We often get asked by clients if there isn’t some way of getting the info from the CV into Placement Partner without having to capture it manually. Everybody is kind of hoping that there is a button which says “Read CV” which then does all the capturing automatically.

Well, for some time now, we have been investigating so-called parsing technology, which is aimed at reading CV’s and categorising the raw data into name, ID number, telephone number, qualifications and so on. It has come a long way since our first tests of various 3rd party systems, but the fact remains that you will never achieve the same level of accuracy as you can achieve using manual capture. The machine makes mistakes.

You basically get two kinds of mistakes: information that is in the CV that doesn’t get filled in, which is the lesser kind of mistake and then the big problem – information that may or may not be in the CV that gets filled in incorrectly. The first type, the missing info, is easy to spot because there is a blank field. The second type, a wrongly completed field, is far harder to identify and likely to be missed. That is, if you are checking at all. If you are parsing hundreds of CV’s, you probably wouldn’t be checking them.

Why is accuracy important? Well you need to be able to trust the data in your database. Firstly, neat automatically formatted CV’s which are sent to clients draw their information from the database. Secondly, if you don’t trust the data, then the value of your database declines. If you have a sneaking suspicion that 10% to 20% of the fields in your database are wrong, your trust in the data disappears and the entire recruiting system relies on your data.

Various accuracies are claimed for these parsing engines, but we have gone a step further and tested them with a random set of 100 Souf Effican CV’s. Everybody is in there from the Dr So-and-so’s perfectly manicured pdf to the barely literate’s hand crafted Word doc. We have found an accuracy of 83% is about the best one can realistically hope for, with an almost equal split in errors between “not filled in” and “wrongly filled in”.

Now we would like to put it to you, our clients, would you be interested in a CV reading module, that could populate your PostBox with CV’s that were on average 83% pre-captured, plus about 8% incorrectly captured info? Some CV’s would be near perfect, others horrible and occasionally not captured at all for some reason, but 83% of the capturing has been done.

This is a question we can’t answer ourselves. If we developed this module and it was offered at the same price as the PostBox & Web Integration module, would you sign up for it?

We would be glad to hear your thoughts.

The Placement Partner Team