Dear HR, Your Analytics Are Lying to You

Dear HR, Your Analytics Are Lying to You

Let’s pretend you are going to open a business that sells high-end denim jeans. They’re cotton that sells for $300 a pair, so it seems like a profitable, high-margin business. But you need to make a decision about what sizes of jeans to have produced. So you do a quick survey of a hundred likely shoppers and you discover that the average waist size of your target audience is 32-inches. As a savvy and data-minded business person, you place an order for 1,000 pairs of jeans, all with a 32-inch waist.

This, of course, sounds crazy. But it’s what you do every time you look at your web analytics. At the top of your report is probably a line telling you how many people made how many visits in the last week or month, how long they stayed and how many pages they saw on average. These are numbers aggregated and averaged out from all the visits. They are there to give you a sense of the volume of traffic overall, not to suggest that each customer is visiting 3.2 pages before they leave. (wearing a tee and a pair of 32-inch waist size jeans!)

These aggregates are worthless from a “valuable insights” perspective. But lots of people — possibly people you see in the mirror every morning, treat them like they mean everything.

For example, if I told you that your traffic doubled last week, you might be thrilled. Instead, you should be asking questions like “Did the number of conversions double, too?” or “Where was the extra traffic coming from?” These aggregate numbers lie because they aren’t really telling you anything useful. At best, they lead to interesting discussion and questions.

You aren’t hiring a crowd, you’re hiring an individual

A great way to get to those interesting questions is to stop thinking about all your traffic in the aggregate. In the same way that you’d separate your jeans customers into men and women, tall and short, and into thin, regular and thick, you need to think about all your web traffic as segments of a whole.

Let’s take an example. You know that 10% of all your traffic converts. That is, they apply for a job on your site. Now, it might be useful to break all your traffic into segments based on referral types. It turns out that all your traffic from Google organic search only has an 8% conversion rate, and your traffic from job boards has a 25% conversion rate. That’s some very basic but still interesting and helpful segmentation. You now know that traffic from the job boards is three times likelier to apply for a job as traffic from a regular search.

But you probably could have guessed that. People coming from job boards have already found an interesting job. The fact that they were on a job board at all means they are well motivated to act. This is helpful to know, in that it confirms knowledge you already possess.

Does this mean that the traffic from Google is bad traffic? No. The fact that it’s converting at a lower rate doesn’t mean that it’s worse; it means that it’s different. Let’s see what we can learn from it.

If you segment out all your traffic from Google, you might see that it has a slightly higher visit duration. Not a huge difference, maybe 10% longer visits. Now think about how people use Google. They aren’t coming from a job board where they already know they want to apply. In fact, Google is making it harder and harder to know what people searched for when they visited you. You can’t know if they were searching for a specific job you had an opening for, jobs in your area, or maybe just your company name. But we can guess that they have different intentions and needs from someone who came from Indeed, for example.

When I do a Google search, it’s pretty common for me to click on likely links just to see what’s behind them. Often, the link isn’t what I want, so I immediately hit my back button. I bet this is what a lot of people do, so maybe looking at all Google traffic in the aggregate isn’t very helpful. So, filter out all the bounce visits to see what Google’s value really is to your site. In this example, it’s going to make my conversion rate jump to 18%, which is great news. It’s also going to make my visit duration(s) jump as well. We go from 3.2 pages a visit to something closer to eight. Suddenly, we can see something: Traffic from job boards is really looking for the fastest way to apply. Traffic from Google is far more interested in learning more about us and our jobs before deciding to act.

When none of your size 32 jeans sold, it was because everyone you knew was either a size 28 or a size 36. Averaging them up not only gave you imperfect information, it actively lied to you. The data told you that your best customers were average, when in fact they were only either thin or… less thin.

Breaking them into segments did more than just help us see the numbers more accurately; they helped us understand the needs to the different groups. These groups weren’t just groups, they became personas: people who are looking to apply versus those looking to learn. With this information, we can start to think about what these different groups want and give it to them.

Giving people what they want and need based on their observed behaviors? Sounds like you are another step closer to becoming a huge fan of analytics.

Take a look at other ways in which analytics can help you make better talent acquisition decisions.

Related Articles

TMP Worldwide
Written by TMP Worldwide

Leave a Reply

Your email address will not be published. Required fields are marked *