In today’s digital world, it can be hard to sort through the various statistics the internet throws at its users. As a data analyst at Appcast, a large part of my job is picking out the signal within the data noise of programmatic job advertising. I’ve compiled a list of some of my best practices, and I’ve included some tips that can hopefully help you navigate through the noise, too.
1. Try not to arrive to a project with preconceived notions.
For me, this point is probably the hardest to execute. Whenever I ask for (or provide) data, it is usually to support or disprove a hypothesis. However, I have learned it is important to not arrive at the conclusion of an analysis before it has even started.
One solid example of this happened to me in the middle of March as I was preparing industry-level cost metrics for a client. At that time, effects of COVID-19 were slowly improving. A vaccine was on the horizon, and a summer minimally impacted by a pandemic was still in play. My message to the client was “stick this market out for just a few more months and hiring can soon return to normal(ish)”. Then I started looking into the data. Conversion rates had plummeted during the past week, and I began to investigate why. At that time, Congress was just passing the 2021 American Rescue Plan, and active job seekers had already begun to react to the extended unemployment benefits and stimulus checks. The gap between supply and demand was quickly growing, and people were noticing. Fast forward a couple weeks, our CEO Chris Forman hosted an informative webinar and Appcast was able to provide supporting documentation about the state of the market. Having a preconceived idea of what to look for had almost caused me to miss this huge change in market behavior.
2. Use a variety of sources.
Here is one of my favorite math quotes of all time: “There are three kinds of lies: lies, damned lies, and statistics” – Benjamin Disraeli
Now I don’t love this one just because of the foul language; I think it has an intriguing message too. There is no doubt that statistics can be manipulated in a variety of ways. In other words, context is everything. In my experience, I have learned that the best way to navigate around this bias is to use multiple sources. Make them noticeably different, too. There is so much more power in using data that has been studied, critiqued, and upheld. What I’m saying is please don’t just take our word for things. Next time you’re on one of our webinars hosted by Andrew Flowers, check his data against the Bureau of Labor Statistics or FRED (Federal Reserve Economic Data). Not only will Andrew love to see a fellow labor economist, but I’m hoping you can ask some insightful questions that the whole audience can benefit from.
3. Translate numbers into actionable items.
Data is only as powerful as the person wielding it. Fractions, percentages, decimals all look really nice on paper, but they are meaningless if you can’t turn them into tangible actions. One example I’m going to use here is our 2021 Recruitment Marketing Benchmark Report. I find the whole report fascinating, but I want to call your attention to page 25, which shows the percentages of applies by day of week. A passive user might look at this graph and notice that the most applies are occurring on Monday and Tuesday. What I’m asking you to do is use this in your favor. If we know that job seekers prefer the beginning of the week to apply, post and refresh your jobs then! Small actions like this can go a long way in meeting goals.
While a lot of these points might seem like huge ideas, practice can help to make them attainable. So, the next time you open our annual Benchmark Report, or listen to a webinar, or even just sit down with a data project, I hope you have Benjamin Disraeli’s voice echoing throughout your head.
