk about what we did, I’d like to call out one big limitation of the layoff lists on layoffs.fyi.
One big limitation of the data: layoff lists are opt-in
From what we can tell, the layoff lists on layoffs.fyi are opt-in. In other words, former employees decide if they want to be included. Sometimes the difference between the layoff count reported in the press and the length of the list is substantial. Peloton, for instance, had 2800 layoffs, but their list only had 400 people.
If job losers’ willingness to opt in depends on their role, they might be overrepresented in the layoff data. Recruiters, for instance, might be more willing to sign on since they know the value of getting their names out there. Our fix is to calculate opt-in rates for each role and use these as weights in the analysis – just like when the Census upweights groups who are less likely to respond to a survey. For more detail on how we did that, please see the section called “Appendix: Methods” below.
The questions we wanted to answer
To disambiguate “tech layoffs” into actual departments and break out how each department was affected, we needed to answer the following questions:
- What % of employees got laid off overall?
- What’s the breakdown of layoffs by department?
- How hard was each department hit?
In the next section, you can see what we learned. If you’re curious how we got to those results, see the section called “Appendix: Methods” at the bottom of the post.
The results, or layoffs by the numbers
What % of employees got laid off overall?
Based on the companies we looked at, at companies where layoffs happened, about 19% of employees were laid off.
What’s the breakdown of layoffs by department?
Below, you can see the % of total layoffs that each department constituted. In other words, of the layoffs that happened in 2022, 20% were salespeople, 5% were engineers, 3% were finance, and so on 2… and that a third of newly unemployed tech workers come from sales and product.
How hard was each department hit?
The graph above doesn’t give the risk of getting laid off, however. All departments don’t have the same headcount. For instance, according to LinkedIn data, engineers make up about 20% of a company’s total headcount on average, whereas recruiters only make up 2%, about 10X less. In other words, if each department were to be hit equally, we’d expect engineers to be laid off about 10X more often than recruiters. That’s clearly not what’s going on.
To get the full picture, we need to correct for department size and see how hard each department was actually hit, which you can see in the graph below.
These results show exactly why using the label “tech” isn’t good enough – there is a big disparity between different departments. According to our data, almost half of HR people and recruiters got laid off, as compared to 10% of engineers and only 4% of salespeople. The salespeople percentage shows the need to correct for department size, given that salespeople constituted the biggest % of overall layoffs.
What does this mean for engineers?
Although engineers didn’t emerge from this year’s layoffs unscathed, our data shows that lumping everyone together under the “tech” umbrella is misleading. These insights don’t change the fact that the people who lost their jobs, regardless of department, are hurting, and, unfortunately, layoffs are likely far from over.
That said, HR teams and recruiters were hit the hardest by far. Sadly these teams are often the proverbial canary in the coal mine – when a recession hits, hiring, and especially outbound hiring, is the first to get cut. Engineering, marketing, product, and design were all affected somewhat, and per capita, sales was affected least.
So what does this all mean for engineers? In our minds, the kind of anxiety produced by the news articles I referenced in the beginning is unjustified. Moreover, though hiring has clearly slowed down in the past quarter, after surveying our users, we saw that a healthy amount of companies are still hiring engineers.3
However, the reality is that engineers are losing some of the leverage they had previously enjoyed. Given this climate, it’s reasonable to expect that more engineers will be competing for fewer open positions than before, which means that:
- Candidate experience is going to degrade somewhat. Why? Sadly, companies caring about candidate experience are not motivated by doing the right thing. Rather, it’s largely a function of how much leverage labor has in the market. As such, we’re likely to see…
- The return of homework assignments and coding quizzes early on in the process, even for senior candidates. When good candidates don’t drop out of your process despite having to jump through hoops, the hoops, which are cheaper than interviewing with a human, will return.
- Interview questions will degrade in quality – when selling in an interview process is less important, companies will put less effort into designing bespoke questions that give candidates a preview of what it’s like to work there and will fall back to doing the easy thing, which is probably leaning more on cookie-cutter Leetcode questions.
- The bar will go up. Because more candidates are competing for fewer positions, companies will be able to get more picky while still hitting their hiring goals.
interviewing.io is both a mock interview platform and an eng hiring marketplace – engineers use us for technical interview practice, and top performers get fast-tracked at companies. Companies actually interview our top performers anonymously, right on our platform, and leave feedback after each interview. If the candidate passes, they unmask and move to the next step (typically an onsite). Feedback is both quantitative and qualitative, and in addition to telling us if the candidate passed, companies also rate them on technical ability, communication ability, and problem solving ability. Technical ability tends to be most predictive and weighed the most heavily.
As such, to check our read on the market, we averaged the technical scores for successful interviews over the last few quarters to see where the bar has been and where it is now. The results are… compelling.
Below is the list of all companies (either US-based or with a significant number of US employees) with layoff lists, through 7/26/22 with Snap (8/31/22) added on. Sadly, there have been more layoffs in the last few months, but at some point we made the call to stop adding new ones so we could actually get this post published. That said, we made the call to include Snap (8/31/22) as a late addition because they had enough layoffs to potentially change the results. Based on layoffs.fyi data, these companies comprise about 15% of tech companies who employed people in the U.S. and had at least one round of layoffs in 2022.4
|7/14||The Mom Project|
For each of the companies above, we did the following to get layoff counts by department:
- Used LinkedIn to get the total number of U.S. employees at each company.5 In cases where a company had both U.S. and non-U.S. employees (e.g. Shopify), we looked at what portion of total employees were located in the U.S. and then adjusted the layoff count proportionately.
- Went through each company’s layoff list and tagged all employees with their department. We included the following departments: Software Engineering, Data Science & Analysis, Product & Design, Marketing, Sales/Account Management/Customer Success, Recruiting, HR (Non-Recruiting), Finance, and Operations.
- Because layoff lists are opt-in (see the section above called “One big limitation of the data: layoff lists are opt-in” for more detail), we needed another source of truth, and LinkedIn seemed like the best bet. As such, for each department:
- For a set of representative companies, we searched LinkedIn for the list of people who had left the company in 2022.6 We chose a subset of companies that included public companies, large startups, and small startups.7
- We compared that count to the count for that department from (2) and used the difference for each department as a multiplier on the number of people in (2). When we did this, we ended up with the numbers below – each bar is our calculated probability that someone from this department would actually opt in to be included on a layoff list.
Once we had layoff counts by department, we also needed to know how many people worked in each department so we could figure out how hard each department was hit. For instance, Stitch Fix has about 4500 employees in the U.S. Of those, 238 are engineers. Therefore, at Stitch Fix, engineers comprise about 5% of employees. To do that, we used the same representative subset of companies as above and cross-referenced LinkedIn to see how many people worked in each department and what % of total headcount that department constituted. After we got all the numbers for the engineering department at our slice of companies, we saw that, on average, engineers constitute about 11% of employees. We repeated this process for each department. This is important information – without knowing the relative size of each department, we wouldn’t be able to reason about how each department was affected.
Thank you to Sam Jordan, Liz Graves, Santiago Munoz, the broader interviewing.io team, David Holtz, and Maxim Massenkoff for their help with this piece.
- Before we get into our methods, a quick word about geography. Please note that this analysis is confined to layoffs in the United States (though we did include companies who weren’t based in the U.S. but had a significant number of U.S. employees, like Shopify). Why? It’s the market we know best at interviewing.io, and we want to limit our analysis to what we understand and also what materially affects our business and the vast majority of our users. We also understand that the findings in this post may not necessarily generalize to other geographies. That said, I want to explicitly call out that just because other countries aren’t our target market doesn’t mean that those layoffs aren’t important or devastating to those affected.
- You might wonder why the numbers above don’t add up to 100 – it’s because our list of departments isn’t exhaustive, and some departments that didn’t consistently come up between companies were omitted in this analysis (e.g., Project Management, IT, Quality Assurance).
- You might ask how this is possible when so many recruiters are getting laid off. The reality is that in this market, more and more good candidates will be inbound, i.e. applying to or getting referred into companies, rather than having to be sourced.
- Given that only 15% of companies actually had layoff lists, the employees who were laid off at these companies may not be representative of the greater set of layoffs.
- We used both LinkedIn Insights and Sales Navigator and cross-checked them against each other – is unfortunately somewhat vague. Job titles aren’t categorized consistently, location data isn’t always available, and people don’t always immediately update their LinkedIn when they get laid off. We assumed that by the time we did this analysis, as several months had elapsed, that most people had indeed updated their LinkedIns. Unfortunately, it’s not possible to search LinkedIn for a snapshot of what headcount looked like for a certain date range.
- One big limitation of this approach is that it’s impossible to distinguish on LinkedIn people who left voluntarily versus those who were laid off. To correct for that, we looked at the average tenure for each department and removed the expected number of people leaving voluntarily. Another limitation is that 2022 is too broad, as some layoffs happened halfway into the year. Not much we could do about either of these, unfortunately.
- The companies we used were Netflix, MasterClass, Stitch Fix, Robinhood, Replicated, Tray.io, Snap, and Policygenius. We tried to get a representative sample that included a range of company sizes and stages.