However, it avoids a perception problem which may be far worse: the hypothetical perception that the people drawing up the layoffs were racist and that they disproportionately affected minorities. They wrote "Layoffs like this can have a more pronounced impact on marginalized communities, so we were particularly focused.." on making sure they did not.
If they ignored race entirely, and just laid off people at random, there's a chance that someone later would say "Twilio employees are 35% female, 33% Asian, 9% Latino, and 6% Black, but the laid-off employees were 11% Latino! Racism!" and turn a mob against them. Once the media/Twitter/mob believe that Twilio HR and Twilio executives are racist, those execs might try to defend themselves by saying: "We weighted based on performance reviews, added a bit of statistical noise, and then did a random selection. We believe those performance reviews to be objective and free of racial bias. We didn't even realize that Latinos were slightly over-represented in the layoffs until you mentioned it, but it looks to be merely statistical chance. We actually fired slightly more white people and more men than the statistics would suggest, we're not bigots."
...Do you think that would work?
It's far easier to defend against the accusation of being overly woke than it is to defend against the accusation of being racist, so while you'd rather be neither, it's better to overtly and publicly fail in the safer direction.