Industry statistics show that it takes 15-20 minutes to enter a detailed candidate profile into a database manually with associated cost of around £2 per CV.

Modern software packages have automated many aspects of the recruitment process increasing consultants’ productivity many times. The main reason for this is that they operate on highly structured data stored in the underlying database. Structured data is used for de-duplication of newly arrived resumes, tracking of applicants through the recruitment process, linking applicants to requirements and clients and many other important tasks. The cost of entering data from unstructured resumes into the structured format of a recruitment database is usually hidden from the main picture but is nevertheless significant.

Industry statistics show that it takes 15-20 minutes to enter a detailed candidate profile into a database manually with associated cost of around £2 per resume. For a company which receives about 100 resumes a day (small to medium size agency) such costs can add up to £35,000-£40,000 a year and for a large recruitment agency it can be several hundreds of thousands of pounds. And resume application volumes increase dramatically every year.

To mitigate high labour demands of manual processing many companies register on the database only a small proportion of received resumes applying some manual pre-sorting. Other companies simply do not accept email applications but require applicants to fill in long web-forms. This reduces the number of entered resumes greatly and affects the chances of finding and placing a good candidate.

The idea of automatically parsing unstructured resumes into structured database fields has been around for quite a long time. The vast majority of such systems are based on keyword and pattern-matching technology where the system could recognize only phrases stored in its lists and combine them into larger fragments using a rigid set of patterns. In general, the accuracy of such systems is 40%-50% and therefore it sometimes takes more time to fix the errors made by the system than to do the entire job manually. Because of this automatic resume processing software has been widely criticised.

Recent advances in several computational disciplines (Machine Learning, Natural Language Processing, Artificial Intelligence and a few others) inspired by close interest of Government and Defense agencies and in particular DARPA, resulted in a new generation of document understanding technology -- very accurate and robust enough to be applied to real world tasks. This is why a few newly started software companies with strong links with leading Universities have been able to develop resume processing technology which is almost as accurate as real people. Such systems operate with a level of accuracy of 85%-90% (people are about 96% accurate on this task) and also have a number of advantages:

One of the most obvious advantages is savings in cost of labour. Automatic systems process resume almost instantaneously with very small associated costs, thus saving tens of thousands a year for small companies and hundreds of thousands for large ones.

Secondly, intelligent automatic systems eliminate resume backlog and make candidate details available for searching within seconds of a new resume arriving. This significantly increases competitive edge in the game 'who puts the right resume on the client’s table first'.

Thirdly, there is no need to compromise over which resumes to put in the database and which to leave unprocessed and therefore inaccessible. Automatic systems can cope with large volumes accurately extracting details from incoming Word documents and emails and, at the same time, the quality of the database can be maintained on the same high level by automatically marking eachresume to belong to a special pool e.g. talent pool, contractors’ pool, speculative pool etc.

The intelligent resume processing proposition fits in between standard manual (or partially automated) in-house data entry and complete data entry outsourcing solutions. Like with the manual in-house process, the automated data entry is done in-house in secure environment and is tightly integrated with in-house recruitment package and business rules. However, the number of personnel to be involved in data entry process is usually reduced by 75%, which makes it similar to a complete outsourcing solution where the internal staff is replaced by the third party. But outsourcing solutions are usually not cheap, have a time delay in resume registration and often require modifications in the underlying database of the recruitment package.

Currently there are just a handful of companies in the world that supply intelligent resume processing software with the majority of them being in the US. In general American systems tend to work well on American resumes but are not so effective on the UK, Irish, Australian and other non-American English language resumes. Another feature that is present mostly in UK/European systems is the ability to process resumes in foreign languages and recognise foreign addresses and other contact information.

Since the new generation of resume processing software is made by specialist companies, it is usually not bundled with any one specific recruitment or HR package but is integrated with several and can be integrated with new packages on demand. Therefore they do not replace currently installed recruitment packages but rather enhance their functionality replacing low-accuracy native data importers.

At Daxtra/Infogistics we recently experienced a significantly increased demand for our resume processing technology from agencies of various sizes. Elan’s Programme Manager Derek Macrae said: "We have reduced our resume registration costs by 50% whilst handling a greater throughput of resumes and with a higher quality of output data." Daxtra technology is also now being rolled out through a number of Elan's foreign offices.

"Headcount reduction versus the cost of software is probably the best ratio I've ever encountered" - adds Steve Higgins at MSB International.

Over the summer Infogistics has signed a number of smaller agencies (some of them under ten people) which still find it very attractive not just to decrease their data entry costs but also to free consultants to concentrate more on the placement process and generate revenue rather than sort and cut and paste resumes.

Recruitment will always remain a people's business, the consultative nature of which is based entirely on interactions between human individuals. However, some routine and mechanical elements such as data entry can now be successfully automated thanks to the latest advances in software technologies.