Author Topic: Crunching some numbers.... (Read 7839 times)

hatman12 · « **on:** May 19, 2008, 18:09 »

In a bored moment I crunched some numbers....

SS is now accepting 52,000 new images every week. The most recent image number is above 12 million and they have 3.6 million images accepted. That suggests an acceptance rate of 30% (rough round numbers). Putting it another way, if 52,000 are being ACCEPTED each week, that implies that 170,000 new images are being SUBMITTED every week from 102,000 contributors.........

Over at iStock the queue is presently 54,000 or so. The most recent image is number 6.1 million and they have 3.1 million accepted implying an acceptance rate of about 50%. That suggests that of the 54,000 in the queue about 27,000 will be accepted.

There are many people wondering about saturation and dilution at SS and these numbers seem to add fuel to that fire. No wonder it is becoming difficult to grow one's income at SS.

There will be many people becoming increasingly disillusioned at SS. Will they stop contributing, or stop taking stock photos? Or will they try again elsewhere? Once the dilution has damaged sales at SS will the same thing happen elsewhere as thousands of disillusioned contributors all try 'other agencies'.

cphoto · « **Reply #1 on:** May 19, 2008, 18:18 »

Actually the SS image number are artificially inflated and you should take that into account. It was noticed a long time ago by Cyril that when you submit a batch of new pics, there are 2 unused numbers between each pic submitted in sequence.

For instance in my latest batch, I have pictures # ...217, ...220 and ...223.

Going back in time in older batches the same was already happening.

So looks 2/3 of numbers are not attributed to pictures... Weird system.

DanP68 · « **Reply #2 on:** May 19, 2008, 18:27 »

Yep, you were bored.

But you are right regarding the difficulty to keep one's sales above water, particularly at Shutterstock. Dilution has to be happening given the rapid growth, but compared to the 12 million + on Alamy, microstock libraries are still quite small.

I'm torn about this Hatman. The Shutterstock search engine is by far the most fair in the industry. Most Popular is simply downloads divided by time online. Nobody receives favoritism, and there are no bizarre community ratings or anything else * up the results. WYSIWYG.

The downside is, the search does not work well over the long term. Images get killed off far too quickly, and once they lose a little exposure the drop is precipitous. Because of the nature of the Most Popular search, technically an image with 4 sales in its first 2 days is considered more popular than an image with 3000 sales in its first 1501 days. This is the reason new images do so well, and older images can get killed off so quickly. Once the 3000 sales image moves down in the search order to make way for unproven newer images, its sales slow down, and it falls further. Vicious cycle. And even if the image with 4 downloads in 2 days flames out quickly, there will 10 more to replace it that week. The older proven image barely stands a chance.

I think a potential fix to this would be to give added weight to total downloads. In other words, give maybe 50% weighting to Downloads per month, and 50% to Total downloads. I don't know if this would help or not in practice, but I do think they need to try something if they want to retain older contributors.

But then again, maybe the nature of subscription sites is to offer new material all the time, regardless of whether more proven images might be of higher quality. I don't know.

CofkoCof · « **Reply #3 on:** May 19, 2008, 18:41 »

Quote from: DanP68 on May 19, 2008, 18:27

The downside is, the search does not work well over the long term. Images get killed off far too quickly, and once they lose a little exposure the drop is precipitous. Because of the nature of the Most Popular search, technically an image with 4 sales in its first 2 days is considered more popular than an image with 3000 sales in its first 1501 days. This is the reason new images do so well, and older images can get killed off so quickly. Once the 3000 sales image moves down in the search order to make way for unproven newer images, its sales slow down, and it falls further. Vicious cycle. And even if the image with 4 downloads in 2 days flames out quickly, there will 10 more to replace it that week. The older proven image barely stands a chance.

Just a small correction of this, from the SS FAQ:

Quote

However, we use an adjustment factor to prevent a photo that has had a quick 5 or 10 downloads in the first day from showing up in the search results. It would take extremely heavy downloading in the first few days for a photo to overcome this adjustment and show up high on the "most popular" list. As time goes by, the weights become exactly proportional to elapsed time. By including a brief period of under-weighting, the "most popular" results can be relied on to return the images that are popular in a sustained way, rather than photos that have been on the site the longest, or photos that were popular for their first few days but had no staying power.

I was wondering today what would happen if they would remove the Newest and Oldest first searches. The images wouldn't get a bump, it would be harder for an image to get high in the most popular images, but it would be good for older images. They would also sell. I think

DanP68 · « **Reply #4 on:** May 19, 2008, 23:45 »

Thanks Cofkocof. I've read that too.

However in practice, that's not how I see things working. Over the last few months, new images which I uploaded gained a few sales per day in their first few days online, and immediately shot up to near the top of the most popular search. When I sort my own portfolio, I typically see very new images take over the top spot once they get a few sales.

There may be an adjustment. But when I look at the real world results, I see very little adjustment.

epixx · « **Reply #5 on:** May 20, 2008, 00:05 »

SS is based on volume sales. To get that, they need to show the customers new photos every day, so that they get "value" for their money. I don't think this will change, and I don't think that SS is what I will count on for my retirement

Last year, SS represented 41% of my microstock income,
ytd 18 May 2008, it was 35%
May this year so far, it's 28%
Last week was 25%

If it continues like this, three or four other micro agencies will pass them long before the end of this year. When they pass 15% on their way down, I will consider to stop uploading to SS. It's not worth selling my photos that cheap for 15% of my total earnings.

leaf · « **Reply #6 on:** May 20, 2008, 01:11 »

I still think shutterstock has one of the best systems.

This year I have been uploading actively bust before that it had been 1.5 years since I had really been uploading at all. During that time my shutterstock earnings dropped 1/3 of what they were at their highest but stayed there for the next 18 months. Earnings do not fall and fall and fall, they go down to where they are without the 'new uploads boost' and stay there. Even after the fall they were still my best performer - and have been ever since I started uploading in 2004

I think the search method works quite well because ALL new images get a chance to shine. They get put on the first page of new uploads and if they are attractive to buyers they get downloaded. if they are very attractive they can end up in the first page of popular images and stick there. In the end it is the buyers who decides what ends up at the top of the best match searches. Good images rise to the top and poor ones sink.

On other sites you could create the best image ever but due to the fancy search algorithms it might end up on page 45 of 90 in the search results. No buyers will end up finding it, it never had a chance to shine and you end up with 0 sales. Yet on shutterstock it might become a best seller.

Freezingpictures · « **Reply #7 on:** May 20, 2008, 01:27 »

Quote from: DanP68 on May 19, 2008, 23:45

Thanks Cofkocof. I've read that too.

However in practice, that's not how I see things working. Over the last few months, new images which I uploaded gained a few sales per day in their first few days online, and immediately shot up to near the top of the most popular search. When I sort my own portfolio, I typically see very new images take over the top spot once they get a few sales.

There may be an adjustment. But when I look at the real world results, I see very little adjustment.

I think the truth is between what you experience and the FAQ's.
I have an image which sits now for year or so in the top position and even if it has for a week less downloads than a new images, it still stays at that place. It was very popular one time and occasionally becomes popular again. But for small periods other newer images are more popular but still are not able to move that image down.

SS's way is kind of nice to see instant results, but for me the long time results are kind of disappointing. I have higher quality and more images online, but it is not reflected in my earnings.

sharpshot · « **Reply #8 on:** May 20, 2008, 01:49 »

SS has been my highest earner every month since I started. I can't make much progress there now but I don't think that has anything to do with dilution. I think it is just because old photos are hard to find with their bias towards new images.

I would like to see them try something like the lucky oliver sideshow. Would be great if they did this for the people in the $10,000+ tier. I can't see them changing though as nothing new has happened there for a long time.

epixx · « **Reply #9 on:** May 20, 2008, 01:56 »

Quote from: leaf on May 20, 2008, 01:11

I still think shutterstock has one of the best systems.

The question is: best for who?

I had my last BME at SS in May 2007. Now, I have almost twice the portfolio and better photos, but I'm not even close to those sales. At the same time, sales at other agencies, particularly FT, IS and StockXpert, are soaring.

DanP68 · « **Reply #10 on:** May 21, 2008, 03:19 »

Regardless of what the FAQ states regarding a weighting on the most popular images sort, Imagist posted this on the Shutterstock boards:

Our four search result sort orders are now - and have always been - transparent. A description of our Most Popular algorithm is in our FAQ: it's a ratio of the total number of downloads the image has enjoyed in its life on the site divided by the total number of days it has been available to buyers. The other three, Newest, Oldest, and Random, are self explanatory.

http://submit.shutterstock.com/forum/viewtopic.php?t=38149&postdays=0&postorder=asc&start=150

Exactly as I have observed. Most Popular = DL's / Days online

CofkoCof · « **Reply #11 on:** May 21, 2008, 03:49 »

Quote from: DanP68 on May 21, 2008, 03:19

Regardless of what the FAQ states regarding a weighting on the most popular images sort, Imagist posted this on the Shutterstock boards:

Our four search result sort orders are now - and have always been - transparent. A description of our Most Popular algorithm is in our FAQ: it's a ratio of the total number of downloads the image has enjoyed in its life on the site divided by the total number of days it has been available to buyers. The other three, Newest, Oldest, and Random, are self explanatory.

http://submit.shutterstock.com/forum/viewtopic.php?t=38149&postdays=0&postorder=asc&start=150

Exactly as I have observed. Most Popular = DL's / Days online

He posted that FAQ describes how the Most popular algorithm works

He probably just didn't wanna go into details on the forums.

The other day one of my illustrations had 25+ downloads on the first day it came up. I checked how it was ranked by Most popular and it was on the 3rd page (using most popular alg.). Then after a few days of up to 10 downloads per day it started climbing, reaching a peak after about a week (that's why I think the adjustment only applies to files newer than 1-2 weeks). I think you would need something like 20 times as much downloads on the first few days on an image to make it most popular.

MicrostockGroup Sponsors

Author Topic: Crunching some numbers.... (Read 7839 times)

hatman12

cphoto

DanP68

CofkoCof

DanP68

epixx

leaf

Freezingpictures

sharpshot

epixx

DanP68

CofkoCof

Related Topics

Sponsors

Microstock Poll Results

Sponsors