MicrostockGroup Sponsors


Author Topic: Too many keywords with some agencies?  (Read 3585 times)

0 Members and 1 Guest are viewing this topic.

« on: March 25, 2022, 03:05 »
0
There is a belief that with some stock agencies, there is such a thing as having too many keywords. After a few years of shooting stock, I still don't really know if this is really a thing or not. On DT, there are some contributors who advise against adding too many keywords as supposedly, this can hurt sales. By contrast, on SS, the general philosophy is to add as many keywords as possible. I actually mentioned on the SS forum that folks over on DT were advising against putting in lots of keywords. An SS contributor replied that this was nonsense and urged me not to listen to them. So since then, I have been adding as many keywords as possible with all the agencies that I submit to.

Though now I'm wondering if I'm doing more harm than good with certain agencies.

AS is another agency that, if I recall correctly, warns against adding excessive numbers of keywords. But unfortunately, I can't help myself.

Actually, some time ago, I did a little test on DT. I tried some searches for one or two of my images on there that had a rather large number of keywords. And I had no trouble locating those images in the search results. So I'm guessing that buyers wouldn't have any trouble either.


« Reply #1 on: March 25, 2022, 04:35 »
+4
Since no agency is really willing to share with us how exactly their search algorithm works, we can all only speculate.
I think too many keywords, of which some are less relevant to the image than others (keywords that are not relevant at all should never be used, no matter what. It just makes the search difficult for customers and hurts us all in the end) might indeed be bad for a certain type of possible algorithm: If a customer seraches for your image with a keyword and buys the image, the image raises in rank. I think this prinicple is part of every agency's algorithm. But some agencies might possibly use an algorithm where, if a customer searches for a keyword, looks at your image, but decides against buying it, your image might actually lose ranks for that keyword.

However, my personal opinion is that microstock just doesn't earn enough to justify the extra time it would take to create different sets of keywords for the same image for every agencies, even if I fully understood how each of their algoirthms worked. I add my kewords to the metadata and that's it.

« Reply #2 on: March 25, 2022, 06:02 »
+1
I think there's a difference between primary keywords that exactly describe the subject and descriptive, secondary keywords, that describe the situation, emotions or shooting conditions.

I try to be as accurate and relevant as possible with my primary keywords. These are main keywords people are looking, so it's worth your time to think them through carefully.
Most of the times, this means 3 to 5 keywords. Less is more indeed. Adding less relevant or wrong keywords here will hurt your sales, as it shows up in the search results without anybody clicking on it.

I fill the rest, secondary keywords, with relevant keywords that describe the situation or shooting conditions as good as possible. 
Example: no people, selective focus, side view, sunny, overcast ... to name some very common ones.
Nobody will actually search for secondary keywords alone, but buyers use them to help them to get a better focused search result.
I guess, when agencies tell you to maximize your keywords, they actually mean adding as much as possible relevant secondary, descriptive keywords.

« Reply #3 on: March 25, 2022, 06:23 »
0
Since no agency is really willing to share with us how exactly their search algorithm works, we can all only speculate.

Generally true. Though unless my memory is playing tricks on me, I thought I recall AS making two recommendations -  avoid using long titles and avoid using too many keywords.

I think too many keywords, of which some are less relevant to the image than others (keywords that are not relevant at all should never be used

Oh yea I only use keywords that are relevant to the image.

However, my personal opinion is that microstock just doesn't earn enough to justify the extra time it would take to create different sets of keywords for the same image for every agencies, even if I fully understood how each of their algoirthms worked. I add my kewords to the metadata and that's it.

Agreed. It's not worth the extra fuss.
« Last Edit: March 25, 2022, 06:36 by dragonblade »

« Reply #4 on: March 25, 2022, 06:35 »
0

I try to be as accurate and relevant as possible with my primary keywords.

Same here. I always choose primary and secondary keywords that are relevant.

Nobody will actually search for secondary keywords alone, but buyers use them to help them to get a better focused search result.

I'm guessing that sometimes buyers will use primary keywords only and other times, they will use a combination of primary and secondary keywords. I have used secondary keywords like 'urban', 'sunny' and 'outdoor' depending on the image.

« Reply #5 on: March 25, 2022, 08:49 »
0

However, my personal opinion is that microstock just doesn't earn enough to justify the extra time it would take to create different sets of keywords for the same image for every agencies, even if I fully understood how each of their algoirthms worked. I add my kewords to the metadata and that's it.

agreed.  My only difference, is once in a while i'll add all the recommended other keywords on SS, since they obviously know better than i do about their algorithm 

« Reply #6 on: March 25, 2022, 10:14 »
0
Where does don't use long titles on Adobe come from?
« Last Edit: March 25, 2022, 10:21 by YadaYadaYada »

« Reply #7 on: March 25, 2022, 18:52 »
0

However, my personal opinion is that microstock just doesn't earn enough to justify the extra time it would take to create different sets of keywords for the same image for every agencies, even if I fully understood how each of their algoirthms worked. I add my kewords to the metadata and that's it.

agreed.  My only difference, is once in a while i'll add all the recommended other keywords on SS, since they obviously know better than i do about their algorithm

I don't submit photos to SS anymore. But when I did, there were many recommended keywords that were irrelevant and had nothing to do with the image that I was submitting. So yea I would be extra careful about which ones I would choose from the automated suggestions made by SS.

« Reply #8 on: March 25, 2022, 18:57 »
0
Where does don't use long titles on Adobe come from?

When I first started contributing to AS, they had some guidelines on their website for new contributors to follow. And that included not having too many words in the title.

However, recently, I saw AS providing an example of an illustrative editorial title and it was quite long. So this contradicts what I read in their guidelines from a few years ago.

zeljkok

  • Non Linear Existence
« Reply #9 on: March 26, 2022, 13:56 »
0
"Less is more" is part of Adobe company culture that goes beyond stock.  Opposite is Alamy discoverability criteria that requires at least 40 tags for "optimized" label (even if this does not probably have practical importance).  Some agencies (IS, AS) do not index title, others (Alamy, DT) put heaviest weight on title.  Then there is IS with clunky and non-portable managed dictionary concept. List goes on.  Bottom line is that it is impossible to come up with identical set of metadata that will be optimized for every agency out there

My take is to focus on the asset without thinking about target agency.  Then as certain photo types sell better on some agencies  refine on individual basis, but only if photo has strong sale potential.  I find Alamy measures tool invaluable as these are real customer search terms, while I largely neglect AI suggested keywords.


Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #10 on: March 28, 2022, 16:02 »
0
Where does don't use long titles on Adobe come from?

When I first started contributing to AS, they had some guidelines on their website for new contributors to follow. And that included not having too many words in the title.

However, recently, I saw AS providing an example of an illustrative editorial title and it was quite long. So this contradicts what I read in their guidelines from a few years ago.

Thanks I always wonder.  :) Just like the people who insist always 50 keywords, even after AS says, do not use more than 49. Hard to tell what's information and myth. Maybe Mat will see this and give us an up to date version.

"Less is more" is part of Adobe company culture that goes beyond stock.  Opposite is Alamy discoverability criteria that requires at least 40 tags for "optimized" label (even if this does not probably have practical importance).  Some agencies (IS, AS) do not index title, others (Alamy, DT) put heaviest weight on title.  Then there is IS with clunky and non-portable managed dictionary concept. List goes on.  Bottom line is that it is impossible to come up with identical set of metadata that will be optimized for every agency out there

My take is to focus on the asset without thinking about target agency.  Then as certain photo types sell better on some agencies  refine on individual basis, but only if photo has strong sale potential.  I find Alamy measures tool invaluable as these are real customer search terms, while I largely neglect AI suggested keywords.



I've seen the same, or think I have, (Alamy and DT) but I stopped making notes and doing anything for any particular agency. I just try to put in the most important words and best description and move on. Just too much time spent trying to figure out how to game the system, when the simple answer is, the best answer, overall.

Keep in mind what others and eventually Adobe said: If you were looking at this image, what words would you use to describe it? And what is the main subject, what is prominent in the image. There you have it: Description and Keywords

Alamy: "detail who or what is in the image and what is happening" and "You shouldnt be including tags that are of little or no relevance to the photo." Your tags should be words to further describe your image they can be literal or conceptual.

What could be easier? No tricks or secrets, describe your image and add keywords that are relevant.


zeljkok

  • Non Linear Existence
« Reply #11 on: March 28, 2022, 17:09 »
0
No tricks or secrets, describe your image and add keywords that are relevant.

Best advice one could get

It is also why clunky IS managed dictionary concept is hugely flawed.  Nobody in their sane mind is going to search for "light - natural phenomenon"  "looking at view" "wave - water" etc.  But there are many photos with such IS Generated phrases at other sites, where they will be completely useless

Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #12 on: March 29, 2022, 16:20 »
0
No tricks or secrets, describe your image and add keywords that are relevant.

Best advice one could get

It is also why clunky IS managed dictionary concept is hugely flawed.  Nobody in their sane mind is going to search for "light - natural phenomenon"  "looking at view" "wave - water" etc.  But there are many photos with such IS Generated phrases at other sites, where they will be completely useless

Yes, while CV is a good thing, if the dictionary was bigger, it's terribly flawed when the real meaning of something is missing and we can't use the real or the actual meaning.

By the way, last time we were discussing this, I started a test on AL for Description, Keyword and Supertags. I used a nonsense word on some images, to see how it affects rank. Well, I have to admit, I've searched and I didn't write down the word, so I don't remember what it was?  ::) (being old can be difficult?) I checked all kinds of images, I can't remember which ones I tagged.

I had to go through again today and add a "nonsense" word that's not used for anything, to a group of images, so I could test the search.

« Reply #13 on: March 29, 2022, 17:48 »
+1
My personal philosophy is to use as many RELEVANT keywords as I can think of. Where we may differ is how I define the term relevant. The word needs to directly apply to the asset being indexed. When you get too abstract is where you can do more harm than good. At Adobe Stock, as most of you know by now, the first 10 keywords you list have the dominant share of impact in search placement. With keywords 11-49, the impact is negligible. The advice we always give is that 15-25 keywords is typically considered the sweet spot. Mosts assets can be described with 25 words or less, but some assets are far more complex and require more. If you can come up with 49 relevant keywords to describe your content, I don't see any harm in doing so. If however, you are stretching the imagination to come up with words that are iffy in the relevance department, you could be hurting yourself. If, for example you have a picture of a banana and you add the keyword apple, you aren't going to trick a customer into buying the photo if they are looking for an apple photo. If your banana shot appears in search, the customer won't click it so the system will assume this is not a sellable image and will push it a bit deeper into search results. The more that happens, the deeper into the later pages of search the image goes, the likelihood of sales happening from people actually looking for images of bananas diminishes.

You can't list more than 49 keywords at Adobe Stock. If you upload an image with 60 keywords embedded, the system cuts it off at 49 and the remaining 11 keywords are simply removed without notification. This is not ideal for anyone so you must be disciplined in that regard.

As far as the question about the title goes, you can list up to 200 characters in the title but we generally say the sweet spot is around 70. Again, no need to be overly strict about this, just use common sense and write out an accurate, descriptive title that reads like a sentence. If an image was captured of me at this exact moment in time I would title the photo "Middle aged bald man with beard typing on laptop computer keyboard." I would then be sure to include the words "bald, man, beard, typing, laptop, computer, keyboard" in my top ten keywords. Words that are listed in the title AND the top ten keywords are supercharged in search.

Everything else beyond that is gravy.

-Mat Hayward

zeljkok

  • Non Linear Existence
« Reply #14 on: March 29, 2022, 21:41 »
0

By the way, last time we were discussing this, I started a test on AL for Description, Keyword and Supertags. I used a nonsense word on some images, to see how it affects rank. Well, I have to admit, I've searched and I didn't write down the word, so I don't remember what it was?  ::) (being old can be difficult?) I checked all kinds of images, I can't remember which ones I tagged

How about "etepelcnu" ("uncle pete" in reverse, one word).  Nobody is going to search for it :) and you can't forget

Re IS managed concept;  it is not that poorly envisioned - indirect phrase concept that expands internally into several indexable keywords.  Problem is that people blindly copy and paste to other agencies, where such phrase won't have any value.  But largest problem with IS to me by far is inability to edit keywords after submission.  This is just dumb,  conditions/knowledge evolves and often you'd like to modify something.  Everyone else allows it, only IS doesn't. Why???  If the intention was to prevent keyword spamming, they should actually QA accurate keywords during submission which of course they don't.   To me IS interface from design standpoint is by margin worst in industry, period

zeljkok

  • Non Linear Existence
« Reply #15 on: March 29, 2022, 21:53 »
0
My personal philosophy is ....

-Mat Hayward

Excellent stuff as always. I will add 2 things:

1) Never automatically accept AI keyword suggestions.  Some will always be inaccurate.  Not such obvious banana/apple example, but more subtle. Example: I live in Canada Banff National Park and shoot lots of landscapes.  Adobe AI will frequently suggest "alps".  It recognizes mountains but naturally can not geographic affinity and mistake happens.  Big deal some might say. It can be a big deal, if you are customer sitting in France and want some nice shot of Euro Alps but are served Canadian Rockies instead

2) Adobe specific: It would be great if we could get customer search keyword(s) that led to sale.  Other agencies support this.   Alamy has Measures tool, Dreamstime directly tells you "Customer was searching for ...", even Shutterstock has "Insights" where such info is available.  This is great help;  since I started adjusting keywords on Alamy based on Measures, my CTR has gone up which impacted my Alamy rank and I am getting more sales.  Just an idea that would benefit everyone, incl. Adobe itself

« Reply #16 on: March 30, 2022, 02:19 »
0
My personal philosophy is to use as many RELEVANT keywords as I can think of. Where we may differ is how I define the term relevant. The word needs to directly apply to the asset being indexed. When you get too abstract is where you can do more harm than good. At Adobe Stock, as most of you know by now, the first 10 keywords you list have the dominant share of impact in search placement. With keywords 11-49, the impact is negligible. The advice we always give is that 15-25 keywords is typically considered the sweet spot. Mosts assets can be described with 25 words or less, but some assets are far more complex and require more. If you can come up with 49 relevant keywords to describe your content, I don't see any harm in doing so. If however, you are stretching the imagination to come up with words that are iffy in the relevance department, you could be hurting yourself. If, for example you have a picture of a banana and you add the keyword apple, you aren't going to trick a customer into buying the photo if they are looking for an apple photo. If your banana shot appears in search, the customer won't click it so the system will assume this is not a sellable image and will push it a bit deeper into search results. The more that happens, the deeper into the later pages of search the image goes, the likelihood of sales happening from people actually looking for images of bananas diminishes.

You can't list more than 49 keywords at Adobe Stock. If you upload an image with 60 keywords embedded, the system cuts it off at 49 and the remaining 11 keywords are simply removed without notification. This is not ideal for anyone so you must be disciplined in that regard.

As far as the question about the title goes, you can list up to 200 characters in the title but we generally say the sweet spot is around 70. Again, no need to be overly strict about this, just use common sense and write out an accurate, descriptive title that reads like a sentence. If an image was captured of me at this exact moment in time I would title the photo "Middle aged bald man with beard typing on laptop computer keyboard." I would then be sure to include the words "bald, man, beard, typing, laptop, computer, keyboard" in my top ten keywords. Words that are listed in the title AND the top ten keywords are supercharged in search.

Everything else beyond that is gravy.

-Mat Hayward

Thanks Mat. Two things:

Thanks for your contribution, on this topic and on others. Really appreciated.
You (and by you, Adobe) lead by example. I can only wish for other agencies doing the same.

Secondly: can I ask you what your philosophy is on long tail keywords?
How to use them and how to avoid keyword spamming which could negatively affect the ranking of content?
Let me give you some examples to clarify the question.
For instance: blue sky. Clear for me this is a long tail keyword, as the color blue only affects the sky, and has no other relation to subjects in the image.
It get's more difficult or complicated when using for instance location names. For instance: locations with Saint in it, which is often abbreviated as St. Or locations with hyphens. Locations which are well known both in native as in English language, or locations which are known by their official name and local slangs.
Different buyers will probably use different search strings to find these images or videos.

So I'm a bit puzzled: should we use all possible naming conventions as different long tail keywords? Or just add all possibilities as single keywords.
My personal take on this depends on the situation, and I mostly use a combination of both. Meaning main and unique part of the location as a single keyword, possible abbreviations, slangs or hyphens as long tail keywords. Not sure it's the right strategy, as it might become close to spam, depending on how the algorithm deals with it.

I gave locations as an example, but of course there are a lot of other different use cases. Easter bunny for instance, which I would always tag as a long tail keyword.


Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #17 on: March 30, 2022, 10:07 »
0
Words that are listed in the title AND the top ten keywords are supercharged in search.

-Mat Hayward

Good reminder. I always do that everywhere, I mean main words and describing the image with best words, but the Adobe "supercharge" is a good point. In the Caption + top ten Words = more value

Can you explain more of the why 70 words as the sweet spot? Or is it just a matter of reasoning, that less words, makes the image more specific and too many, is a weaker vague match in terms of what the buyer wants to see? Meaning, Adobe doesn't change the rank based on if someone has more or less words, it's just a fact that 70 is around the right number for a good match for a customer search?



 

Related Topics

  Subject / Started by Replies Last post
10 Replies
6061 Views
Last post August 21, 2013, 09:56
by roede-orm
2 Replies
5294 Views
Last post March 18, 2014, 12:27
by einstein
6 Replies
6205 Views
Last post September 11, 2015, 19:04
by FlowerPower
5 Replies
7529 Views
Last post August 14, 2016, 06:05
by Niakris
10 Replies
3943 Views
Last post December 19, 2020, 16:47
by Uncle Pete

Sponsors

Mega Bundle of 5,900+ Professional Lightroom Presets

Microstock Poll Results

Sponsors