LinkedIn ad targeting data
It’s quite simple to pull data on the female and male employee estimates for companies.
First, we created a draft text ad to use as a placeholder to get to the audience targeting tool. Targeting data is available without purchasing or publishing an ad.
For each company, we found the estimated number of the:
- Total company workforce on LinkedIn.
- Female employees of each company with LinkedIn profiles.
- Male employees of each company with LinkedIn profiles.
Data for this study was collected Feb. 28-March 1, 2017.
A massive data set
LinkedIn ad targeting gave us a much larger sample than used in most corporate research we’ve come across.
For the 50 companies in the study, LinkedIn ad targeting could determine gender for roughly 1.839 million employees. (We arrived at this number by adding together the total male and female estimates for all 50 companies.)
First name inference
LinkedIn states that gender is “inferred in English from first name of member.”
This practice is not unique to LinkedIn. While companies use a variety of methods to identify gender at scale, first name determination is a common one. Solutions such as Genderize.io, GenderPredictor, SexMachine and Beauvoir rely on first names to identify gender.
Initially, we were concerned that there could be a potential bias toward assigning gender to predominantly English names. But when we checked our results against recent diversity reports, the targeting data was pretty close to the reported gender ratios. For this reason, we did not investigate English name bias further for this study.
Finding the data
Though LinkedIn gives several options for ad audience targeting, we used current company and gender.
LinkedIn gives audience estimates for total workforce, male employees, and female employees rounded to the nearest thousand.
For example, the target audience of female Facebook employees on LinkedIn was estimated at “5,000+”. We used the number 5,000 as the female employee estimate for Facebook in our study.
Companies in our study had to have 1000+ employees
LinkedIn gives the message “your audience is too narrow” when it estimates an audience to be less than a thousand. That is why we only included companies with thousands of employees in this study.
Gender unknown profiles and ratios
We also considered how the total male and female estimates compared to the total company estimates on LinkedIn.
Of the 50 companies we looked at, male and female estimates represented an average of 82 percent of the total company estimates on LinkedIn. Meaning, LinkedIn could not determine the gender of an average of 18 percent of the company’s employees with LinkedIn profiles.
Since female employee estimates were drastically lower than male estimates for most of the companies we looked at, it was unlikely that the names LinkedIn could not assign a gender to would significantly alter the male/female ratio in all but a few instances. For example, Concur and SAS.
Of the 50 tech companies in our study, Concur and SAS had the highest estimated percentage of female employees in their workforce.
Concur was the only company to have equal male and female employees estimates, though LinkedIn was unable to identify a gender for 20 percent of the Concur employees with LinkedIn profiles.
Women are still a minority in major tech companies
Here are some of our findings:
- On average, women represented 24 percent of their workforces. Men represented 58 percent of their workforces on average. An average of 18 percent of the workforce was gender unknown.
- Women made up only 20-30 percent of the workforce for most of the companies we studied.
- Not one of the companies we studied had 50 percent or higher estimates for women in the workforce.
Gender unknown LinkedIn profiles could shift the male/female ratios. However, in most cases, it is unlikely the change would be drastic unless all the unassigned profiles were women.