After I was surprised by the low numbers of images for a range of ethnicities on iStock, I analyzed annual downloads (Total DLs/age of image) to see if the low result counts were justified.
Average Age of Image in BM vs. Download Sorts
- Given the high number of Agency images in Best Match, I wasn’t surprised by the low average age of BM results.
Conclusions
For yesterday’s post on Summer images, I analyzed data on 1,000 images across ten summer-related searches. I was curious about the relationship between the age of an image and the number of downloads it had.
Total Downloads vs. Age of Image
- The images with the most lifetime downloads are not the oldest ones — rather they are between 3-5 years old.
- The correlation between age & downloads is very close to zero which suggest little to no correlation between the two.
Annual Downloads vs. Age of Image
- The distribution looks very similar here.
Conclusions:
- Just because an image is old, doesn’t mean it will get a lot of downloads.
- New images have come in that are downloaded at a rapid rate.
- Again, as we saw with contributors, age doesn’t appear to be the dominant factor.
I searched for ‘business’ on iStockphoto, and analyzed the top 100 results returned for Best Match & Downloads sorts. For each image, I recorded:
- Total Views & Downloads
- Upload Date
- Search Position
Using the data above, I was able to calculate the age of each image, and therefore, calculate the average annual downloads for each image. I used the median instead of the mean for averages, to prevent outliers from skewing the results too badly.
Median Age of Image
- This is the difference between the upload date of each image and the date of the analysis – April 30, 2011.
Median Value for Total Downloads
- This is the median value of the total downloads for each images in the Top 100 search results.
Median Value for Annual Downloads
- This is the Median Value of the Total Downloads divided by the age of each image.
Things That Stood Out to Me
- 46 of the 100 images were present in both Best Match and Downloads sorts.
- I expected a bigger spread in median age of image between Best Match and Downloads.
- I was also surprised by the parity between the median annual downloads in each case.
Over the next few days, I’ll analyze these searches in more detail, and compare them to other subject categories.
If there are topics that interest you, please let me know in the comments.
I studied the difference in counts and downloads between people vs. non-people images by running searches on iStock for: business, education, industrial, lifestyle, and medical images. I recorded and analyzed the number of images and also calculated the number of downloads for the Top 200 images.
Image Count
- 7.6% of the images in this sample had nobody in them. (Note: there may be some distortion here due to limited use of the keyword ‘nobody’. But, the ‘No People’ filter includes images of body parts, which I don’t want to count in this analysis).
- 16.4% of the images in the industry category had no people in them (2.3x average).
- I expected business and lifestyle to account for more than 10% of iStock’s eight million or so images, overall.
Average DLs/Image for the Top 200 Images in Each Search
- I obtained the average DLs per image by summing the downloads of the Top 200 images in each search and dividing by 200.
- Overall average DLs/image for the categories shown: 1,153 DLs/image.
- Overall average DLs/image for images with ‘nobody’: 588 DLs/image.
Some Caveats:
- This analysis only applies to the chosen categories. I think the performance of people vs. nobody will be different when looking at subjects like ‘Christmas’.
- I used ‘nobody’ instead of the ‘No People’ filter to exclude images of handshakes, or fingers pointing etc. (Also, the ‘No People’ filter doesn’t work.)
- The reason lifestyle & medical are at exactly 1000 DLs/image is because each of the Top 200 images are in the 1000+ download band.
- Data is cumulative.
