MyCareer’s Search Results
September 10, 2008 by Geoff Jennings
Back in July, I reported on MyCareer’s search for excellence. MyCareer were testing a new search box on their home page with keyword and category search versus keyword only.
Seems that the winner is the old keyword and category search. It would be interesting to see the A/B test results to see how the keyword-only search fared.
Update: 16/09/08
I don’t know what is going on at MyCareer now…the keyword-only box is back up. Perhaps they’re still doing A/B testing?
An over-exaggerated amount of attention has been given to the importance of keyword-only search in our space – largely driven by society’s adoption of Google over the last 10 years.
Many recruitment advertisers have not realised that there is a substantive difference between a Google-style search and a job search.
Generally on Google, users have a pretty good idea of what they’re looking for. They enter some criteria, and then make a selection from a list.
In jobsearch, the reverse is true. Jobseekers have a fairly good idea of what they don’t want, and are fundamentally looking to eliminate these things so to be left with a list of relevant career options from which they can choose.
Our research shows that ‘relevance’ to an individual can include both accounting and ski instructing in a single list of job results, reflecting the possible career options for an individual user.
In other words, Google;’s algorythms are geared towards an inclusion model, whereas jobsearch (and dating search and some others) are more effective when geared towards an exclusion model.
When it comes to keyword-only search, the mathematics of the search interaction can be described as getting 200,000 jobs down to abnout 100. Keyword-only search is simply not a blunt enough instrument for users to achieve an effective elimination of irrelevance, compared to classification taxonomies, or structured data like salary ranges, locations, employment-type, posting recency and so on.
That said, on employment websites with small inventories of job ads, highly ambiguous classifications or a niche market segment, keywords can be a more effective tool to deal with the mathematics of job search.
Accordingly on SEEK, you find that around 60% of job searches do not contain any keyword at all.
Of those searches that do use keywords, a majority are using a keyword that is a classification e.g. accounting or sales.
The number of google-style keyword-only searches hovers around the 10% mark.
This varies substantially by industry segment of course – you’ll find that keywords are much more used in I.T. than they are in Customer Service, simply because of the higher volume of acronyms and abbreviated unambiguous technology names that can be readily found using keywords.
Keywords are also highly effective for those roles that have a unique unambigiuous nature where only one word in the language captures the entire discipline exclusively e.g payroll or taxation.
For words or phrases at the other end of the spectrum and / or where there is a very high volume of advertising accross multiple disciplines (‘account manager’ or ‘business analyst’), keyword searching is probably about as useful as a chocolate teapot.
It is therefore no surprise to see where the A/B testing ended up.