Performing a Search for Difficult to Fill Role

Created: Feb 04, 2015 05:02PM PST      Updated: Oct 20, 2015 10:58AM PDT
We already know that TalentBin excels at finding passive technical candidates for open source software engineering roles.

However, let's take a closer look at how to search for hard-to-find candidates for a much more specialized software engineering role.

For this example, let's say we've been tasked with finding software engineers who were previously employed by Santander Bank and used their proprietary financial software Banksphere or BKS. Ideally, we'd like this person to be located in Boston.

That's our example. But the search strategies below can easily be applied to research any purple squirrel using TalentBin.

Getting Started
1. Create a folder for your search.
2. Click on the Advanced Search and add a few search fields.

Skills-Based Searching
Remember, TalentBin does best with skill-based searching. So the first thing you'll want to search on in the most specialized skill that your candidate is required to possess. Every new search facet you add is weighted approximately 50% less than the search term before it.

As you're building your search, feel free to hit the search button from time to time to vet your search results. This is especially helpful when sourcing specialized roles. In fact, it can be quite instructive to see how each term impacts your search results.

For example, we believe Banksphere to be a specialized skill but how do we prove that to our client? Simply enough, run an unrestricted search on that word using the default "Located anywhere" as the location. This will perform a worldwide search for passive candidates who match that keyword.

As you can see below, there are only 52 matches. A rare bird indeed! That doesn't mean that there are only 52 people with this particular skill in the world but rather that there are 52 who have mentioned "Banksphere" on one or more of the 44 or so public datasources crawled and collated by TalentBin.

Figure One: Performing a Simple Search.

Since we know BKS is a synonym for Banksphere, we can try adding that either as a custom synonym or as part of an "Or Search" (on the same line).

When is it best to create a custom synonym?
A: You can create one anytime you like. However, custom synonym lists are best for words that you'll be using repeatedly in later searches within your bin. If we know we're going to be doing a lot of future searches around Banksphere, we could create a custom synonym list, but since we believe this will be a one-off search, let's just add the word BKS as an "Or Term" to our search.

Now when we perform the search, we can see our search results have gone up significantly. We've gone from 52 matches to 1,337 matches worldwide. It's still a rare bird but not quite as rare as we'd previously thought. On the other hand, that is such a large jump that we have to safeguard against the likelihood that we've introduced a number of false positives into our search results.

Filtering Out False Positives
We can choose to deal with false positives in a few different ways. One would be to manually vet each candidate and simply begin staging as either "promising" or "bad fit" based on other information in the profile. With a small dataset this is humanly do-able, but greater efficiency can be obtained by looking for other ways to further refine the results before we begin to vet profiles individually.

Figure Two: Refining the Search.

Let's scroll through the first page or so of the search results and see what kind of people we're returning. Does it look like all of the BKS search results are related to Banksphere, or are we getting some false positives? If so, what else might BKS stand for and how might we exclude that from our search? lists a number of things which BKS might stand for besides Banksphere. If any of these are falsely increasing our results, we can attempt to remove them by adding a prohibited term to our search. For example, what if we wanted to remove "Barnes & Noble Bookstores." Let's try that to see if it has any appreciable difference to our search results.

Removing "Barnes & Noble" did result in a slight dip in search results. The results dropped from 1,337 to 1,319. However, this may simply be due to people who liked "Barnes & Noble" as part of their data exhaust. 

Figure Three: Refining Search Results.

Are there any other synonyms we could add to our search list for Banksphere? If so, how would we find them? One way to locate them would be to search through the existing results to see if there are other words being repeated on the profiles, i.e. words that passive candidates are using to describe this particular skill. This is also a great way to identify false positives or additional words that can also be removed as you continue to refine your search.

Another strategy would be to do a simple Google search on the word to see if there are other closely associated words.

In this example, the phrase "BKS Bank AG" comes up with enough frequency that we'll want to research further to determine if that is something we should filter out or potentially add in order to find additional candidates. A simple Google search reveals that "BKS Bank AG" is actually a separate financial institution based in Austria. They probably have nothing to do with Santander Bank or Banksphere so let's remove that term from our search results.

Removing that term takes us from 1,319 results to 1,314 results, so it wasn't particularly prevalent. You can continue to fine tune in this manner as you wish.

Restricting by Location (starting to close the net)
Let's see how many of these passive candidates are available in the United States. Now, why would you want to do that? Well, again, it's useful demographic data you might wish to pass on to your client or the hiring manager in terms of managing expectations. Remember, the client asked us to find this person in Boston if at all possible.

Figure Four: US results only.

Choosing United States brought the search results down from 1,314 to 622. As we scroll through the results, however, we can see we're still getting all sort of unusual things which really have nothing to do with Banksphere. There may be other conditional matches on "BKS" that will now reveal themselves.

Refining by Role
But let's set that aside for a moment and see what happens if we choose to add the role "Software Engineer" to the search and/or do a prohibited search on most groups of non-software engineers by excluding roles with the explode terms of "Sales", "Leadership" and "Recruiting", since this will also be an effective way of removing false positives as well.

When we add the explode term "Software Engineer" we quickly find that the search has become too restrictive. There are only 12 results. And it isn't even clear from the summary whether any of them have worked with Banksphere.

Where do we go from here? Probably a good time to pour a stiff drink. I'll wait.

Backing Out
Whenever a search becomes overly restrictive, you can ease out of it by making one or more of your search terms optional or nice to have by choosing the + sign to the left of the search field. This will prioritize those matches at the top of your search results. Let's try doing that for "Software Engineer."

We're now down from 622 candidates to 523 candidates. But are we any closer to finding our elusive purple squirrel?

Figure Five: Removing Non-Technical Results.

A quick look at the summary of the results would indicate that we're still pulling in lots of non-technical people. Perhaps at some point, we simply have to admit defeat and realize that adding BKS has muddied the waters too much. The results are simply too unreliable and not worth additional manual vetting.

What if we tried to add something like "BKS-Banksphere" or "BKS, Banksphere"? Let's try that now. No appreciable change in the search results.

What if we add "B.K.S."? A dramatic increase in the results but many (most?) again seem to be false positives.

Let's just remove "BKS", "B.K.S.", "BKS-Banksphere", and "BKS, Banksphere."

Now, let's try adding "Financial Services Companies" or "Santander Bank" as a Company facet.

No results. What?

Widening the Net More
Let's open up the search again so that we're looking worldwide.

Now we're back to 42, which is 10 less than our original one word search a half hour ago! But looking through the search results, it would appear that everyone is well-qualified. If they're not developers, they're IT Analysts and Managers and, for a rare position such as this one, certainly worthy of our time in terms of reaching out. For example, if they aren't interested, they may be able to give us the names, emails and phone numbers of other people they've worked with in the past. Never discount the power of a referral network.

There seems to be a large cluster of Banksphere expertise in Madrid. Madrid is a long ways from Boston. However, one or more of these passive candidates may have other Banksphere software engineering referrals, even in the United States. Only one way to find out.

Saving Two Searches
Let's add email required from the Available Contact accordion and search once more. 16 with email. Not a bad place to start. Let's save that search and then save the search again without email (really, it's with or without email, i.e. email not required).

Now, we have two saved searches in our BKS bin. After further vetting, we're going to create a customized outreach template and email all of the passive candidates for whom we have email addresses. As soon as we contact them, they'll be staged as "Attempting to Contact."

Next, we're going to load our other search and request all of the missing email addresses. This will be a snap since TalentBin's folders automatically remove duplicated profiles from a search that have been staged. And since TalentBin is successful finding missing emails approximately 70% of the time, we'll likely be able to email another 18 or so candidate emails, for a total of 34 candidates.

And for those candidates who don't respond, we can also consider contacting them using other vectors exposed in TalentBin such as Twitter, Meetup, and LinkedIn.

Remember, we're attempting to build a Banksphere engineering pipeline. Our goal is to hire a software engineer. We're going to have to be especially resourceful given how rare this particular skill appears to be.

Q​uestions? Thoughts? Either c​ontact your Product Specialist, reach out to us at anytime, or leave a comment below!
Comment Form is loading comments...

Contact Us