During the recent months at Techloop, we have introduced several new features and improvements that we shared. One of the most prominent changes is the new public job portal mentioned last month.
One thing we secretly launched and are revealing now, is that Techloop released a whole new matching algorithm, replacing our previous search engine Algolia, for ElasticSearch. So why did we do this and what are the benefits? 🧐
The key element for Techloop has always been to quickly find the best match by selecting appropriate tech skills, specialisations, location, or language to identify relevant candidate profiles. The search criteria would then be shown as tags, that represent your required search results. This is used for automatically listing all matching candidates when creating an Open Position, but also for searching specific results without any Open Position created (Search). Thanks to this, users leveraged our unique dataset over Candidate profiles, making the whole sourcing very easy.
While this is something our users have seen since the launch of Techloop, one of the core issues was that you could only filter out specific profiles that matched the exact search criteria. Result? You were shown only a handful of profiles that matched perfectly but could not see other profiles which have a very similar skillset. This gave our users a feeling that there are not many candidates to choose from 🤷♂️
It is clear that HR Managers, Recruiters and Tech leads responsible for sourcing candidates want the best match to their Open Positions. Since Techloop filtering & matchmaking offers a wide range of data points sometimes it was tricky to see only results that exactly match all criteria.
With the shortage of Candidates on IT markets (750K IT professionals are lacking in Europe only) sometimes it is necessary to compromise on some criteria which are not crucial. We saw many times that a Candidate matching 6 out of 7 criteria is at the end better fit than 7 out of 7 matches. We wanted to show those Candidates to our clients and give them the chance to connect even though they are not the perfect matches at first glance.
Introduction of Elasticsearch! 🚀 After doing careful analysis, our decision was to completely revamp how search is done, without breaking the simple user experience we had before. Today you can still filter out your criteria as before, but the effect shows on average 8x more results due to our unique matchmaking algorithm.
From a user perspective, results on your search today are categorized in 2 main ways:
Best Matches – these profiles will be at the top of your search results, showing exact matches based on your search criteria.
Suggested match (To discover) – after reviewing Best Match profiles, we are able to identify similar profiles that meet your search criteria between 90-60% which is shown on the indicator of each profile. This allows you to browse similar profiles that may be close to your expectations, resulting in an interesting profile worth contacting.
The magic behind explained! 🎩 We use the ElasticSearch engine to create our own unique matchmaking algorithm. This algorithm is used in many places within our platform, for example: listing all matched Candidates, for your Open Positions, for listing all search results, for matchmaking emails and much more. Every Open Position search we calculate a weighted score for each Candidate profile and sort them from the best. Candidates with 100% score are shown as the Best Match.
The score considers how the Candidate meets the criteria of Open positions/search plus we add our platform data such as Response Rate %, Candidates activity etc. If you have exactly two same profiles, the one with higher activity would promote a higher score in the results. We always show only relevant Candidates with relevant tech stacks.
What other positive effects has this had?
- Smarter geo matching based on location requirements
- Powerful full-text search, allowing you get results from information in the work experience or bio of a candidate profile
- Matchmaking emails, sending you a list of recently matched profiles directly to your inbox
- Open Positions are suggest to candidates weekly as well based on their skillset
We are still at the beginning of an incredibly powerful tool, and will be constantly tweaking our algorithm to bring even better matching results. What’s ahead of us? First of all we want to add possibility to promote specific filters and give them higher weight/score ratio. For instance, some of our users consider location to be extremely important criteria, while for other users it’s not that important as they hire remotely. Other users may require language skills to be more important and so on. With a filter promotion ability, Companies will create their unique algorithm and can benefit from ElasticSearch even more. Last phase will be implementing machine learning on our data sets and analysing them giving even more precise results.
Let us know your thoughts and feedback! 🚀
Andrew Elliott – CEO & Co-founder at Techloop
