RecBuzz remains an essential gathering for professionals in the tech recruitment field. It fosters discussions on important topics like performance-based pricing, workforce shortages, and the impact of AI integration.
Yana Levchenko, Country Manager for France at Jooble, joined the RecBuzz conference in Barcelona on April 16-17. She discussed Jooble’s data usage experience, emphasising its benefits and potential pitfalls if misunderstood.
Not a success story?
“I will tell you a short story about how Jooble–a proudly data-driven company–tricked itself with its own data. I should say that this isn’t a success story yet,” – Yana said. “At conferences like this, it’s much more important to raise questions and share thoughts with industry peers than to provide immediate answers.”
Yana encouraged the audience to picture themselves searching for a job. They often come across many job ads, all showing the job title. Some might also mention the salary and a short job description. The main thing is, there are lots of similar job ads on every site.
She asked, “Have you ever thought about how you choose which job to apply for when you’re scanning through the search results on every job site?”
At Jooble, we found that each job posting typically catches a job seeker’s attention for only 3 to 10 seconds. But is this enough time to decide if a job is right for you? We don’t believe so. Plus, we’ve noticed that almost every job site gets a lot of mobile traffic.
This means users might behave similarly on job sites as they do on Instagram: quickly swiping through posts without paying much attention, like when they’re waiting in line or riding the bus home.
We’ve been trying hard to extend this attention span, but we’ve realised that we’re limited by how the human brain works. We understand that job seekers often make decisions quickly and based on subconscious factors. And these factors aren’t just important to us – they matter to them too.
At Jooble, we’ve recognised that to create a top-notch job search product, we need to understand what matters most in predicting these subconscious factors. It wouldn’t be so challenging if job seekers were more willing to assist us in finding the perfect job fit.
The more data, the better
We’ve found that fewer than half of the users who visit Jooble for the first time are searching for a specific profession, such as a nurse at a hospital or an accountant at a bank, and so on. Another sizable portion of our users is seeking jobs based on criteria like salary expectations or whether the job is remote or in an office.
There’s also an interesting group of users who are simply looking for “any job.”
For those involved in the job sector, the term “any job” doesn’t necessarily imply any job at all.
When we encounter the word “job,” it’s merely a search keyword for us, devoid of any specific meaning. However, our users can attach a wealth of information to it. They might have expectations regarding salary, employer benefits, and other factors, as each individual has their own set of expectations.
Here’s how it unfolds on a daily basis:
Jooble collects millions of data points, which we use to analyse what search keyword was entered:
Everything is important, and the more data, the better. So, such an analysis helps us build a system of recommendations based on similar behaviour patterns of users looking for jobs without precising any particular criteria.
We can see that at one point, we have this data, we know how to collect it, analyse it, and segment it, and we have two primary criteria to segment the audience based on them: the country and search keyword.
Segments
Consequently, we established segments to pinpoint common preferences and began promoting jobs that are likely to appeal to job seekers searching for any type of job.
In examining the UK market, we’ve observed that individuals seeking any job tend to click on job titles as depicted in the tag cloud in the image below:
We’ve implemented a comparable tag cloud for each profession and criteria-based search.
This strategy has proven effective. Following its implementation, we’ve achieved positive outcomes: increased interactions with job vacancies in France and remarkably high conversion rates in Germany.
But why was this not a success story?
Key Insight
Despite our excitement over the positive results, we overlooked an important factor: while job recommendations may perform well at a country level, their effectiveness can vary based on other criteria, such as region. For instance, every country has distinct industrial areas, business hubs, and tourist destinations, leading us to hypothesise that job preferences may differ in these regions.
Conducting some initial tests confirmed our hypothesis. In London, for instance, individuals seeking any job often click on Amazon Flex or waiter positions, given the abundance of restaurants. In Manchester, customer service and HR roles are popular choices, while in Bristol, airport and porter positions attract more attention.
This insight was incredibly valuable to us, and we acknowledged our oversight in not recognising it sooner. It serves as a significant learning experience, and we anticipate implementing this new approach to further enhance our results, as demonstrated below:
In conclusion:
Big data and the models derived from it are effective.
You may have brilliant minds on your team who can conceive groundbreaking features that drive success for your business and clients. However, it’s always prudent to complement creativity with data-backed insights for more reliable expectations and predictions.
The larger your dataset, the more crucial segmentation becomes.
With abundant data, there’s a higher risk of overlooking valuable insights when analysing information from a single perspective. Examining data from multiple angles helps uncover subtle details that can strengthen your business, benefit your clients, and support job seekers in finding suitable employment.
Impressive outcomes may conceal future growth opportunities.
Achieving positive results shouldn’t signal the end of innovation. Instead, it’s an opportunity to explore further enhancements and uncover hidden potential for growth.
To develop an exceptional job search product, thorough analysis of regional economies and the integration of AI models are essential, regardless of the order of implementation. The key lies in diligent information analysis. Jooble intends to expand this approach to its primary markets—Germany, France, and the United Kingdom—in the near future, with plans to extend to additional markets later on.
Learn more about understanding user intent with Jooble from Yana Levchenko’s presentation.
