We have added NLP to our recruitment chatbot to help you map your data and start targeted conversations
As long as we’re hiring humans (and not robots), “Hello!” will always beat “fill out this form” as a first point of contact. Having a chat simply comes more naturally to most of us than filling a form. Recruitment chatbots can deliver a higher conversion rate for your career site or your Facebook page because of this simple fact. While conversion rates are great, the problem often comes after that, when you try to decipher all these “conversations” into meaningful data points. In other words, collecting the data is fine, but organizing it can be tricky. Mapping out responses to open-ended questions used to be a nightmare manual task.
Let me illustrate why… If you ask candidates where they are currently working will elicit many potential responses. If you ask “Who’s your previous employer?”, expect some candidates to answer “I used to work at Google” , other to say “Employed at Google for past 2 years”, other to say “I am currently working at Google” and thousands of other potential responses.
The ability to extract all this data (“past employer”, “duration of employment”, “current employer”, etc.) consistently used to be 100% manual… which is why people used to defer to forms over chatbot: less room for interpretation.
Enter NLP or the art of making your chatbots “understand”
This is where Natural Language Processing (NLP) comes in. With NLP, chatbots can understand the intent of a statement and find the information it is after. An “understanding” of the candidate’s response also lets chatbots interact more naturally which keeps the conversation going longer.
In the Talkpush CRM, where you can configure each questions in a given campaign, you can now map the type of data field that you are expecting to receive (e.g. “previous employer”) in response to a particular question.
In the example above, you can detect the sentence construction to determine which part of the sentence actually contains the actual value of the candidates “employer” as well as “employment period” which are pre-defined data fields in Talkpush.
Talkpush Data Fields
We now allow our users to map custom data fields to chatbot questions. When creating your interview questions, you can select the data field associated with each question. The NLP engine then takes care of the rest. With each answer, the candidate’s profile gets populated and the data is organized. The NLP continues to work in the background.
What’s next? More data fields and more segmented conversations
The default list of data fields which you can map to your questions is limited to a few standard fields for now (e.g. “previous employer”, “spoken languages”) but that list will be built and grown every week and should eventually contain hundreds of standard candidate fields. And we’d be happy to add some data fields at your request, if you simply email us what you need.
Next step? You might want to re-activate old candidate databases, by simply asking them: are you currently looking for a job? That’s the kind of data point that is worth its pound of gold!
As the depth of your candidate data map grows, so will the quality of your conversation with them. Increasingly, you will be able to have segmented engagements with subgroups within your talent pool. For example, you can reach out to all the “Java programmer” within a 2 mile radius of Bangalore (read about the GeoSearch feature here), with the certainty that you are not missing the target.
What is Talkpush? Talkpush is a conversation-first CRM which allows employers to connect with millions of job seekers digitally and to engage in conversation via chatbots over career websites, social media and SMS. Talkpush is being used by high-volume employers in over a dozen countries to improve the candidate experience and reduce the cost-per-hire. Find out more on www.talkpush.com or sign up here for a demo of the software.