The Good, the Bad and the Ugly - Wide Variation in Keywording on Fotosearch Site

Keywording spam, under-keywording, inaccurate keywords, misspelling and inconsistency have been highlighted in a survey of keywording found on image portal Fotosearch.

The survey, conducted by keywording company Keedup, returned a wide spread of results with scores of 93 out of 100, to as low as 29 amongst the 21 rights-managed photo libraries represented on the Fotosearch site.  The average score was 68.

"There was some really excellent keywording - clear, relevant and accurate - but other keywording was well below par," said Keedup's Stock Keywording Manager, Jess Wilson.

As the contributing libraries used their own metadata, this was not a reflection on Fotosearch itself, but on the wide variation in quality between the libraries.

"Considering that the success of any librarie's images on a site like this is dependent on their images being found before those of their rivals, poor keywording amounts to taking a knife to a gunfight.  Libraries with good keywording will inevitably do better, at the expense of their rivals.  It's as simple as that."

The keedup survey was a qualitative assessment of the keywords, looking at such factors as spelling errors, consistency, whether the keywording included extraneous detail, or left out vital information.

"What we hadn't been prepared for was how many inaccurate and just plain strange some keywording was," Miss Wilson said.

One example was a set of keywords for a picture of an old woman leaning on her hand, close up from chin to forehead.

The keywords included were:
‘accomodation, annuity insurance, bench, boarding-house, calm, care, dementia, granny, hands, health, health insurance, home, need care, older, park, park bench, pelt, pension, rest, seniors, silence, sitting, skin, support, weakness, age’

"It's hard to see the park bench, or a pelt.  Whilst the woman might be at home, using the keyword 'home' is hardly relevant.  And although she might be sitting, you can't actually tell. Was she in a boarding house? Who knows.  There were other obvious words that could have been put in that weren't, such as 'elderly' or 'woman'."

For more information about the survey results, email